i the Phonology and Phonetics of Jamaican Creole ( Pdfdrive )

THE PHONOLOGY AND PHONETICS OF
JAMAICAN CREOLE REDUPLICATION
DISSERTATION
Presented in Partial Fulfillment of the Requirements for

the Degree of Doctor of Philosophy in the
Graduate School of the Ohio State University
By
Shelome A. Gooden, B.A., M.A.
******
The Ohio State University

2003
Dissertation Committee: Approved by
Professor Donald C. Winford, Co-adviser ________________________

Co-advisor
Associate Professor Elizabeth V. Hume, Co-adviser Linguistics Graduate Program
Professor Mary M. Beckman
________________________
Co-adviser
Linguistics Graduate Program
i
© Copyright by
Shelome Gooden
2003
i
ABSTRACT
This dissertation is an extensive treatment of the phonological and phonetic
properties of Jamaican Creole (JC) reduplication. While reduplication is thought to be a
typical feature of Creole languages and has been studied in the past, to date little work
has been done on the phonetic or morpho-phonological properties of the process.
Complementing the analysis of reduplication developed in this work, is an analysis of the
prosodic system of the language. The analysis posited, treats the prosodic system of JC as
a stress-based system in which lexical contrasts are signaled by differences in the
alignment of the F0 contour with the word.
Reduplication processes in JC are similar in form and semantics to those found
in other Caribbean English Creoles. The processes are described and analysed from an
Optimality Theoretic perspective. The phonological aspect focused on delimiting the
constraints on the segmental properties of reduplication processes. The observation is
that the JC reduplicant is a prosodic foot which copies its base completely. Further, it is
shown that when the required phonological conditions cannot be satisfied, there is no
reduplication.
The phonetic aspect investigated how the phonological constraints on
reduplication interact with the phonetic properties of reduplicated words. Particular
ii
attention is paid to intensive and distributive reduplication processes, which yield
segmentally identical words. Specifically, it is shown that distributive reduplications
pattern like other words in the language. Consequently, they are treated as a single
prosodic word with a single nuclear pitch accent. Intensive reduplications pattern
differently and are analysed as two prosodic words with a prenuclear pitch accent and a
nuclear pitch accent.
This work has import firstly for our understanding of the productivity and scope
of reduplication in Jamaican Creole. It also contributes to our understanding of the
relationship between word-level prosody and the semantic functions of reduplicated
words in Creole languages in general. Thus, this dissertation represents an important step
towards our understanding of the relatively understudied area of Creole prosodic systems.
iii
Dedicated to my grandparents, Lois and Vaughn
iv
ACKNOWLEDGMENTS
For the existence of this dissertation I am indebted to a great many people. For my
intellectual growth, I owe a great deal to my advisors. Donald Winford’s advice both
academic and personal has helped me to keep focused over the years. I consider myself
blessed to have found teacher, mentor, advisor and friend in him. Elizabeth Hume has been
supportive, insightful and patient during my own quest to understand the inner workings of
my ‘mother tongue’. Mary Beckman has provided excellent council and support for my
research on this project.
Thanks also to my former teachers Hubert Devonish and Silvia Kouwenberg at the
UWI Mona campus who planted the seed of interest in academic research. I especially
thank Hubert Devonish for continual discussions of my research which include but are not
limited to, long phone calls and prolonged office hour meetings. Thanks also for giving me
access to the research facilities in the Linguistics Department at Mona during my visits
home and helping me find ‘hard to find’ books and manuscripts.
My fieldtrips to Jamaica were supported by the Department of Linguistics, the Office
of International Affairs and the Graduate School at OSU. I thank all those
v
involved with the different travel committees for helping me obtain the financial support
for my research. Thanks also to the administrative and computer support staff in the
department, especially Jim Harmon, for technical help and for giving me more than fair
share of server space to store my many sound files.
I appreciate the support and help I received from my fellow graduate students. I
want to thank especially Hope Dawson, Michelle Ramos-Pellicia, Misun Seo and Huang
Tsan for friendship, and more importantly, sisterhood. Members of the Phonies and
Changelings discussion groups in the department deserve thanks as well for providing
forums for discussing my work.
The incessant prayers of my mother, Hazel, and my grandmother, Agatha, have
sustained through the years and my daddy, Haskel taught me that possibilities are endless
and encouraged me to pursue my goals.
Finally I would also like to acknowledge the support of the citizens of the Top Alston
community who have been patient enough to endure my many questions about patwa.
Special thanks to Sister Gloria, Miss Addy and Miss Brown for taking the time to provide
me with meals.
vi
CURRICULUM VITAE
September 6th, 1973 ......................................Born - Clarendon, Jamaica W.I.
1993.................................................................B.A. (Honors) Linguistics,
University of the West Indies,
Jamaica
2002.................................................................M.A. Linguistics, Ohio State University
1997- 2001.......................................................Graduate Teaching and Research Associate
The Ohio State University
2001 - 2003......................................................Editorial Associate
Journal of Pidgin and Creole Languages
PUBLICATIONS
Research Publication
1. (2002). Past Time Reference in Belizean Creole. UPenn Working Papers in Linguistics.
Vol. 8:3. Selected Papers from NWAV 30.
2. (2003a). Reduplication in Jamaican Creole: Semantic Functions and Prosodic

Constraints. In S. Kouwenberg (ed). Twice as Meaningful: Reduplication in Pidgin
and Creoles. Westminster Creolistics Series Volume 8. London. Battlebridge
Publications. pgs. 93-103.
vii
3. (2003b). with Silvia Kouwenberg and Darlene LaCharité. Jamaican Creole
Reduplication: An Overview. in S. Kouwenberg (ed). Twice as Meaningful:
Reduplication in Pidgin and Creoles. Westminster Creolistics Series Volume 8.
London. Battlebridge Publications. pgs. 105 -110.
4. (2003c). with D. Winford, H. Dawson and R. Dodsworth. OSU Working Papers in

Linguistics. Studies in Language Contact and Language Change.
5. (2003d). Prosodic Contrast in Jamaican Creole Reduplication. In Ingo Plag (ed). The
Phonology and Morphology of Creole Languages. Linguistische Arbeiten. Tübingen.
Niemeyer. pgs. 193 - 208.
FIELDS OF STUDY
Major Field: Linguistics
Minor Fields: Phonology, Creole Linguistics, Sociolinguistics
viii
TABLE OF CONTENTS
Page
Abstract................................................................................................................................ii
Dedication………………………………………………………………………………...iii
Acknowledgement...............................................................................................................v
Vita....................................................................................................................................vii
List of Maps......................................................................................................................xiii
List of Tables....................................................................................................................xiv
List of Tableau ..................................................................................................................xv
List of Figures.................................................................................................................xviii
Chapters:
1. Introduction......................................................................................................................1
1.1. Overview.................................................................................................................. 1
1.2. Goals of the Study.................................................................................................... 4
1.3. Phonological Aspects............................................................................................... 5
1.4. Phonetic Aspects.................................................................................................... 10
1.5 Broader Implications............................................................................................... 13
1.6. Theoretical Assumptions ....................................................................................... 14
1.6.1. Optimality Theory....................................................................................... 15
1.6.2. Metrical Theory .......................................................................................... 20
1.7. Structure of the dissertation ................................................................................... 25
ix
2. Reduplication in Jamaican Creole and Other Caribbean English Creoles.....................28
2.1. Introduction........................................................................................................... 28
2.2. Reduplication in Jamaican Creole..........................................................................30
2.2.1. Iterative Reduplication................................................................................. 32
2.2.2. Characteristic Reduplication........................................................................ 33
2.2.3. Intensive Reduplication ............................................................................... 35
2.2.3.1. Complex Input Types...................................................................... 35
2.2.3.2. Simple Input Types ......................................................................... 37
2.2.4. Distributive Reduplication ........................................................................... 38
2.2.5. Other Types.................................................................................................. 40
2.2.6. Discussion .................................................................................................... 42
2.3. The Semantics of CEC Reduplication ................................................................... 43
2.3.1 Iterative.........................................................................................................43
2.3.2 Intensive........................................................................................................44
2.3.3 Characteristic................................................................................................46
2.3.4 Distributive...................................................................................................47
2.3.5 Attenuative....................................................................................................49
2.3.6 Stative...........................................................................................................50
2.3.7 Other Meanings.............................................................................................51
2.4. Full vs. Partial Reduplication................................................................................ 52
2.5. Prosodic Properties of CEC Reduplicated Words ................................................ 53
2.6. Morphological Function of Reduplicated Words ................................................. 59
2.7. Chapter Summary ................................................................................................. 60
3. Jamaican Creole in its Sociocultural Context................................................................62
3.1. Introduction............................................................................................................ 62
3.2. Sociohistorical Background ................................................................................... 63
3.2.1. Impact of Language Contact on Word-Level Prosody ................................ 66
3.2.2. Impact of Language Contact on Reduplication ........................................... 67
3.3. Contemporary Language Situation ........................................................................ 68
3.3.1. Variation in Creole Continua ....................................................................... 70
3.4. Community Profile................................................................................................. 71
3.5. Data Collection ...................................................................................................... 74
3.5.1. Subjects ........................................................................................................ 74
3.5.2. Production Tasks.......................................................................................... 75
3.6 Chapter Summary ................................................................................................... 76
4. Word Prosody: Evidence for JC as a Stress-Based System………………………….....78
4.1. Introduction…………………………………………………………………….78
4.2. The Phoneme Inventory…………………………………………………….......81
4.3. Prosodic Structure…………………………………………………………...…84
x
4.3.1 Syllable Structure………………………………………………………..84
4.3.2. Foot Structure and Moraicity……………………………………………85
4.4. JC Word-Level Prosody……………………………………………….. ............85
4.4.1.Previous Analyses of JC Prosody……………………….………………..86
4.4.2. Criteria for Prosodic Classification………………………………………97
4.5. Evidence for Stress in JC……………………………………………………...100
4.5.1. Theoretical Assumptions………………………………………………..101
4.5.1.1. Prosody ................................................................................................101
4.5.1.2. Notations..............................................................................................102
4.5.1.3. F0 Plots.................................................................................................103
4.5.2. Tone or Stress: F0 Pattern in Bisyllabic Words ..........................................105
4.5.2.1. Summary and Discussion....................................................................115
4.6. Discovering Contrasting Cues to Word Prosody: Techniques………...……... 117
4.6.1 Strategies………………………………………………………………..118
4.6.1.1. Discussion…………………………………………………...125
4.6.2. Method for Data Collection…………………………………………….126
4.6.2.1 Speakers…………………………………………………….. …..126
4.6.2.2. The Data………………………………………………….............127
4.6.2.3. Elicitation Technique……………………………………………129
4.7. F0 Pattern in other Monomorphemic Words and Compounds……………...…132
4.7.1. F0 Pattern in Trisyllabic Words………………………………...............132
4.7.1.1. Interim Summary……………………………………………..137
4.7.2. .. F0 Pattern in Quadrasyllabic words…………………………………. 138
4.7.2.1. Interim Summary……………………………………………..142
4.7.3. F0 Pattern in Compound Words………………………………………..143
4.7.3.1. Interim Summary……………………………………………..147
4.7.4. Section Summary………………………………………………………148
4.8. Chapter Summary ............................................................................................149
5 Word Level Stress in Jamaican Creole………….……………………………………151
5.1 Introduction………………………………………………...……………........151
5.2 Stress Patterns………………………………………………………………...152
5.2.1 Bisyllabic Words .....................................................................................152
5.2.2 Trisyllabic Words ....................................................................................156
5.2.3 Quadrasyllabic Words .............................................................................162
5.2.4 Interim Summary.....................................................................................163
5.2.5 Compound Words.....................................................................................165
5.3 Summary…………...………………………………………………...……… 168
5.4 Predicting Stress Assignment in JC - An Optimality Theoretic Account ..........169
5.4.1 General Pattern ..........................................................................................169
5.4.2 Alternate Patterns.......................................................................................187
5.4.3 Section Summary.......................................................................................194
5.5 Chapter Summary…………………………………………………….…….. ..195
xi
6 Phonetic Study of the Prosody of Reduplicated Words…………………...…………196
6.1 Introduction…………………………………………………………………...196
6.2 F0 Pattern of Reduplicated Words…………………………………… .............199
6.2.1. Monosyllabic Base Words…………………………………………….199
6.2.1.1 Distributive Reduplication…….……………………………..199
6.2.1.2. Intensive Reduplication…….………………………………..203
6.2.1.3. Summary………………….………………………………....208
6.2.2. Bisyllabic Based Words……………………………………………….210
6.2.2.1.....Distributive Reduplication………………………………….211
6.2.2.2. .... Intensive Reduplication…………………………………….214
6.2.2.3.... F0 Pattern on Narrow Focus Phrases………………………..216
6.2.2.4.... Summary……………………………………………………219
6.3. Discussion……………...……………………………………………………..222
6.4. Chapter Summary………………………...…………………………...……... 225
7. An Optimality Theoretic Account of JC Reduplication............................................226
7.1. Introduction.......................................................................................................... 226

7.2. Iterative Reduplication......................................................................................... 227
7.2.1.Interim Summary ........................................................................................ 235
7.3. Characteristic Reduplication................................................................................ 235
7.3.1. Interim Summary ...................................................................................... 244
7.4. Intensive Reduplication (Complex Input)............................................................ 245
7.4.1. Interim Summary ....................................................................................... 251
7.5. Prosodically Disambiguated Reduplications ....................................................... 252
7.5.1. Distributive Reduplication ......................................................................... 255
7.5.1.1. Interim Summary .......................................................................... 264
7.5.2. Intensive Reduplication (Simple Input) .................................................... 264
7.5.2.1 Interim Summary .......................................................................... 271
7.5.3. Section Summary ...................................................................................... 272
7.6. Previous Accounts of JC Reduplication .............................................................. 273
7.7. Chapter Summary ................................................................................................ 275
8. Conclusion .............................................................................................................. 279
Appendices:
A. Sample Informant Data Sheet................................................................................285

B. F0 Pattern in Reduplicated Words Produced by Other Speakers...........................286
Bibliography ...............................................................................................................289
xii
LIST OF MAPS
Map Page
3.1. Sources of Jamaican population (Lalla and D’Costa 1990:4) ................................... 64
3.2. Map of Jamaica Showing the Fourteen Electoral Parishes........................................ 72
xiii
LIST OF TABLES
Table Page
3.1: Informant Demographic Data .........................................................................................75
4.1: JC Consonant Inventory ..................................................................................................82
4.2: JC Vowel Inventory.........................................................................................................82
4.3: Summary of F0 Patterns……………………………………………………………..115
4.4: Elicited Words used in Analysis...................................................................................128
4.5: Prosodic Conditions for Target Words.........................................................................128
4.6: Summary of Pitch Accents in Trisyllabic, Quadrasyllabic and Compound

Words............................................................................................................................149
5.1:General Stress Pattern in Monomorphemic words of different syllable types……….164
6.1: Elicited Reduplicated Words used in Analysis……………………………………...198
6.2: Summary of Pitch Accents in Intensive and Distributive Monosyllabic Based

Words.............................................................................................................................210
6.3: Summary of Pitch Accents in Intensive and Distributive Bisyllabic Based

Words............................................................................................................................221
xiv
LIST OF TABLEAUX
Tableau Page
1.1: [nyamnyam] ‘food’ .......................................................................................................19
5.1: ["kaSa ] ‘type of thorny tree’ .........................................................................................171
5.2: ["kaSa ] ‘type of thorny tree’………………………………………………………....172
5.3: [mo"zEla] ‘type of yam’………………...…………………………………………....173
5.4: [ÆalI"gEta] ‘alligator’.......................................................................................................173
5.7: [a"kam ] ‘keloid’ ............................................................................................................176
5.8: [a"kam ] ‘keloid’ ............................................................................................................177
5.9: ["gandZa] ‘marijuana’.....................................................................................................178
5.10: ["gaNgalu] ‘ruffian, bully’ ............................................................................................179
5.11: [su"sumba] ‘wild bean’ ................................................................................................180
5.12: [Ækala"ban] ‘box like trap’ ..........................................................................................181
5.13: [ÆgjalI"mEnta] ‘type of wood’.................................................................................…..181
5.14: ["gaalin] ‘egret’ ............................................................................................................182
5.15. ["baNgaÆraN] ‘useless odds and ends’ ...........................................................................183
5.16: ["baNgaÆraN] ‘useless odds and ends’..........................................................................184

xv
5.17: [ma"SietId] ‘dilapidated, of ill appearance’…............................................................185
5.18: ["kaNkantar] ‘type of top’.............................................................................................185
5.19: [ba"fan] ‘clumsy person’............................................................................................189
5.20: [fan"daNgl] ‘elaborate ornament’...............................................................................190
5.21: [ad"vantidZ ] ‘to take advantage of, cheat’ ..................................................................190
5.22: [pro"vIdZan ] ‘generic term for yam, potatoes, etc’.....................................................191
5.23: ["kumiÆna] ‘type of religious dance ceremony’............................................................192
5.24: ["babiÆlan] ‘rastafarian term for non-believer or policeman’.......................................193
5.25: ["bagaÆbu ] ‘insects, worms’ .........................................................................................194
7.1. [pItoppItop] ‘to spit on repeatedly’ ..............................................................................229
7.2. [maakopmaakop] ‘to mark repeatedly’.......................................................................230
7.5. [njakaop] ‘to cut/chop crudely’ ...................................................................................232
7.8. [naasInaasI] ‘having nasty, filthy characteristics’.......................................................237
7.9. [laafIlaafI]‘inclined to laughter’ ..................................................................................238
7.10. [laafIlaafI]‘inclined to laughter’ ................................................................................239
7.11. [dZukIdZukI] ‘prickly or needle-like’ ........................................................................241
7.12. [grIEnIgrIEnI] ‘having grainy, coarse characteristics’ ...............................................242
7.13. [*tSupIdtSupId] ............................................................................................................243
xvi
7.14. [*tSupIdItSupIdI] ........................................................................................................244
7.15. [ maalIhakop ] ‘to make a tremendous mess of…………………………………...248
7.16. [sIdoNop] ‘to sit upright/relaxed for a prolonged period’ ..........................................249
7.17. [baksbaksop] ‘to hit hard in the face’........................................................................250
7.18. [naasInaasIop] ‘to make completely filthy’ ..............................................................250
7.19: [ÆgjalI"mEnta] ‘type of wood’.................................................................................254
7.20: ["gaalin] ‘egret’ ...........................................................................................................254
7.21. [ ÆgriEnI"griEnI] ‘grainy all over’ ..................................................................................257
7.22. [ Ægrin"grin] ‘green all over’..........................................................................................257
7.23. [ Ægrin"grin] ‘green all over’..........................................................................................258
7.24. [ÆblakI"blakI] ‘black all over’ .......................................................................................259
7.25. [Ægrin"grin] ‘green all over’..........................................................................................261
7.26. [ÆgomI"gomI] ‘gummy all over’...................................................................................262
7.27. [ÆgriEnI"griEnI] ‘very long’ ...........................................................................................263
7.28. ["laN"laN] ‘very long’.....................................................................................................266
7.29. ["gomI"gomI] ‘very gummy’.........................................................................................266
7.30. ["grin"grin] ‘very green’................................................................................................267
7.31. ["grin"grin ] ‘very green’ .............................................................................................269
7.32. [ "gomI"gomI] ‘very gummy’........................................................................................270
7.33. ["gomI"gomI] ‘very gummy’.........................................................................................271
xvii
LIST OF FIGURES
Figure ............................................................................................................................Page
4.1 F0 contour of the statement Mieri want wan buriburi wan ‘Mary wants
one that has burrs all over’ (dotted line) and the related question Mieri
want wan buriburi wan Does Mary want one that has burrs all over?
(solid line)...........................................................................................................104
4.2. Annotation conventions.......................................................................................104
4.3. Fundamental frequency contour (upper panel) and spectrogram (lower panel)
of the word "mada ‘mother’ produced in isolation. This word has primary stress
on the initial syllable............................................................................................106
4.4. Fundamental frequency contour (upper panel) and spectrogram (lower panel)of
the word "mada ‘mother’ produced in statement intonation in final position
som yam fi mi mada ‘Some yams for my mother’ (left graph) and in non-final
position mi mada yam dem ‘My mother’s yams’(right graph). ..........................107
of the word "mada ‘mother’ produced in question intonation in final position
a fi mi mada ‘Is it my mother’s?’ (left graph) and in non-final position a
fi mi mada yam ‘Is it my mother’s yam?’(right graph).....................................108
of the word ma"da ‘female religious leader’ produced in citation form.
This word has primary stress on the second syllable...........................................109
of the word ma"da ‘female religious leader’ produced in statement intonation
in final position, som yam fi mi mada ‘Some yams for my female religious
leader’ (left graph), and in non-final position, mi mada yam dem
‘My female religious leader’s yams’(right graph)...............................................110
xviii
of the word ma"da ‘female religious leader’ produced in question intonation
in final position a fi mi mada ‘Is it my female religious leader’s?’ (left graph)
and in non-final position a fi mi mada yam ‘Is it my female religious
leader’s yam?’(right graph)..................................................................................111
4.9. Fundamental frequency contours (upper panels) and spectrograms (lower

panels)of the word "mada ‘mother’ produced in statement intonation in
final position, som yam fi mi mada ‘some yams for my mother’ (left graph),
and ma"da ‘female religious leader’ produced in statement intonation in
final position, som yam fi mi mada ‘Some yams for my female religious
leader’ (right graph).............................................................................................112
4.10. Fundamental frequency contours (upper panels) and spectrograms (lower panels) of
the word word "mada ‘mother’ produced in statement intonation in non-final
position, mi mada wantsom yam ‘my mother wants some yams’ (left graph), and
ma"da ‘female religious leader’ produced in statement intonation in non-final
position, mi mada want som yam ‘My female religious leader wants some yams’
(rightgraph).............................................................................................................113
4.11. Fundamental frequency contours (upper panels) and spectrograms (lower panels)
of the word "mada ‘mother’ produced in question intonation in final position,
a som yam fi mi mada ‘Are these some yams for my mother?’ (left graph),
and ma"da ‘female religious leader’ produced in statement intonation in
final position, som yam fi mi mada ‘Are these some yams for my female
religious leader?’ (right graph)...............................................................................113
of the word word "mada ‘mother’produced in question intonation in
non-final position, mi mada want som yam ‘Does my mother wants some
yams?’ (left graph), and ma"da ‘female religious leader’ produced in
statement intonation in non-final position, mi mada want som yam ‘Does
my female religious leader wants some yams?’(right graph)..................................114
4.13. Fundamental frequency contour of (upper panel) and spectrogram

(lower panel) mi"riina ‘men’s undershirt’ produced in citation form. This
word has primary stress on the penultimate syllable..............................................132
4.14. Fundamental frequency contour (upper panel) and spectrogram of

(lower panel) mi"riina ‘men’s undershirt’ produced in statement intonation
final position, im want wan miriina ‘He wants a men’s undershirt’ (left graph)
and non-final position, im want wan miriina an som yam ‘He wants a
men’s undershirt and some yams’ (right graph)...................................................133
xix
of mi"riina ‘men’s undershirt’ produced in question intonation final position,
im want wan miriina ‘Does he want a men’s undershirt?’ (left graph) and
non-final position, im want wan miriina an som yam ‘Does he want
a men’s undershirt and some yams?’ (right graph).................................................134
of the word Ina"mEl ‘inamel’ produced in citation. This word has primary stress
on the final syllable..................................................................................................135
4.17. Fundamental frequency contours (upper panels) and spectrograms

(lower panels) of the word Ina"mEl ‘inamel’ produced in statement intonation
final position, im wan we a inamel ‘He wants one that is (made of) enamel’
(left graph) and non-final position, im wan we a inamel an som yam ‘He wants
one that is (made of) enamel and some yams.’ (right graph)...................................136
of the word Ina"mEl ‘enamel’ produced in question intonation final position, im
wan we a inamel ‘Does he wants one that is (made of) enamel?’ (left graph)
and non-final position, im wan we a inamel an som yam ‘Does he wants one
that is (made of) enamel and some yams?’ (right graph).......................................137
of ÆgjalI"mEnta ‘type of wood’ produced in citation. This word has primary
stress on the penultimate syllable...........................................................................138
of ÆgjalI"mEnta ‘type of wood’ produced in statement intonation in final position,
im want wan outa gyalimenta ‘He wants one of galimeta wood’ (left graph) and
in non-final position, im want wan outa gyalimenta an mahuo ‘He wants one
of galimeta wood and mahoe’ (right graph)............................................................139
of ÆgjalI"mEnta ‘type of wood’ produced in question intonation in final position,
im want wan outa gyalimenta ‘Does he want one of galimeta wood?’
(left graph) and in non-final position, im want wan outa gyalimenta an mahuo
‘Does he want one of galimeta wood and mahoe?’ (right graph)...........................140
of ÆalI"gEta ‘alligator’ produced in citation. This word has primary stress
on the penultimate syllable......................................................................................141
xx
of ÆalI"gEta ‘alligator’ produced in statement intonation in final position,
im want wan aligeta ‘He wants an alligator’ (left graph) and in non-final
position, im want wan aligeta an som yam ‘He wants an alligator and
some yams’ (right graph).........................................................................................141
of ÆalI"gEta ‘alligator’ produced in question intonation in final position, im want
wan aligeta ‘Does he want an alligator?’ (left graph) and in non-final position,
im want wan aligeta an som yam ‘Does he want an alligator and some yams?’
(right graph).............................................................................................................142
of ÆbElI"uman ‘pregnant woman’ produced in citation.............................................143
of ÆbElI"uman ‘pregnant woman’ produced in statement intonation final position,
im want wan beliuman ‘He wants a pregnant woman’ (left graph) and in
non-final position, im want wan beliuman an som yam ‘He wants a pregnant
woman and some yams.’ (right graph) ...................................................................144
of ÆbElI"uman ‘pregnant woman’ produced in question intonation final position,
im want wan beliuman ‘Does he want a pregnant woman?’ (left graph) and in
non-final position, im want wan beliuman an som yam ‘Does he want a
pregnant woman and some yams?’ (right graph).....................................................145
of ÆbElI"uman ‘pregnant woman’ produced in statement intonation final
position, (im) want wan beliuman ‘He wants a pregnant woman’ and in
non-final position, (im) want wan beliuman an som yam ‘He wants a
pregnant woman and some yams.’ (right graph).....................................................146
of ÆbElI"uman ‘pregnant woman’ produced in question intonation final position
yu want wan beliuman ‘Do you want a pregnant woman?’ (left graph) and in
non-final position yu want wan beliuman an som yam ‘Do you want a
pregnant woman and some yams?’ (right graph) ....................................................147
of the reduplicated word Ægrin"grin ‘scattered green’ produced in citation
form by speaker SG. This reduplicated word has primary stress on the
second syllable.........................................................................................................200
xxi
of the reduplicated word Ægrin"grin ‘scattered green’ produced in statement
intonation in final position in the sentence, im want wan we griingriin ‘He
wants one that is scattered green’ (left graph) and in non-final position,
in the sentence, im want wan griingriin wan an som yam ‘He wants a
scattered green one and some yams’ (right graph)..................................................201
of the reduplicated word Ægrin"grin ‘scattered green’ produced in question
intonation in final position in the sentence, im want wan we griingriin ‘Does
he want one that is scattered green?’ (left graph) and in non-final position
in the question, im want wan griingriin wan an som yam ‘Does he want a
scattered green one and some yams?’ (right graph).................................................202
of the reduplicated word "grin"grin ‘very green’ produced in citation form by
speaker RP..............................................................................................................204
of the reduplicated word "grin"grin ‘very green’ produced in citation form by
speaker SG...............................................................................................................205
of "luo"luo produced in citation form with an intensive meaning ‘very low’
by HF.......................................................................................................................206
of the reduplicated word "grin"grin ‘very green’ produced in statement intonation
in final position in the sentence, im want wan we griingriin ‘He wants one
that is very green’ (left graph) and in non-final position, in the sentence,
im want wan griingriin wan an som yam ‘He wants a very green one and
some yams’ (right graph). .......................................................................................207
of the reduplicated word "grin"grin ‘very green’ produced in question intonation
in final position in the sentence, im want wan we griingriin ‘Does he want one
that is very green?’ (left graph) and in non-final position, in the sentence, im
want wan griingriin wan an som yam ‘Does he want a very green one and
some yams?’(right graph)........................................................................................208
of niernier produced with a distributive meaning ‘not quite near’ (left graph)
and with an intensive meaning ‘very near’ (right graph) by EW...........................209
xxii
for the reduplicated word ÆjElo"jElo ‘scattered yellow’ (distributive meaning)
produced citation.....................................................................................................211

(lower panels) for reduplicated word ÆjElo"jElo ‘scattered yellow’
(distributive meaning) produced in statement intonation in final position
in the sentence, im want wan we yeloyelo ‘He wants one that is scattered
yellow’ (left graph) and in non-final position in the sentence, im
want wan yeloyelo mango ‘He wants a scattered yellow mango’
(right graph).............................................................................................................212

(lower panels) for the reduplicated word ÆjElo"jElo ‘scattered yellow’
(distributive meaning) produced in phrase final position in yes-no
question intonation in the phrase yu waant wan we yeloyelo ‘Do you want one
that is scattered yellow?’ (left graph) and in non final phrase position in the
question yu waant wan yeloyelo mango ‘Do you want a scattered
yellow mango?’ (right graph)..................................................................................213
6.13 Fundamental frequency contour (upper panel) and spectrogram (lower panel)
for the reduplicated word "jElo"jElo ‘very yellow’ (intensive meaning)
produced in citation..................................................................................................214

(lower panels) for the reduplicated word ÆjElo"jElo ‘very yellow’
(intensive meaning) produced in statement intonation final position in the
sentence Di wan we in want yeloyelo ‘The one he wants is very yellow’
(left graph) and in non-final position in the sentence, im want wan yeloyelo
mango ‘He wants a very yellow mango’ (right graph)...........................................215

(lower panels) for the reduplicated word ÆjElo"jElo ‘very yellow’
(intensive meaning) produced in question intonation in the sentence final
position in the sentence yu waant wan we yeloyelo ‘Do you want one that
is very yellow?’ (left graph) and in non final position in the sentence yu
waant wan yeloyelo mango ‘Do you want a very yellow mango?’
(right graph).............................................................................................................216
6.16 Fundamental frequency contour (upper panel) and spectrogram (lower panel)
for the phrase "jElo"maNgo ‘yellow mango’ produced in citation............................217
xxiii
(lower panels) for the phrase "jElo"maNgo ‘yellow mango’ produced in
statement intonation in final position in the sentence, im want wan yelo mango
‘He wants a yellow mango’ (left graph) and in non-final position in the
sentence, im want wan yelo mango an som yam ‘He wants a yellow mango
and som yams’ (right graph)....................................................................................218

(lower panels) for the phrase "jElo"maNgo ‘yellow mango’ produced in
in yes-no question intonation in final position in the question yu want yelo
mango ‘Do you want a yellow mango?’ (left graph) and in non-final position
in the question yu wan yelo mango an som yam ‘Do you want a yellow mango
and some yams?’ (right graph)................................................................................219

(lower panels) of jElojElo produced in citation with a distributive meaning
‘yellow all over - scattered’(left graph) and an intensive meaning
‘very yellow’ (right graph) by HF...........................................................................220
xxiv
CHAPTER 1
INTRODUCTION
1.1 Overview
Reduplication is a morphological process in which all or part of an existing word is
replicated to form a new word. The process is attested in a wide range of languages
where it often marks grammatical or semantic contrasts. For example, in Ju|’hoansi the
word n|ù’í ‘to be shiny’ is reduplicated to form the word n|ù’ín|ù’í ‘to cause to be shiny’;
in Turkish gyzel ‘beautifully’ reduplicates to form gyzelgyzel ‘very beautifully’; in
Guyanese Creole the word wâytíi ‘whitish’ is reduplicated to form wâytíiwâytíi ‘whitish
all over’; and in Tagalog lakad ‘walk’ reduplicates to form lalakad ‘will walk’.
Though this process is known to be typical of Caribbean English Creole
languages (CECs) (cf. Hall, 1966 in Sebba, 1981), the majority of studies have focused
on providing descriptions of the processes or discussions about their origin.
Comparatively little is known about the morpho- phonological or phonetic properties of
the process. As I show in this dissertation, morpho-phonological as well as phonetic
descriptions are crucial to the understanding of the differences between contrasting pairs
of segmentally identical reduplicated words. In fact, it has been argued elsewhere that
-1-
where reduplication creates segmentally identical words, creole languages manipulate the
prosodic properties of the words such as stress or tone, to effect differences in meaning.
For example, Devonish (2003) describes verbal reduplication in Guyanese Creole, in
which the verb has an iterative meaning if there is a high pitch throughout the word, e.g.
rónrón ‘to run continuously’. However, when the reduplicated verb has a high-low
(falling) pitch on the first syllable and a high pitch on the second syllable e.g. rônrón the
meaning is a distributive one, ‘to run in fits and starts’. Similarly, Sylvain (1936)
observed that in Haitian French Creole, differences in meaning between segmentally
identical reduplicated adjectives are conveyed by differences in tone. For example, when
the meaning is ‘intensive’ the initial syllables of words are pronounced with a high tone
e.g. píkepíke ‘very pricking’. However, where the meaning is ‘attenuative’ the initial
syllables are pronounced with a low tone e.g. pìkepìke ‘slightly pricking’.
Cross-linguistic studies on reduplication processes have shown that the
processes tend to have specific phonological conditions imposed on them, which are a
vital part of the inherent identity relationship between base and reduplicant. This study
accounts for several types of reduplication attested in Jamaican Creole (JC) with the aim
of delimiting the constraints on reduplication in the language. Particular attention is paid
to two types of reduplication processes that yield segmentally identical words. I test the
hypothesis that the contrast between these reduplications is a result of differences in
prosodic properties.
Jamaican Creole (JC) has a rich morphological subsystem in which there are
several productive processes of reduplication operative in all major word classes yielding
-2-
different semantic types. In distributive reduplication, the meaning of the reduplicated
word has the sense of scattered, all over the place, here and there or occasionally. In
iterative reduplication, the meaning of the reduplicated word describes an action
performed repeatedly and in intensive reduplication the degree of quantity (noun), the
degree of action (verb) or the degree of quality (adjective) has more intensity or emphasis
in the reduplicated word than in the unreduplicated word. The last type is characteristic
type reduplication where the meaning expressed by the reduplicated form incorporates
the feature X, where X is a semantic component of the base. For example, the word 
‘to pierce’ reduplicates as follows:
(1) (a)  ‘prickly object’

(b)  ‘characteristically prickly all over/ pierced all
over’ (Distributive)
(c)  ‘very prickly’ (Intensive)
(d)   ‘to prick repeatedly’ (Iterative)
1
The form in (a) is a noun which incorporates a distributive meaning; the form in (b) is an
adjective and has a distributive meaning. The form in (c) is also an adjective but has an
intensive meaning and the form in (d) is a verb and has an iterative meaning.
Interestingly, two of these reduplication processes yield segmentally identical forms that
are arguably prosodically distinct. For example, the form in (b) with a distributive
meaning is segmentally identical to the form in (c) with an intensive meaning. The aim
of this study is to discover the nature of the prosodic contrasts between these identical
reduplicated words.
1
This can occur in both predicative and attributive position. In predicative position it indicates a process and in
attributive position, it indicates a result of an action. There appears to be no prosodic differences between the two.
-3-
This study builds on earlier work (Gooden, 2002; 2003a; 2003b) in which I
developed the hypothesis that where different reduplication processes produce
segmentally identical words, there are additional requirements at the prosodic level so
that semantic distinctions are maintained.
1.2 Goals of the Study
This dissertation investigates how phonological constraints on JC reduplication interact
with phonetic properties to define the set of wellformedness conditions on copying. The
dissertation is an extensive treatment of the phonology of Jamaican Creole reduplication
that also draws on phonetic evidence in an effort to shed light on its prosodic properties.
The study focuses in particular on the word-level prosodic properties of two reduplication
processes that yield segmentally identical words: distributives and intensives.
Three main issues are addressed. First, the study defines the full range of
reduplicative processes in JC. Second, an extensive phonological analysis is provided
which accounts for both the segmental and prosodic properties of the reduplicated words.
Third, the study describes the phonetic characteristics of segmentally identical
reduplicated words and shows how these phonetic characteristics are interpreted in the
phonology.
As was noted above, while there is some agreement that differences in pitch may
play a role in reduplication in creole languages, there is disagreement over whether the
relevant phonological implementation is stress or tone. For example, stress was shown to
be an important factor in describing Mauritian Creole reduplication (Baker 1999),

-4-
whereas tone was the crucial factor in Krio (Nylander 1999). For Sranan however, the
picture is less straightforward. Dioncie (1959 in Sebba, 1981) for example, suggests that
tone distinguishes between segmentally identical reduplicated words whereas Adamson
and Smith (1999) identify stress as the distinguishing feature. An adequate phonological
analysis of reduplication in these languages must account for these differences as well as
interpret them within the context of the relevant prosodic system. As I discuss in Chapter
4, with regard to pitch, there are crucial differences in the phonetic implementation of the
phonological categories stress and tone. With regard to JC I show that a stress is the
relevant phonological category rather than tone.
It is important to note, also that while descriptions of the phonetic properties of
CEC reduplication are informative, they are based on impressionistic auditory analyses.
There is no empirical evidence for whether it is absolute pitch differences (the perceptual
correlate of F0) alone or differences in how the F0 associates with the stressed syllables of
words that accounts for lexical contrasts. The analyses of the JC data proposed in this
dissertation provide a background against which to make more robust claims about these
prosodic differences in JC, by providing a precise description of the phonetic properties
of the words. I show in Chapter 4, that lexical contrasts are accounted for by referring to
differences in the alignment of the F0 contour with words.
1.3 Phonological Aspects
Several issues will be addressed in this dissertation with regard to the phonological
properties of reduplication: the shape and size of the copied material (reduplicant); the
-5-
location of the reduplicant (prefixed, suffixed); the failure of reduplication in some
words; the presence of invariant segments in both the reduplicant and the base which are
not observed in the input; and the role of the phonetic properties of the words,
specifically, pitch accents, in devising an accurate phonological description of
segmentally identical reduplications.
Shape and size of reduplicant. Reduplication in CECs is full reduplication in the
majority of cases but there are also cases of partial reduplication (Sebba, 1981;
Kouwenberg and LaCharité, 1998; Gooden, 2003a). In both instances, we need to know
what the requirements on reduplication are. In cases of whole word reduplication, any
discussion of a fixed shape or size of the reduplicant appears to be rendered superfluous
since the entire base is copied. Recent work (Alderete, 1993; Gooden, 1999; 2003a) on
the morphophonology of reduplication in JC suggests, however, that there are in fact
restrictions on how much of a base is copied and further, what such a copy should look
like. Gooden (2003a), for example, argues that the reduplicant of characteristic
reduplication is a bisyllabic foot. If the reduplicant is too small, a segment has to be
inserted, as seen in (2). As discussed in Chapter 7, this vowel is also inserted to satisfy a
morphological function.
(2) laaf ‘to laugh’ laafIlaafI ‘inclined to laughter/jovial’
swit ‘sweet’ switIswitI ‘characteristically sweet’
daag ‘dog’ daagIdaagI ‘characteristically doglike’
-6-
If the reduplicant would be larger than two syllables, reduplication is blocked, as in (3).
(3) lIzad ‘lizard’ *lIzadIlIzadI ‘characteristically lizard like’
pIkInI ‘child’ *pIkInIIpIkInII ‘characteristically childlike’
Location of the reduplicant. Again, in dealing with whole word reduplication, it
is difficult to argue for a specific location of the reduplicant with respect to its base,
whereas in partial reduplication the argument is more straightforward. Wilbur (1973) and
Moravcsik (1978), for example, suggest that in locating the reduplicant in total
reduplication, a cue can be taken from general patterns of affixation in the language, if
available. For example, if the language has suffixes, then one can assume that the
reduplicant is suffixed to the base or if the language has prefixes, that the reduplicant is
prefixed to the base. Marantz (1982) on the other hand, suggests that prefixation is the
unmarked pattern in reduplication. Designating reduplication as prefixing or affixing by
this method is problematic for Creole languages since they tend to have few affixes. For
example, GC has a suffix –ii which attaches to intransitive verbs or adjectives to form
adjectives (Devonish, 2003). JC has a suffix -øp which attaches to verb stems to form a
kind of complex verb e.g. dZUk + op = dZUkop2. When this form reduplicates it can be
either fully copied or partially copied giving different meanings, which overlap in some
cases. By Wilbur’s suggestion, we could characterize JC reduplication in this case as
suffixing reduplication since there are other suffixes in the language. However, by
2
It is not clear at this point what the precise semantics of the new form is. Some informants indicate that the word + op
gives a sense of deliberate action, or repeated action e.g. dZUk ‘to pierce’; dZUkop ‘multiple acts of piercing’;
‘deliberate act of piercing’.
-7-
Marantz’s metric we might characterize it as prefixing reduplication since prefixation is
the unmarked pattern, i.e. observed more frequently cross-linguistically.
Odden and Odden (1985 in McCarthy & Prince 1995:76) report a case of verbal
reduplication in Kihehe that illustrates another way of telling whether the reduplicant is
prefixed or suffixed to the base. The prefixes /n/ and /ku@/ (kw before vowels) appear
before the root, a penultimate H tone occurs in the infinitive. One could argue that the
reduplicant is prefixed to the base since the root tones are not copied to the reduplicant.
In the examples below, from Kihehe, the portion of the reduplicated word that maintains
the prosodic properties of the root (tones) would be the base and the part that does not
would be the reduplicant.
(4) /haata/ ku@-haa@ta ku@haata-haa@ta ‘to ferment’, to start id.’
/ita/ kw-i@ita kwi@ita-kwii@ta ‘to spill, to pour a bit’
/tiitu/ n-i@i@tu niitu-ni@i@tu ‘black, blackish’
The issue of suffixing versus prefixing reduplication is related to the issue of
directionality of copying. Under an OT analysis such as that presented in this
dissertation, the issue of directionality is rendered superfluous in the absence of evidence
such as that presented for Kihehe in (4). This is because OT allows for a two-way
correspondence between base and reduplicant (section 1.6.1). In the analysis of JC
reduplication, I compare processes of reduplication with compounding to see whether and
how the component parts of the words are differentiated phonologically. I show that both
-8-
reduplicated words and compounds have main stress on the right-most foot in the word.
With this, I suggest that while reduplication in JC might be analysed as prefixing
reduplication which attaches the reduplicative morpheme to a prosodic word it need not
be since forms with prefixing and suffixing reduplication are phonetically identical.
Failure of reduplication. As discussed above, though reduplication in JC is a
productive process there are instances where reduplication is expected but does not occur.
I show that in these cases, reduplication does not apply if a potential base is the wrong
foot type or if the base is too large (cf. Gooden, 2003a).
Fixed segmentism. Where a base is augmented to satisfy prosodic foot
requirements as in (2) above, the inserted segment is // in the majority of cases and // in
a few cases. In addition to addressing the facts concerning the appearance of a fixed
segment in characteristic reduplication, the status of these segments with respect to the
general phonology of the language is addressed. I show that the inserted segment serves
both a phonological function of permitting the base to achieve the proper foot size and a
morphological function of denoting the class of characteristic reduplicated adjectives.
Prosodic properties. Gooden (2003a) examined the prosodic requirements for
some classes of reduplication in JC. For example, it was argued that the intensive
reduplicant must be a trochaic foot whereas iterative reduplicants need only be a
bisyllabic foot. Based on the results of the phonetic study in this dissertation, I show that
the prosodic requirements on JC reduplication extend beyond requirements on the well-
formedness of foot structure to include requirements for pitch accent association at higher
levels of prosody, i.e. the prosodic word.
-9-
A secondary goal of this dissertation is to provide preliminary answers to
questions about the realization of prominence in JC. For example, is it tone as in
Saramaccan (Bakker, 1987) stress (LePage, 1960), or a combination of both as has been
proposed for Guyanese Creole (Devonish, 1989) and Papiamentu (Rivera-Castillo,
1998)? I give an overview of the different arguments presented in support of claims for a
stress-based system and a tone-based system in JC. I examine the JC data and establish
criteria for characterizing the prosodic system of the language. While an extensive
description of the prosodic system of the language as a whole is not undertaken, it is
hoped that the description of the prosodic properties of the reduplicated words will shed
light on the wider prosodic system.
1.4 Phonetic Aspects
Another objective of this dissertation is to gather acoustic evidence for the types of
contrasts that have been suggested as differentiating segmentally identical reduplications.
Ultimately, the question that needs to be addressed is the implications of the observed
phonetic differences for the specification of the phonological properties of reduplication.
As mentioned in section (1.3), there are reduplication processes in some CECs
that produce segmentally identical words. The two known processes in JC are intensive
and distributive reduplications. Several researchers have presented data from other
related creole languages suggesting that segmentally identical reduplicated words are
similar only at the segmental level since they have different prosodic qualities (Sylvain,
1936; Dioncie, 1959; Adamson and Smith, 1999; Devonish, 2003). It was noted also that
- 10 -
we do not have sufficient information about the types of phonetic differences. While
suggestions have been made about possible differences in stress or tone, we need to
discover exactly what phonological property we are dealing with in order to properly
understand the nature of the phonetic differences between the reduplicated words.
Therefore, the goal of the phonetic study of JC reduplication is to find out the
exact nature of the phonetic differences between the segmentally identical reduplications.
The dissertation is thus largely a study of the production of these words and focuses on
the prosodic properties of the words rather than on quantitative measures of specific
aspects of the system. The main question is the nature of the prosodic differences
between intensive and distributive reduplications.
The acoustic-phonetic analyses described, results from a two-part experiment
aimed at providing an adequate description of the prosodic properties of JC reduplicated
words. The first part is a pilot perception and production study with one speaker. This
pilot study was used to formulate hypotheses about the phonological characterization of
the phonetic differences between the reduplicated words. This is a combination of
instrumental and auditory observations of the reduplicated words in different sentence
positions and in different types of intonation patterns e.g. statement intonation versus
question intonation. The second part is a replication of the production part of the pilot
study with several other speakers to test whether and how the phonetic properties
observed in the speech of the single speaker differed from those of other JC speakers.
The design of the production study was complicated by the fact that differences in
meaning between segmentally identical reduplicated words are sometimes associated
- 11 -
with different syntactic functions. For example, Migge (2003) argues that there are two
kinds of segmentally identical reduplications in Surinam Eastern Maroon Creoles, which
have different meanings but which are generally used in mutually exclusive syntactic
environments (see section 2.3). There is one exception to this; both forms occur in
prenominal position. In these cases, it is not clear whether the listener relies on
contextual cues rather than prosodic cues to identify the meaning of the word. The
question is what distinguishes such reduplications when they occur in the same syntactic
environments. As I show in Chapter 6, it is possible to keep the syntactic context
constant for the target JC forms to see if these reduplications are differentiated, while
varying another aspect of the sentential forms — the intonation contour as a whole. This
strategy also helps us to determine whether prosodic cues make for a distinction in
meaning in identical syntactic environment or whether a particular meaning is conveyed
with similar prosodic cues in different syntactic environments. Given the complexity of
the matter, the present study addresses only the first issue. Recordings were done of
words in isolation as well as in appropriate sentence frames in several different sentential
contexts and with several different intonation contours. The sentence frames allow both
intensive and distributive reduplications to be freely alternated in the same position
without affecting the grammaticality of the sentence as illustrated in (5)
(5) Im want wan yEloyElo mango an som yam

He wants a yellowyellow mango and some yams.
He wants a very yellow mango and some yams. (Intensive Meaning)
He was a scattered-yellow mango and some yams. (Distributive Meaning)
- 12 -
1.5 Broader Implications
This study is intended firstly as a contribution to our knowledge of the phonology and
phonetics of reduplication in Jamaican Creole but it has wider implications for the study
of creole phonology in general. Cassidy (1961) claimed that Jamaican pronunciation has
greater variation in intonation than American or British English. He carefully noted that
this is an impressionistic description and that for a more accurate picture we need to call
in mechanical aids to “catch the speech with a recording device and afterwards take it
apart laboriously - in short a job for a phonetician…” (pg. 26). In characterizing the
prosodic properties of reduplicated words in JC, this dissertation partially takes up this
forty-two year old mandate, trusting that it is well answered. I provide a prosodic
analysis of the reduplicated words in JC, set against the background of my own
characterization of the prosodic system of the language itself.
This type of study is important since the study of the prosodic systems of creoles
is a largely unexplored area of research. As far as Caribbean English lexified creoles are
concerned, there are large numbers of varieties that are historically connected, but which
are nevertheless distinct. This topic therefore provides fertile soil for comparative work
using prosodic analyses to shed light on discussions about the development of the
prosodic systems of the different CECs and the relationships among them as well as other
issues. The JC reduplication data discussed in this work provides the type of groundwork
needed for this type of research (cf. Devonish, 1989; 2002; 2003). This study also
contributes to our understanding of the relationship between word-level prosody and
morphological structure in creole languages as well as in other languages in general.
- 13 -
1.6 Theoretical Assumptions
In languages that have reduplication, in many cases there is co-dependence on
phonological processes as well. Reduplication involves addition of material as in other
types of affixation, however, the identity of the added material is partially or wholly
determined by the base. Some part of the input is repeated to the left, right, or
occasionally in the middle (Spencer, 1991). In one of the earliest accounts of
reduplication, Wilbur (1973) notes that reduplication resembles a phonological rule, since
statements can be made about the shape, size and location of the reduplicant. In this way,
it can be said to have a structural description and a structural change. However, Wilbur
notes that the rule is determined by grammatical information. Thus, reduplication seems
like some sort of affixation in which a process is applied to the base rather than a mere
concatenation of one morpheme with another, as in ‘typical’ cases of affixation. Since
reduplication has both morphological and phonological properties, its interaction with
morphological and phonological rules is of particular interest to phonologists and
morphologists alike. Exactly how this interaction takes place has been the source of
much debate in the literature. Three of these issues are pertinent to the study of JC
reduplication.
There is the problem of transfer (cf. Clements, 1985). In analyses of
reduplication, there is a problem of accounting for the transfer of information about the
prosodic structure of the base to the reduplicant (copied material). Clements highlights
four types of information which any successful analysis of reduplication must account
for: (a) the size of the copied material; (b) what is copied (the base); and (c) where in
- 14 -
relation to the base the reduplicant will reside, prefixed, suffixed or infixed. Yet another
problem is accounting for invariant material in reduplicated forms, fixed segmentism
(Alderete et al 1999). For example, in JC characteristic type reduplication, an //
consistently surfaces in the reduplicant and base even though it is not present in the input
and as such might be treated as a fixed segment. The final problem is that of location of
the reduplicant with respect to its base. Broselow and McCarthy (1983) for example
observed that reduplicative infixes often attach to a prosodic constituent in Samoan. As I
discuss in Chapter 7, while the location of the JC reduplicant is not crucial to the analysis,
the output forms must be well-formed prosodic constituents.
1.6.1 Optimality Theory
The analyses in this study are done in a Correspondence Theory framework, an offspring
of Optimality Theory (OT) (Prince & Smolensky, 1993; McCarthy & Prince, 1995b).
McCarthy & Prince (1993b, 1995b et seq.) proposed that the reduplicant and its base are
output strings related to each other in a correspondence relationship. This approach
essentially capitalizes on the mate relationship proposed by Wilbur (1973). OT differs
significantly from ruled-based accounts in that there are no rules. Instead, a given surface
form is ‘derived’ from an underlying form by means of a universal set of ranked and
violable constraints on well-formedness. It is the interaction of constraints, which results
in the selection of the actual surface form. Only the candidate which fares the best with
respect to the constraint hierarchy is selected as the output. Therefore, even a candidate
that is completely faithful to the input may be ruled out in order to satisfy some higher
- 15 -
ranked constraint. Further, unlike derivational accounts, constraints on reduplicative
identity are evaluated in parallel with other constraints on the structure of the output, thus
the problem of transfer of prosodic material is immediately overcome.
Correspondence Theory (CT) (McCarthy and Prince, 1995b) places greater
emphasis on the correspondence relations, generalizing over various types of faithfulness
relations. This is formally expressed in (6).
(6) Correspondence Given two strings S1 and S2, correspondence is a function

from the elements of S1 to those of S2. Elements S1 and
S2 are referred to as correspondents of one another when
.
There are separate correspondence relations depending on how S1 and S2 relate to each
other. The constraints on reduplication considered here are of two kinds. The first set
minimizes differences between the input and the output (I-O Faithfulness), as between an
underlying form (S1) and a surface form (S2). The second set minimizes differences
between the base and the reduplicant (B-R Identity). In this case, the base is S1 and the
reduplicant is S2. The inherent claim of this model is that the reduplicant copies the base,
in order to achieve complete identity with it, as is assumed by copy and association
models. Unique to this model, however, is the permissible correspondence between both
portions of the reduplicated word as illustrated in the diagram in (7). This is important in
accounting for cases of so-called backcopying (overapplication), in which the
phonological properties of the base may be imposed on the reduplicant.
- 16 -
(7) Full Model: Input /AfRED + Stem/
I-R Faithfulness I-O Faithfulness

Output R B
B-R Identity
A direct result of correspondence is that there is no intermediate stage, where segmental
material is copied and then erased (Marantz, 1982, Bell 1983, Broselow and McCarthy,
1983) or truncated (Steriade, 1988).
Another aspect of the theory which is relevant for the analysis of JC reduplication
is Generalized Alignment Theory (GAT), which introduces the idea that reduplicative
morphemes are specified for morphological category. GAT’s conceptualization of the
reduplicative morpheme as phonologically empty differs in an important way from
template based approaches in that there are no actual templates. By eliminating templates
from the grammar, GAT seeks explanations elsewhere by making use of a general set of
constraints on the interaction between prosody and morphology. Urbanczyk (2001)
argues that GAT has greater explanatory power than template based accounts since the
latter only explain the shape of reduplicative morphemes and say nothing about
segmental content. Urbanczyk argues that specific aspects of both the shape and
segmental content of a reduplicative morpheme in Lushootseed can be derived from
morphological classification. In this way, differences between reduplicative morphemes
can be related to a difference in morphological classification. In this dissertation, each
reduplicative morpheme is marked for its respective subtype as denoted by the semantic
- 17 -
properties of the output form. For example, REDINT denotes the intensive reduplicant and
REDITER denotes the iterative reduplicant and so on. The phonological analysis shows
that a constraint-based approach that permits two-way correspondence between the base
and the reduplicant, is well suited to account for these reduplication processes.
For the uninitiated reader, I give a brief illustration of an OT analysis in this
section3. In cases of full reduplication, as seen in the creoles discussed in Chapter 2, there
must be complete segmental similarity between the base and the reduplicant, as in
nyamnyam ‘food’ (Sranan). In OT, this is accounted for by means of the Identity
constraint in (8):
(8) Maximise Base-Reduplicant (MAXBR): Every element of the base has a
correspondent in the reduplicant (McCarthy & Prince, 1995b).
A given form incurs a violation (*) of MAXBR if a segment in the base is not also in the
reduplicant. A fatal violation (*!) rules out the form completely. Two possible candidates
generated from the input are shown in Tableau 1.1.
3
For a more detailed description I refer the reader to McCarthy, John. 2002. A Thematic Guide to Optimality Theory:
Cambridge University Press.
- 18 -
/ RED + nyam / MAXBR
! a. nyamRnyamB
b. nyaRnyamB *!
c. nyRnyamB *!*
d nRnyamB *!**
Tableau 1.1: nyamnyam ‘food’
The actual surface form shown in (a) (marked by !) copies the entire base. All of the
segments of the base are seen in the reduplicant, satisfying MAXBR. Candidate (b) does
not have all of the segments of the base in the reduplicant. In particular, the /m/ of the
base does not appear in the reduplicant. Thus, it fails to satisfy MAXBR and is ruled out
as a possible output form. Likewise, candidates (c) and (d) do not satisfy MAXBR
because the fail to copy all the segments of the base. Candidate (c) does not copy the /a/
and the /m/ and candidate (d) does not copy three segments /y/, /a/ and /m/.
The inherent claim of this model is that the reduplicant copies the base in order to
achieve complete identity with it (as is also assumed by Copy and Association models, cf.
Marantz, 1982; Bell 1983, Broselow and McCarthy, 1983; among others). As will be
illustrated in detail in the phonological analysis in Chapter 7, complete identity in JC
reduplication is not always attainable under these terms, since the process is regulated by
prosodic requirements. Unique to this model, however, is the correspondence between
both portions of the reduplicated word, the base and the reduplicant, which will be shown
to be important in accounting for variable shapes of reduplicants in JC.

- 19 -
Another relevant issue is that of fixed segmentism. In OT, cases of fixed
segmentism observed in reduplication are viewed as cases of the emergence of the
unmarked (TETU). The idea is that a phonologically unmarked structure that emerges in
reduplicated forms is not a condition that is necessarily required in the language as a
whole. Unmarked structures are so because they obey the phonoconstraint and marked
structures occur when this constraint is violated. The ranking is as follows:
(9) I-O faithfulness>> Phonoconstraint>> B-R identity (McCarthy and Prince, 1995b)
In JC characteristic reduplication, an // always surfaces as the final vowel of the base and
the reduplicant whether or not it is in the input form. The phonoconstraint would be one
that penalizes the insertion of any other vowel and would be highly ranked since other
vowels are not inserted.
1.6.2 Metrical Theory
In this section, I present an overview of the metrical theory of word stress (Liberman and
Prince, 1977; Halle and Vergnaud, 1987; Hayes, 1995) under which the analysis of JC
stress is subsumed. While these approaches differ with respect to the formal
representation of stress, they all share a common assumption that stress is a rhythmic
phenomenon encoded in the strong-weak relationship that exists between syllables in a
word. Metrical theory is essentially a study of the relative prominence of syllables and
higher-level units of prosody such as the foot, which is a rhythmic unit (Kager, 1995;
- 20 -
1999). Based on the properties of stress systems, metrical theory claims that the
alternating prominence (rhythm) and sensitivity to syllable weight are among the most
important determinants of stress patterns in languages. The potential conflicts between
the need to have alternating stress, the need to match syllable weight with syllable
prominence (quantity-sensitivity) and the need to mark the edges of morphological
domains with metrically strong syllables (edge-marking), are very well suited for the OT
analysis which I propose for the JC stress system.
In determining stress patterns in a language, syllables must be organized into feet.
We must then determine whether these feet are right-headed (iambic) or left-headed
(trochaic), headedness being determined by whether the stressed syllable is rightmost or
leftmost in the foot. The strong syllable in a foot is referred to as the head and the weak
syllable as the non-head. These relationships are illustrated below.
(10) (a) Right-Headed (b) Left-Headed (c) Degenerate
F F F
w s s w
σ σ σ σ σ
Iambic Trochaic
As shown in (10), feet are parsed into binary units of syllables. This is based on an
assumption of metrical theory that feet are only of two maximal sizes, binary or
unbounded. We will see later, that under a moraic theory analysis feet can also be
- 21 -
comprised of a single syllable with two moras. Structures like that in (10c), have only
one syllable and are referred to as degenerate feet. To give a concrete example, let us
consider the JC word pokomienya ‘revivalist cult’. We can divide the word into two
metrical feet of two syllables each as follows:
x Word level
(x ) ( x ) Foot level
x x x x Mora/syllable level
po ko mie nya
Note that we can talk about a strong-weak relationship between feet within a prosodic
word as well as between syllables within a foot. So, at the foot level, the syllables po and
mie are metrically stronger than the syllables ko and nya respectively. Likewise at the
prosodic word level the foot mienya is metrically stronger than the foot poko. The idea
of a maximal size for feet also incorporates the concept of extrametricality.
Extrametricality designates a particular prosodic constituent as invisible for ‘rules’ of
stress assignment, (Hayes, 1995: 57). That is, stress is assigned without considering the
extrametrical entity.
In its representation of stress, metrical theory assumes a set of universal prosodic
categories based on the prosodic hierarchy. The premise of the prosodic hierarchy is that
every prosodic category, mora, syllable, foot, prosodic word, has as its head an element
of the next higher level category, (Selkirk, 1984).
- 22 -
Prosodic Hierarchy
IP Intonational Phrase
PhP Phonological Phrase
PrWd Prosodic Word
Ft Foot
σ Syllable
μ Mora
In talking about syllable weight, I will refer to Hayes’ (1989) theory of syllable weight
and quantity i.e. moraic theory. Rules of stress assignment tend to make a distinction
between syllables that are heavy and those that are light. According to moraic theory,
syllable quantity is a function of the number of weight-bearing units or moras in a
syllable. It is important to note that in evaluating the weight of a syllable, the segment
count is not important since a light syllable may have more segments than a heavy one.
For example, in English pri of ‘prison’ has three segments and is light but the syllable it
has two segments and is heavy. Under moraic theory, a short vowel has a single mora
and a long vowel has two moras. We will see in Chapter 4 that in the JC case, this
distinction is better captured as a distinction between a non-branching versus a branching
rhyme since diphthongs are also treated as heavy. Languages differ on whether or not
they allow postvocalic consonants to bear a mora and thus contribute to syllable weight.
Hayes (1989) refers to this as a principle of syllabification, weight-by-position, that

- 23 -
assigns a mora to a postvocalic consonant in a language. In example (11), coda
consonants contribute to syllable weight, so that CVC, CVV and CVVC sequences all
count as heavy syllables, having two moras each.
(11) CV - Light CVC, CVV, CVVC - Heavy

(a) σ (b) σ σ σ
μ μ μ μ μ μ μ
C V C V C C V C V C
([CV:]) ([CV:C])
(adapted from Hayes, 1995:52)
In some languages, only CVV and CVVC count as heavy syllables while CV and CVC
syllables are treated as light (see (12)).
(12) CV, CVC - Light CVV, CVVC - Heavy

(a) σ σ (b) σ σ
μ μ μ μ μ μ
C V C V C C V C V C
([CV:]) ([CV:C])
(adapted from Hayes, ibid:52)
- 24 -
The ideas of metrical structure play an important role in prosodic morphology and
intonational phonology. The central tenet of the prosodic morphology hypothesis is that
word formation processes, including reduplication, make use of authentic units of
prosody (McCarthy and Prince, 1993b; 1995a). In chapter 7, I show that JC reduplication
has the ‘foot’ as its prosodic target. This is important to the discussion of JC stress since
according to (Hayes, 1995), the kind of foot required in a language’s morphological
system is the same that is required of its stress system. In notating the different foot types
I follow the moraic theory notation, using L for light feet and H for heavy feet.
With respect to intonational phonology, pitch accents and boundary tones are seen
as components of prosodic constituents. For example, boundary tones are associated with
the edge of phrases defined by the prosodic constituent intonational phrase, while pitch
accents may be associated with the syllable or the mora constituent.
1.7 Structure of the dissertation
The dissertation is outlined as follows. Chapter 2 gives a detailed description of the data
outlining the different types of reduplication observed in JC that need to be accounted for
in the phonological analysis. It also reviews previous studies on reduplication in creole
languages in general, but focuses especially on descriptions of reduplication in Caribbean
English-lexicon Creoles. This places the descriptions of JC reduplication in the context
of reduplication processes in Caribbean English-lexicon Creoles in general.
Chapter 3 provides background information relevant to the study of Jamaican
Creole. This includes a historical background of the language, a description of the

- 25 -
contemporary language situation and the rural community chosen for this study. The
chapter also includes details of the methods of data collection. Methods used to gather
sociodemographic data as well as methodologies employed in the production study are
described.
Chapter 4 provides a general description of JC phonology. Drawing on phonetic
evidence the chapter provides a characterization of the JC prosodic system as a stress-
based system. Former treatments of JC prosody are also reviewed along with studies on
the acoustic properties of segmentally identical words. The purpose is to compare
strategies used to discover acoustic cues to contrast in stress-accent systems like that
proposed for JC.
Chapter 5 presents an overview of the different stress patterns and gives a formal
analysis of the stress system in an Optimality Theoretic framework.
Chapter 6 presents the phonetic analysis of word-level prosody of reduplicated
words. This is a detailed description and analysis of reduplicated words in different
sentential contexts. The results provide evidence for the prosodic differences between
distributive and intensive reduplications and also fortify the characterization of word-
level prosody in JC as stress-accent.
Chapter 7 provides an Optimality Theoretic account of the different types of
reduplications taking into account the issues raised in Chapter 6 regarding the prosodic
properties of intensive and distributive reduplications. An assessment is made of the
impact of prosodic differences on the reduplication process. Finally, Chapter 8 presents
- 26 -
the concluding remarks and a summary of the analyses and discussions presented in
previous chapters.
- 27 -
CHAPTER 2
REDUPLICATION IN JAMAICAN CREOLE AND OTHER CARIBBEAN ENGLISH
CREOLES
2.1 Introduction
This chapter summarizes and evaluates previous studies on reduplication in creole
languages in general but focuses especially on descriptions of reduplication in Caribbean
English-lexicon Creoles (CEC). The discussion is organized around four main issues: the
semantic properties of reduplicated words; the types of processes involved in CEC
reduplication, i.e. full versus partial reduplication; the use of word-level prosody to
contrast the meanings of segmentally identical reduplications; and finally, the
classification of the processes according to their morpho-phonological function. I show
that the types of reduplication processes observed in Jamaican Creole (JC) as well as the
range of meanings expressed, are also seen in other CECs.
Reduplication in CECs produces words with different meanings that are
classifiable into six different semantic classes: stative, intensive, iterative, distributive,
attenuative and characteristic. In stative reduplication, the reduplicated word conveys the
sense that the quality referred to by the base has been persisting; for example,
- 28 -
taaku ‘be evil’ reduplicates as taakitaaku ‘to be in an evil state’ (Ndyuka). In distributive
reduplication, the meaning of the reduplicated word has the sense of scattered, all over
the place, here and there or occasionally, for example, naasti ‘nasty’ reduplicates as
naastinaasti ‘nasty on occasion’ (Guyanese Creole). In iterative reduplication, the
meaning of the reduplicated word describes action performed repeatedly, for example,
‘steal’ reduplicates as ‘to steal repeatedly’ (JC). In intensive reduplication the
degree of quantity (noun), the degree of action (verb) or the degree of quality (adjective)
has more intensity or emphasis in the reduplicated word than in the unreduplicated word,
for example,  ‘busy’ reduplicates as  ‘very busy’ (Sranan). Attenuative
reduplication results in a weakening of the meaning expressed by the input word. For
instance, a reduplicated verb would mean ‘to perform an action to a small degree’, an
adjective would yield the sense of ‘possessing a quality to a small degree’ and nouns
would give a ‘diminutive’ meaning. For example,  ‘long’ reduplicates as
‘longish’ (Sranan). Finally, in characteristic reduplication, the meaning
expressed by the reduplicated form incorporates the feature X, where X is a semantic
component of the base, for example  ‘to eat’ reduplicates as  ‘inclined to
eat’ (JC).
Reduplication in these languages involves full reduplication in the majority of
cases but there are also cases of partial reduplication (Gooden, 2003a; Kouwenberg and
LaCharité, 1998; Sebba, 1981). All the languages in this survey have full reduplication,
e.g. Gullah, Sranan, and some have partial reduplication as well, e.g. Sranan, Ndjuka,
Jamaican. None of the languages have only partial reduplication, which is not a
- 29 -
surprising finding given Moravscik’s (1978) typological survey of languages that have
reduplication. According to Moravscik, if a language has partial reduplication, by
implication it also has full reduplication.
Several CECs in this survey are shown to make use of differences in word-level
prosody to distinguish between segmentally identical reduplicated words, e.g. Sranan,
Guyanese, Jamaican.
I first provide a detailed description of the reduplication processes in Jamaican
Creole in section 2.2. In subsequent sections, I discuss how these processes are manifested
in other CECs. I discuss the semantic properties of reduplicated words in section 2.3.
Section 2.4 reports on the occurrence of full versus partial reduplication processes observed
in the languages, then in section 2.5 I discuss the interaction of word-level prosody with
processes of reduplication. In 2.6 I survey the classification of the reduplication processes
in terms of their morpho-phonological function in each language and summarize the main
points of the chapter in section 2.7.
2.2 Reduplication in Jamaican Creole
In this section, I discuss the phonological properties of reduplicated words in Jamaican
Creole (JC) which serves as the basis for a formal Optimality Theoretic (OT) analysis in
Chapter 7. Unless indicated otherwise, these data are based on my own fieldwork or are
from Cassidy and LePage (1967). First, I examine five processes of productive
reduplication i.e. intensive (two subtypes), iterative, distributive and characteristic, then I
- 30 -
look at one process which is not as productive or which is arguably obsolete, i.e. scalar
reduplication. In the final section, I give an overview of the different phonological
properties of JC reduplication which need to be accounted for in the phonological
analysis.
Cassidy (1957) made early observations on reduplication processes in JC. Cassidy
viewed a reduplicated form as a single combined form with a close juncture, usually with
reduced stress on the second portion of the word. Citing a database of 175 forms, he
identified reduplication as a productive process of word formation in JC. Cassidy
reported that of the 175 forms, nouns were the most numerous (47%) followed by
adjectives (34%), verbs (16%) and adverbs (3%). All the cases mentioned are cases of
full reduplication, applying to all word classes both within the same category and
changing from one category to the next. Examples are shown in (13)
(13)(a) wasswass ‘wasps’ Noun

(b) talkytalky ‘to talk a lot’ Verb
poppop ‘to break all over’
(c) foolfool ‘foolish’ Adjective
tieftief ‘thieving’
weewee ‘very small’
(d) plentyplenty ‘plentifully’ Adverb
(e) boutbout ‘round about’ Preposition
According to Cassidy, the meaning of the reduplicated form depends on the meaning of
the base and he identified the following four meanings: iterative, repetition or continuous
action; distribution; intensification - with adjectives; and attenuation (weakening of
qualities).
- 31 -
Based on analyses presented in (Gooden, 2003a) I classify the different patterns
of reduplication in terms of the meanings they express. In this way, each semantic class
defines a distinct subtype. As I discuss further below, what is problematic is that some of
the words are segmentally identical but have different meanings. Reduplication in these
cases conveys either a distributive meaning or an intensive meaning. Examples are given
in (14) to illustrate the contrast and additional examples are discussed in sections 2.2.3
and 2.2.4.
(14) swEl ‘swollen’ swElswEl

‘swollen in different areas (distributive);
very swollen’ (intensive)
gomI ‘gummy/sticky’ gomIgomI ‘gummy/sticky all over (distributive); very
gummy’ (intensive)
In following chapters, I focus on the class of intensive and distributive reduplicated
adjectives. I review phonological and phonetic evidence demonstrating that these
reduplications are distinguished prosodically. Further, I argue that this prosodic
difference is a key aspect of the semantic interpretation of the reduplicated words.
2.2.1 Iterative Reduplication
In iterative reduplication, the meaning of the reduplicated word describes an action
performed repeatedly. The base for iterative reduplication is either a complex stem
formed from a root and a suffix, e.g. blakop ‘to blacken’ or a simple stem, e.g. kot ‘to
cut’ comprised of just a root.
- 32 -
(15) Root + suffix = Stem Reduplicated Form
(a) bU°odop ‘to seal with boards’ bU°odopbU°odop ‘to seal with boards repeatedly’
pItop ‘to spit’ pItoppItop ‘to spit repeatedly’
maakop ‘to mark on’ maakopmaakop ‘to mark on repeatedly’
blakop ‘to blacken’ blakopblakop ‘to blacken repeatedly’
kotop ‘to cut into pieces’ kotopkotop ‘to cut into pieces repeatedly’
(Cassidy, 1957)
(b) tiif ‘to steal’ tiiftiif ‘to steal repeatedly’
(Kouwenberg and LaCharité,
1998)
kot ‘to cut’ kotkot ‘to cut repeatedly’
(c) tSambaop ‘to cut/chop *
coarsely’
njakaop ‘to cut/chop crudely’ *
kjatao°ut ‘to scatter *
(deliberately)’
As shown in (15) the input forms are both simple and complex stems. The iterative
reduplicant copies its entire base which is equivalent to the stem in this case, giving the
interpretation of action done repeatedly. In the cases in (15a) where reduplication is
observed, the stems are bisyllabic; the monosyllabic forms in (15b) are also reduplicated.
The forms in (15c) on the other hand, with trisyllabic stems, have no reduplicated forms.
2.2.2 Characteristic Reduplication
This type of reduplication expresses the characteristic nature of the base form. That is,
displaying the characteristic X, where X refers to semantic properties of the base
(Kouwenberg & LaCharité, 2001; Gooden, 2003a). The input forms are nouns, verbs or
adjectives but the output form is always an adjective. This type of reduplication may also
be seen as full reduplication in the sense that the base and the reduplicant are identical.
- 33 -
However, a segment not seen in the input form is present both in the reduplicant and in the
base. This will be shown to be directly related to the claim that this type of reduplication
process is prosodically based. As I show in Chapter 7, the inserted vowel also has a
morphological function of marking these types of adjectives. As seen below, the
reduplicants in (16a) and (16b) are bisyllabic. Notice however, that the stems
corresponding to the bases of the forms in (16a) are also bisyllabic, whereas those in (16b)
are not. The forms in (16c) also have bisyllabic stems but have no corresponding
reduplicated forms. Likewise the form in (16d) is trisyllabic and has no reduplicated form
either.
(16)
(a) grIEnI ‘grainy’ grIEnI-grIEnI ‘having grainy, coarse
characteristics’
naasI ‘nasty, filthy’ naasI-naasI ‘having nasty, filthy
characteristics’
(b) laaf ‘to laugh’ laafI-laafI ‘inclined to laughter’
dZuk ‘to pierce’ dZukI-dZukI ‘prickly or needle-like’
blak ‘black’ blakI-blakI ‘having black spots or areas’
bwaI ‘boy’ bwaII-bwaII ‘characteristically boyish
(c) lIzad ‘lizard’ *lIzadI-lIzadI
tSupId ‘stupid’ *tSupIdI-tSupIdI
arIndZ ‘orange’ *arIndZI-arIndZI
(d) pIkInI ‘child’ *pIkInI-pIkInI
A suitable phonological account needs to explain why a vowel is inserted in some words
and not in others, why the inserted vowel appears in both the base and the reduplicant,
why the inserted segment does not appear in the all reduplicated forms, and also why
there are no corresponding reduplicated words for some input stems.
- 34 -
2.2.3 Intensive Reduplication
In intensive reduplication the degree of quantity (noun), the degree of action (verb) or the
degree of quality (adjective) has more intensity or emphasis in the reduplicated word than
in the unreduplicated word (Huttar and Huttar, 1997). The input form is either a
morphologically complex stem, e.g. grIn Æ grIngrInop ‘to smile a lot with someone’ or
is a simple stem e.g. swiit Æ swiitswiit ‘very sweet’.
2.2.3.1 Complex Inputs
Inputs to this type of reduplication are morphologically complex stems with the suffixes
op, out or aaf, which give the interpretation of an action done intentionally or of an
accomplished action (Gooden, 2003a). Notice that the suffix is not copied in any of the
reduplicated forms. The result is that in each case the reduplicant is exactly one syllable
smaller than the stem, i.e. partial reduplication.
(17) Root + suffix=Stem Reduplicated Form

(a) grIn-op ‘to grin…’ grIngrInop ‘to smile with someone a
lot’
blak-op ‘to make black’ blakblakop ‘to make very black’
bU°od-op ‘to seal with boards ’ bU°odbU°odop ‘to seal completely with
boards’
baks-op ‘to hit in the face..’ baksbaksop ‘to hit hard in the face’
laaf-op ‘to laugh with’ laaflaafop ‘to laugh with a lot’
INk-op ‘to soil/cover with ink ’ INkINkop ‘to soil/cover with ink a
lot’
a°Il-op ‘to soil/cover with oil ...’ a°Ila°Ilop ‘to soil/cover with oil a
lot’
krI°Ep-o°ut ‘to scrape out’ krI°EpkrI°Epo°ut ‘to scrape out...’
wa°Ip-o°ut ‘to wipe out’ wa°Ipwa°Ipo°ut ‘to wipe out…’
njam-o°ut ‘to eat some of’ njamnjamo°ut ‘to eat some of…’
brok-o°ut ‘to break out’ brokbroko°ut ‘to break out completely’
kri°Ep-aaf ‘to scrape off’ kri°Epkri°Epaaf ‘to scrape off completely’
- 35 -
wa°Ip-aaf ‘to wipe off’ wa°Ipwa°Ipaaf ‘to wipe off completely’
njam-aaf ‘to eat completely’ njamnjamaaf ‘to eat completely…’
(b) naasI-op ‘to make filthy ...’ naasInaasIop ‘to make completely
filthy’
plaasta-op ‘to daub/plaster ...’ plaastaplaastaop ‘to daub/plaster ...’
dotI-op ‘to soil ...’ dotIdotIop ‘to soil...’
(c) arI°EndZ-op ‘to arrange things’ *
sIdoN-op ‘to sit upright/sit relaxed
for an extended period’
maalIhak-op ‘to make a mess/ to ruin’ *
batabruz-op ‘to bruise extensively...’ *
Where the stem is bisyllabic, the reduplicant is monosyllabic (17a); where the stem is
trisyllabic the reduplicant is bisyllabic (17b). The forms in (17c) show trisyllabic and
quadrasyllabic stems which do not have corresponding reduplicated forms. The fact that
there is no reduplication in these instances raises interesting questions. This cannot
simply be attributed to a prohibition on copying the suffix in reduplication, since iterative
reduplication illustrates that some input forms with these suffixes do undergo
reduplication. With regard to this type of reduplication, we need to account for (a) how
much of the base is copied, (b) why the reduplicant only partially resembles the input,
and (c) why some inputs do not have reduplicated forms.
- 36 -
2.2.3.2 Simple Inputs
As seen in the examples in (18), single morphemes may also serve as inputs to intensive
reduplication.
(18) Stem Reduplicated Form

(a) blak ‘black’ blakblak ‘very black’
swEl ‘swollen’ swElswEl ‘very swollen’
tINk ‘foul scented’ tINktINk ‘very foul scented’
laN ‘long’ laNlaN ‘very long’
fUl ‘full’ fUlfUl ‘very full’
lU°o ‘low’ lU°olU°o ‘very low’
slU°o ‘slow’ slU°oslU°o ‘very slow’
wayt ‘white’ waytwayt ‘very white’
rayp ‘ripe’ rayprayp ‘very ripe’
laas ‘lost’ laaslaas ‘very lost’
grin ‘green’ gringrin ‘very green’
ril ‘real’ rilril ‘very real’
swit ‘sweet’ switswit ‘very sweet’
ful ‘fool’ fulful ‘very foolish’
flat ‘flat’ flatflat ‘very flat’ (Cassidy, 1957)
nof ‘a lot’ nofnof ‘more than enough’ (Kouwenberg
and LaCharité, 2001)
(b) maaga ‘skinny’ maagamaaga ‘very skinny’
grIenI ‘grainy’ grIenIgrIenI ‘very grainy’
gomI ‘gummy’ gomIgomI ‘very gummy’
oglI ‘ugly’ oglIoglI ‘very ugly’
jala~jElo ‘yellow’ jElojElo ‘very yellow’
taakI ‘to talk’ taakItaakI ‘to talk a lot’ (Cassidy, 1957)
plEntI ‘plenty’ plEntIplEntI ‘plentifully’ (Cassidy, 1957)
Notice that the words in (18a) have monosyllabic stems while those in (18b) have
bisyllabic stems. In both cases, there is full reduplication. As I discuss just below in
section 2.2.4, this subtype of intensive reduplication produces words that are segmentally
identical to words with a distributive meaning. In Chapter 6 I show that these two types
- 37 -
of reduplications are different prosodically and as such we need to account for the nature
of the differences between the two types of reduplications.
2.2.4 Distributive Reduplication
In distributive reduplication, the meaning of the reduplicated word has the sense of
scattered, all over the place, here and there or occasionally. Of interest is the observation
that there are some reduplicated forms which have identical segmental shapes to those
seen above in section 2.2.3, which have an intensive meaning. In the examples in (19),
all the input forms are comprised of a single morpheme and are copied completely.
Distributive reduplications that are derived from a monosyllabic base appear to take two
forms: bisyllabic and quadrasyllabic. All the output forms in (a) are bisyllabic and all the
forms in (b) are quadrasyllabic with /I/ as the final vowel of the base and the reduplicant
in the majority of cases and /a/ occasionally. Notice that the quadrasyllabic forms in (c)
are derived from monosyllabic bases and have the same meaning as bisyllabic forms in
(a), e.g. waytwayt or waytIwaytI meaning ‘white all over’.
(19)Stem Reduplicated Form

(a) blak ‘black’ blakblak ‘black spots all over’
swEl ‘swollen’ swElswEl ‘sort of swollen/swollen in spots’
tINk ‘foul scented’ tINktINk ‘sort of foul scented’
laN ‘long’ laNlaN ‘long in parts’
slU°o ‘slow’ slU°oslU°o ‘sort of slow/slow at times’
wayt ‘white’ waytwayt ‘whitish all over’
rayp ‘ripe’ rayprayp ‘sort of ripe/ripe in spots’
laas ‘lost’ laaslaas ‘sort of lost’
daag ‘dog’ daagdaag ‘scattered dogs/many different dogs’
bwaI ‘boy’ bwaIbwaI ‘various kinds of boys’
- 38 -
grin ‘green’ gringrin ‘green spots all over/greenish all over’
swit ‘sweet’ switswit ‘sweetish all over’
bluo ‘to blow’ bluobluo ‘toy whistle that makes intermittent
noises’
krIEp ‘to scrape’ krIEpkrIEp ‘scrapings’4
(b) grIEnI ‘grainy’ grIEnIgrIEnI ‘ grainy all over’
gomI ‘gummy’ gomIgomI ‘ gummy all over’
oglI ‘ugly’ oglIoglI ‘ ugly all over/in parts’
*mInI mInImInI ‘scattered spots seen when dizzy’
jala~jElo ‘yellow’ jElojElo ‘yellowish all over’
haafa ‘halved’ haafahaafa ‘in halves’
pisa ‘spliced’ pisapisa ‘in pieces’
*chaka ‘untidy’ chakachaka ‘untidy everywhere’
*nyaka ‘jagged cuts’ nyakanyaka ‘jagged cuts all over’
(c) blak ‘black’ blakIblakI ‘black spots all over’
grin ‘green’ grinIgrinI ‘green spots all over’
swit ‘sweet’ switIswitI ‘sweet ingredients all over’
wayt ‘white’ waytIwaytI ‘white spots all over’
pop ‘to break’ poppop ‘to break all over’
While a detailed semantic description of the meanings is beyond the scope of this
work, some observations are notable. The distributive meanings expressed can be
grouped into semantic types such as physical property (e.g. rayp ‘ripe’), dimension (e.g.
laN ‘long’) colour (e.g. blak ‘black’). We might also consider that there are some words
whose lexical aspect makes them intrinsically distributive e.g. grain(y). If this is the case,
then the meaning of the reduplicated item is necessarily distributive. However, some of
these forms can have an intensive meaning as well. For the intensive interpretation,
grIEnIgrIenII for example could be glossed as ‘an extraordinary amount of grain/granular
particles’. It is perhaps also worth mentioning that the morphosyntactic distinction
between true adjectives, which modify nouns, and predicate adjectives which can
function as intransitive verbs (Winford, 1997; Migge, 1998), is not relevant here since the
4
Note that these are by nature distributed over a given area and so are legitimately classifiable as distributive.
- 39 -
main foci are the phonological and phonetic properties of the words. However, I consider
these distinctions in the construction of materials for the production study (see Chapter
6).
2.2.5 Other Types
DeCamp (1974) describes other types of reduplication in JC which have been said to
have vowel harmony related to the semantic interpretation of the words. I will refer to
these as scalar reduplication since the vowel alternations are said to correspond to
semantic distinctions in the size and scale of the object or activity. I include these here
for the sake of completeness but will not describe them in the formal analysis since they
are not consistently recognized by my informants, if at all. Some of these reduplicated
words are, thus still in use but cannot always be related to productive bases or often times
to other members of the class as described by DeCamp. These like other reduplicated
words which lack ‘productive’ bases arguably have strictly lexical uses, and are treated
here as peripheral to synchronic reduplication processes such as those I focus on in this
dissertation.
(20)a. maka maka ‘mud’

moko moko ‘thick mud’
meke meke ‘thin, watery mud’
b. mak mak — less intensive than the forms in (a).

mok mok
mek mek
- 40 -
c. maki maki — familiar or jocular versions of the forms in (a).
moki moki
meki meki
(21) a. taga taga ‘to drag’

togo togo ‘to drag something heavy’
tege tege ‘to drag something light’
b. laga laga ‘to carry, lift’

logo logo ‘to carry, lift something heavy’
lege lege ‘to carry, lift something light’
(22) graNgi graNgi ‘brushwood, sticks’

*groNgo groNgo
*greNge greNge
*graNga graNga
In all the examples seen above in (20)- (22), the forms are fully reduplicated. That is, the
reduplicant is completely identical to the base. DeCamp (1974) and Alderete (1993) note
that the absence of non-high vowels in these forms correspond to the observed semantic
distinctions. Notice also that all the input forms are no more than two syllables long. In
addition, all the words that allow for scalar meanings are of the shape CVCV. The form
in (22) with CVCCV segmental shape does not undergo scalar reduplication. Alderete
(1993) proposes that the segmental shape of the input forms is related to the
permissibility of scalar reduplication. The restriction is that the process applies within the
domain of a foot, more specifically a bimoraic foot. Since the form in (22) is trimoraic,
non-high vowel harmony does not apply.
Alderete’s paper is one of the first (to my knowledge), to apply a prosodic
morphology approach to the study of JC reduplication. One issue which is not addressed
in sufficient detail, however, is the nature of the data described. Based on data from
- 41 -
Kouwenberg and LaCharité, (1998 et seq.) and Gooden (1999 et. seq), it is clear that
many of these forms are no longer in contemporary use by speakers and more
importantly, some processes like the process of vowel-harmony are not observed often if
at all. In fact, I would argue that this is not a process of vowel harmony at all but an
effect of full reduplication. It is significant that the vowels in the input word are identical
to those in the output form. The example in (22) grangigrangi supports this alternative
view since vowel harmony would predict either *grangagranga or *gringigringi, neither
of which is attested.
2.2.6 Discussion
This section gave a detailed description of the data outlining the different types of
reduplication observed in JC. In the majority of cases observed reduplication involved
complete copy of an input form, i.e. full reduplication. However, we observed one case
of incomplete copy of the input form and another case in which the entire input form was
copied and additional segmental material was added. Additionally, there were cases in
which no reduplication was observed. An adequate phonological analysis of JC
reduplication needs to account for the differences and similarities between each subtype
of reduplication; that is, the shape and size of the copied material (reduplicant), the
presence of invariant segments in the reduplicant and the base which are not observed in
the input, the failure of reduplication in some words, the location of the reduplicant
(prefixed, suffixed) and finally, the nature of the prosodic differences between
segmentally identical reduplicated forms.
- 42 -
As noted above, while JC reduplication expresses a range of meanings, there are
some processes which produce segmentally identical forms. In particular intensive
(simple input) reduplications seen in section 2.2.3.2 are identical to distributive
reduplications seen in 2.2.4. I propose that similarity between the two types of
reduplications is a result of differences in the prosodic properties of the words, as I show
in Chapter 6.
2.3 The Semantics of CEC Reduplication
The semantic notions expressed by reduplicated forms in JC are also expressed by
reduplicated forms in other CECs. In the following sections, I give examples of
reduplicated words from different CEC languages that express the different meanings.
2.3.1 Iterative
Kouwenberg and LaCharité, (1998 et seq.) examined patterns of reduplication in several
Caribbean Creole languages. They include iterative reduplication among the types
discussed. They develop a distinction between inflectional reduplication which they
argue is iconic and derivational reduplication which they argue is non-iconic.
Kouwenberg and LaCharité argue that inflectional reduplication in verbs marks a
subcategory of imperfective aspect, i.e. iterative, as in the examples in (23).
(23) Saramaccan bia ‘to turn’ biabia ‘to wind (of river)’
Sranan taki ‘to speak’ takitaki ‘to gossip’
Ndjuka suku ‘to look for’ sukusuku ‘to keep looking’
- 43 -
Sebba (1981) also reports this type of reduplication in Sranan; some examples are
shown in (24).
(24) waka ‘to walk’ wakawaka ‘walk up and down, saunter’

tan ‘to stay’ tantan ‘stay intermittently’
Other cases of iterative reduplication are reported for Saramaccan. Baker (1987) claims
that iterative meaning is always expressed through reduplication as shown in (25).
(25) jamba ‘to walk’ jambajamba ‘to walk up and down’

bendi ‘to bend’ bendibendi ‘to totter’
2.3.2 Intensive
Another semantic subtype common to CECs is intensive. Huttar and Huttar (1997),
examine several processes of reduplication in Ndjuka, including one they refer to as
‘augmentation’, which is somewhat comparable to intensive reduplication, and other
types that express approximation. In many cases, the explanations of the meanings
expressed by reduplication are precise and clear. However, there is a problem with their
account of the category ‘augmentation’. In particular, the meanings they subsume under
this label in fact belong to different semantic types. Included in the category,
‘augmentation’, are meanings of distribution, as well as intensification and duration5.
Dividing this semantic category, augmentative, into distributive and intensive, provides a
better classification of the types of meanings expressed by reduplication.
5 Duration is said to be expressed by iteration rather than reduplication.

- 44 -
The intensive meaning gives a greater quantity or greater degree of quality or
more intensity or emphasis in the reduplicated form than in the unreduplicated form. The
syntactic categories involved include time adverbials (26a), numerals (26b), adjectives
(26c), and adverbs (26d).
Ndjuka
(26) a. dyonso ‘soon, recently’ dyonsodyonso ‘very soon, very recently’
fuuku ‘early’ fuukufuuku ‘very early’
b. dunsu ‘thousand’ dunsudunsu ‘many thousands’
c. pikin ‘small’ pikinpikin ‘myriad small holes’
gaan ‘big’ gaangaan ‘very big’
duungu ‘drunk’ duunguduungu ‘very drunk’
d. saafi ‘slowly’ saafisaafi ‘very slowly’
Kouwenberg and LaCharité (2001) include meanings of emphatic, intensive or
augmentative used in the literature under the inflectional category for adjectives.
However, all these meanings can be accommodated for by the term intensive as
illustrated by the examples in (27).
(27) Saramaccan langa ‘long’ langalanga ‘very long’

Sranan bisi ‘busy’ bisibisi ‘very busy’
Ndjuka tuu ‘true’ tuutuu ‘very true’
Hopkins (2000) reports a few cases of intensive reduplication in Gullah which
applies to both nouns and verbs as seen in (28).
(28) baN ‘a loud noise’ baNbaN ‘a very loud noise’

dE ‘there’ dEdE ‘exactly there’
fIl ‘to fill’ fIlfIl ‘to fill entirely’
waIt ‘white’ waItwaIt ‘very white’
- 45 -
In Sranan, reduplicated adjectives that are derived from adjectives intensify the
meaning of the input form as in (29) (Sebba, 1981).
(29) (a) bisi ‘busy’ bisibisi ‘very busy’

bruya ‘confused’ bruyabruya ‘very untidy’
libi ‘living’ libilibi ‘very lively’
(b) krin ‘clean’ krinkrin ‘entirely’
wan ‘one’ wawan ‘alone, only’
The forms in (29) can be classified as intensive in some sense if we see intensive as
including entirety and completeness as parts of its meaning.
Similar reports have been given for Saramaccan as seen in the examples in (30),
(Bakker, 1987).
(30) langalanga ‘too long’

bumbunu ‘very good/holy’
wante ‘right now’ wantewante ‘immediately’
awaa ‘now’ awaaawaa ‘at this very moment’
djunsu ‘soon, recently’ djunsudjunsu ‘just now, very soon’
2.3.3 Characteristic
The third type of meaning expressed in CEC reduplication is the characteristic meaning.
According to Kouwenberg and LaCharité (2001), the reduplicated word in these cases
conveys an ‘X-like quality’, where X denotes the meaning of the base. The specific
meaning of the reduplicated word depends on the word class of the base, i.e. noun,
adjective or verb. If the base is a noun, the interpretation is attribution of the presence of
the object denoted by the noun as a typical characteristic, or similarity to the object as in
- 46 -
(31). If that base is an adjective, the interpretation is ‘similarity to the quality described
by the adjective’ as in (32) and finally, if the base is a verb, the interpretation is
attribution of the activity described by the verb as a characteristic activity.
(31) Saramaccan baafu ‘soup’ baafubaafu ‘souplike’

Sranan tifi ‘tooth’ tiifitiifi ‘cogged, indented’
frowsu ‘whim’ frowsufrowsu ‘capricious’(Sebba, 1981)
(32) geligeli ‘yellowish

guuunguuun ‘greenish’
baafubaafu ‘souplike’
puilipuili ‘powderlike’ ( Bakker, 1987)
2.3.4 Distributive
Devonish (2003) makes similar reports of a type of x-like reduplication in his discussion
of Guyanese Creole (GC) reduplication. However, in this case the meaning of the
reduplicated word has a distributive interpretation. Devonish makes a distinction between
compounding reduplication and non-compounding reduplication. This distinction is
partly semantic and partly phonological. I discuss the semantic distinctions here and the
phonological distinction in section 2.5 when I discuss the prosodic properties of CEC
reduplication. Compounding reduplication may have a derived base. Devonish argues
that a derivational morpheme –ii attaches to intransitive verbs or adjectives to form
adjectives with a distributive meaning. The meaning is ‘having the feature of X, all over’
where X is the base. Further, the precise meaning of the reduplicated word is dependent
- 47 -
on the lexical category of the input. The forms in (33a) have verb inputs; the forms in
(33b) have adjective inputs and the forms in (33c) have noun inputs.
(33) (a) pik ‘picked’ pikii ‘as if picked’ pikiipikii ‘as if picked all over’
juk ‘pierced jukii ‘as if pierced’ jukiijukii ‘as if pierced all
(b) wait ‘white’ waitii ‘whitish’ waitiiwaitii ‘whitish all over’
wet ‘wettish’ wetii ‘wettish’ wetiiwetii ‘wettish all over’

(c) gyorl ‘girlish’ gyorlii ‘girlish’ gyorliigyorlii ‘girlish in many
aspects of behaviour
brik ‘stone’ brikii ‘stony’ brikiibrikii ‘stony all over’
The examples in shown in (34) illustrate that distributive reduplication in GC
applies to all word classes.
(34) (a) tSuk ‘to prick’ tSuktSukin ‘prick(ing) from time to time’
tap ‘to tap’ taptapin ‘tap (ing) from time to time’
(b) naasti ‘nasty’ naastinaasti ‘nasty on occasion’
red ‘red’ redred ‘red off and on’
(c) haid ‘hide’ haidhaid ‘hide from time to time’
daans ‘dance’ daansdaans ‘dance from time to time’
(d) wan ‘one’ wanwan ‘in occasional ones’
plentii ‘many’ plentiiplenti ‘in occasional groups of several’
Huttar and Huttar (1997) describe the meaning in Ndjuka, as giving the sense of
people or things being divided or parceled out. They also include the notion of variety,
which conveys the sense of several groups or kinds or actions dividing things into groups
or kinds as illustrated in (35). All of these can be classified as distributive according to
the definition I use here.
- 48 -
(35) waata ‘water’ waatawaata ‘various liquids
soutu ‘kind’ soutusoutu ‘many kinds’
kulu ‘group’ kulukulu ‘different groups’
tutu ‘two’ tuutuu ‘two, two-by-two’
A few examples from Saramaccan are given in (36), (Baker, 1987).
(36) wa(n)wan ‘one by one’

tutu ‘two by two’
2.3.5 Attenuative
Attentuative reduplication is also referred to as approximative reduplication in the
literature. Approximative reduplication is said to give the reduplicated word
‘approximative semantics’ (cf. Huttar and Huttar, 1997; Migge, 2003). Migge notes that
what she terms approximative reduplication is also referred to as diminutive,
imperfective, or pejorative reduplication in the literature (ftn 9). According to Migge, in
approximative reduplication, the reduplicated word refers only to part of the quality
referred to by its base or it may have a distributive interpretation, for example, nyoni
‘small’ > nyoninyoni ‘smallish’. Verbs are reduplicated to form verbs or attributive
adjectives. I refer to these collectively as attenuative. Examples are shown in (37),
(37) nyoni ‘small’ nyoninyoni ‘smallish’

langa ‘long’ langalanga ‘longish’
piiti ‘torn’ piitipiiiti ‘torn in several places’
nyan ‘eat’ nyannyan ‘eat in several places’
- 49 -
The examples in (38) are reported for Ndjuka (Huttar and Huttar, 1997).
(38) nyan ‘to eat’ nyannyan ‘nibbled’

lebi ‘red’ lebilebi ‘reddish
lontu ‘round’ lontulontu ‘roundish’
2.3.6 Stative
The fifth semantic category reported in the literature, is stative reduplication. This is said
to give the reduplicated word stative semantics. The reduplicated word conveys the sense
that the quality referred to by the base has been persisting, for example, fon ‘to beat’ >
fonfon ‘be in a beaten state’ (cf.Alleyne, 1987; Migge, 2003). Verbs are reduplicated to
form adjectives that function as both predicates and modifiers of nouns. The following
examples in (39) are from Ndjuka (Migge, 2003). The reduplications in question include
those in (a) which refer to activities that result in visible or ascertainable states and those
in (b) that refer to concepts of human propensity. Winford (1997) makes similar
distinctions among reduplicated items in Sranan.
(39) a. booko ‘to be broken/to break’ bookobooko ‘(be in a) broken (state)’

fon ‘to beat fonfon ‘(be in a) beaten state’
b. giili ‘be greedy’ giiligiili ‘(be in a) greedy (state)
taaku ‘be evil/be ugly’ taakutaaku ‘(be in a) evil/ugly state’
Stative reduplication is also reported for other Creoles. Kouwenberg and LaCharité
(2001) give the examples in (40) from their survey. Although enough details are not
- 50 -
given about this type of reduplication, they appear to be similar to Migge’s stative
reduplications.
(40) Ndjuka bai ‘buy’ baibai ‘bought’

Saramaccan singi ‘sink’ singisingi ‘sunken’
Sranan not attested
Other examples that resemble the stative reduplications are in (41) from Saramaccan
(Baker, 1987).
(41) lailai ‘loaded’

dEEdEE ‘dried’
hopohopo ‘lifted’
2.3.7 Other Meanings
Reduplication also yields other forms which do not easily fit into any of the semantic
classes established here. For example, Kouwenberg and LaCharité (2001) discuss a type
of so-called derivational reduplication involving the formation of deverbal nouns and
derived adjectives. In both of these cases, they argue that the meanings of the
reduplicated forms are not predictable from the meanings of their bases and as such they
are non-iconic. For example, in deverbal noun formation we can get meanings such as
those expressed by the words in (42).
(42) Saramaccan koti ‘to cut’ kokoti ‘cicatrix, ornamental scar’

tai ‘to tie’ tatai ‘string’ (instrument)
Sranan nyam ‘to eat’ nyamnyam ‘food’
tai ‘to tie’ tatai ‘bundle’ (instrument)
- 51 -
Bakker (1987) reports similar cases in Sranan. For example, nominal reduplication
typically applies to nouns referring to animals, plants and body parts as in (43). All of
these are morphologically frozen forms since they do not have productive bases.
(43) wiwiri ‘weed, plant, leaf, hair’

konkoni ‘rabbit’
gorogoro ‘throat, gullet’
grangran ‘dry bushes’
2.4 Full vs. Partial Reduplication
The following cases of reduplication reported on in this section are particularly
interesting in that the input word is not completely copied. Based on the data seen above,
the majority of processes are processes of full reduplication. The following examples in
(44) are from Sranan. Some researchers such as Migge (2003) do not treat these as cases
of partial reduplication whereas other do, for example, Bakker (1987) and Kouwenberg
and LaCharité (1998).
(44) redi ‘red’ redredi ‘reddish’

fatu ‘big, fat’ fatfatu ‘fattish’
Although it appears that this might be an instance of final vowel deletion (paragoge), a
process which is generally operable in the language (Plag and Uffman, 2000), paragoge
alone does not explain the presence of the vowel in the base. It is not clear why the
vowel should be deleted from one portion of the word and not the other.
- 52 -
Huttar and Huttar (1997) also identify cases of partial reduplication in Ndjuka,
which is said to copy the first C(C)V of a base as seen in (45).
(45) (a) sibi ‘to sweep’ sisibi ‘broom’

wan ‘one’ wawan ‘just, only’
tyai ‘to carry’ tyatyai ‘padded headcloth’
2.5 Prosodic Properties of CEC Reduplicated Words
In this section, I discuss issues related to word-level prosody in CEC reduplication.
Based on the preceding discussion, we saw that reduplication in many Creoles yielded
different semantic types. Where reduplication produces segmentally identical words, it
has been observed that there are prosodic differences between the reduplicated words that
cue contrasts in meaning. However, while there is some agreement that prosodic
differences play a role in reduplication, there is disagreement over whether the relevant
phonological category is stress or tone. As I discuss in Chapter 4, this has to do with the
related issue of the prosodic classification of CECs. For example, for Sranan, Dioncie
(1959 in Sebba, 1981) suggested that tone distinguishes between segmentally identical
reduplicated words whereas Adamson and Smith (1999) identify stress as the
distinguishing feature. Devonish (2003) describes verbal reduplication in Guyanese
Creole (GC), in which differences in the pitch pattern differentiate between an iterative
meaning, rónrón ‘to run continuously’ and a distributive meaning, rônrón ‘to run in fits
and starts’.
- 53 -
The fact that these Creole languages emerged out of contact between tone
languages and stress or pitch-accent languages is relevant to the discussion of the
characteristics of their prosodic systems and by implication to the discussion of word-
level prosodic differences in reduplication. We do not yet know how much, if any, of the
prosodic system of the input languages is maintained in these Creole languages. As
discussed in Chapter 3, the precise impact of language contact on the prosodic system of
CECs is still largely unclear (see Devonish, (2002) for some discussion). Perhaps this
would account for the differences in the characterization of the same language’s prosodic
system by different researchers. With regard to JC, Lawton (1963) argues that the
language has a tonal system, whereas Cassidy & LePage (1967) place JC in the same
category as British RP, having a stress system. In undertaking research on the prosodic
properties of reduplicated words, it is imperative to clarify which system is being dealt
with since reduplication word prosody can only be properly understood in the context of
the wider prosodic system. Chapter 4 gives a more detailed discussion of analyses of the
JC prosodic system where I show that JC has a stress-based prosodic system.
Interestingly, Sebba (1981) notes that homophony restricts reduplication so that
where there are homophonous words in different lexical categories, only one will
reduplicate. Elsewhere, he argues that some forms are reduplicated in order to avoid
homophony. Recall that it has been reported that in other creoles, reduplication often
creates segmentally identical words. It is not clear whether by homophony Sebba means
similarity at both the segmental and the prosodic levels. His claim potentially conflicts
with data he cites from Dioncie (1959) in a footnote (ftn.2). Dioncie showed that a
- 54 -
reduplicated adjective may give a ‘defirmative’ (attenuative) or ‘intensive’ meaning as
illustrated in (46).
(46) rédiredi ‘fiery red’ rediredi ‘fairly red’

(very red – my gloss) (reddish – my gloss)
According to Dioncie, this difference in meaning (which I interpret as intensive versus
attenuative) is given by a higher tone on the first part of the word for intensive
reduplication. In addition, Adamson and Smith (1999; 2003) report the use of stress to
differentiate the meaning of segmentally identical reduplicated forms in Sranan, giving a
diminutive meaning (attenuative) as in (47a), an iterative meaning as in (47b) or an
augmentative meaning (intensive) as in (47c).
(47) a. férfiférfi ‘to paint a bit’

b. férfiferfi ‘to paint several times’
c. ferfiférfi ‘to paint too much’
These reduplicated forms can also function as adjectives with similar meanings. For
example, an under painted house, (47a); an over painted house, (47c) and an oft painted
house, (47c). We therefore have two reports of the interaction of word-level prosody
with Sranan reduplication, one in terms of stress and one in terms of lexical tone.
Migge’s (2003) overview of the reduplication processes in Eastern Maroon
Creoles raises the issue of whether segmentally identical reduplications are differentiated
only by syntactic function or their position in the sentence rather than by prosodic
differences. For example, in Migge’s data, both stative and approximative reduplications
can function as adjectives in prenominal position though they are differentiated in other
- 55 -
syntactic environments (Migge, 2003:68). The question is whether the words are
differentiated by prosody when they occur in identical contexts since the listener cannot
rely on syntactic information. As I discuss in Chapters 4 and 6, commuting contrasting
pairs of segmentally identical words in the same syntactic frame is a useful strategy for
discovering what the acoustic cues to the contrasts are as well as for resolving the issue of
the contribution on context to the interpretation of the reduplicated words.
Migge does report however that in SEM (and Sranan Tongo) reduplicated forms
cannot undergo a second process of reduplication. Instead their form stays constant and
their meaning changes. In these cases there is a difference in the tone pattern of the
reduplicated item. She notes however that this was not observed in the data reported for
EMC or Sranan.
A number of researchers make a distinction between real and apparent
reduplicated words based on their phonological or phonetic properties (Carter, 1987,
Huttar and Huttar, 1997; Migge, 2003; Devonish, 2003). Huttar and Huttar (1997) and
Migge (2003), for example, analyse reduplication as “the repeating of all or part of a
word (more than a single segment), the result still being a phonological word, with its
pitch and stress pattern” (Migge, 2003:2). Migge argues that intensive reduplication does
not constitute (true) reduplication but rather repetition, since each part of the word has
independent phonological and semantic qualities. Kouwenberg and LaCharité (2001)
require that a base has a ‘non-emphatic prosody’ and that the corresponding reduplicated
item be ‘produced under a single intonation contour’ (pg.1). Their requirements
effectively rule out cases of whole word reduplication in which prosodic differences
- 56 -
signal differences in meaning. As noted earlier, cases like this are reported for several
Creoles, for example, Guyanese Creole (Devonish, 2003), Jamaican Creole (Gooden,
1999 et seq.) and Sranan (Dioncie, 1959; Adamson and Smith, 1999; 2003). Another
problem is the term ‘non-emphatic prosody’ itself. It is not clear what this means or how
this is realized in the different Creole languages being described. A similar problem
arises with their notion of ‘a single intonation contour’. Since their database consists of
several Creoles some of which are unrelated and have different source languages, it is
reasonable to assume that they do not necessarily have identical intonation patterns or
prosodic systems. As noted above, this is especially important since prosodically
contrastive reduplications of verbs, adjectival and adverbial modifiers have been reported
for other English-lexicon Creoles. As I show in Chapters 6 and 7, an understanding of
both the phonological and phonetic properties of segmentally identical reduplications is
an important aspect of understanding the nature of the contrast between them.
According to Bakker (1987), Saramaccan has lexical tones and in reduplication
the tonal pattern of the reduplicated word is identical to the tonal pattern of the input
words. It is not clear though what the tonal melody of either part of the word is before
reduplication because nothing is said about it. This is especially important since it has
been shown that even in cases of full copy, different restrictions apply to the base or to
the copied material (McCarthy and Prince, 1995a; McCarthy and Prince, 1995b;
Urbanczyk, 2001). If Bakker’s observations are correct however, then this would be a
case of total reduplication in which everything including prosodic information is
transferred in reduplication. However, Bakker notes that there are exceptions to this rule.
- 57 -
In isolation, the last tone of the reduplicated word is always low and all low tones
between high vowels become high tones. Further exceptions are words that were
borrowed from Sranan. In this case, they retain the accent patterns of Sranan.
Carter (1987) made two distinctions between phonologically identical forms in
Guyanese Creole (GC), i.e. iteration versus reduplication. Carter views iteration as a case
in which each part of the repeated form is a separate ‘tone group’; these give an intensive
interpretation (48). The words in (48b) are classified as reduplication, which for Carter is
a subtype of compounding. Reduplication in GC under her analysis has H-tone deletion
for the first component and a nuclear H-tone on the first syllable of the second
component of the word, irrespective of the pattern in isolation. Carter claims that this
gives the word a distributive meaning though this it not reflected in the glosses given (the
! marks downstep).
(48) (a) ta@ll ta@ll!ta@ll ‘very tall’ (b) tallta@ll ‘rather tall’
hole@y hole@y!hole@y ‘very full of holes’ holeyho@ley ‘rather holed’
Devonish (2003) argues that Guyanese Creole has both morphological
reduplication that copies words and syntactic reduplication that copies phrases. As in
Carter’s analysis, Devonish makes a distinction between compounding and non-
compounding reduplication. As was noted earlier, this distinction is partly semantic and
partly phonological. With regard to the phonological distinction, compounding
reduplication gives a distributive meaning and forms a single tone phrase while non-
compounding reduplication gives a repetitive/continuous meaning (iterative) and does not
- 58 -
form a single tone phrase. Notice that this distinction is similar to Carter’s distinction
between iteration and reduplication. Compounding reduplication is morphological
reduplication in Devonish’s treatment and non-compounding reduplication is syntactic
reduplication.
Devonish also discusses the tonal description of reduplicated forms. In his
analysis, all lexical items in GC have an obligatory HL melody and its location in the
word is lexically specified. He assumes also that GC forms tone phrases (roughly
analogous to Carter’s tone group), which are constructed from right to left. Words with
more than one HL melody are permissible only as the final item in a tone phrase. Finally,
the formation of a tone phrase is said to be dependent on compounding.
Unless indicated otherwise, all the reduplicated words in the GC examples in (49)
have a HL tone on the first portion of the word and an H-tone on the second portion. In
bisyllabic words, the H-tone is on both syllables.
(49) tSuk ‘to prick’ tSuktSukin ‘prick(ing) from time to time’

naasti ‘nasty’ naastinaasti ‘nasty on occasion’
haid ‘hide’ haidhaid ‘hide from time to time’
wan ‘one’ wanwan ‘in occasional ones’
2.6 Morphological Function of Reduplicated Words
The output of several reduplication processes in CECs produces words that have been
classified as property items in the literature. Winford (1997) and Migge (1998), for
example, refer to predicative adjectives such as siki ‘sick’, fatu ‘fat’ as property items
- 59 -
reflecting the observation that they refer to the ‘properties, qualities or characteristics of
referents’(Thompson, 1988). Migge states a preference for the latter morphosyntactic
label since it is neutral with respect to the syntactic category of these words. Migge also
notes that crosslinguistically, property items fall into several syntactic categories such as
adjective, verb, or noun depending on their discourse function, (Migge, 1998:116).
Based on the grammatical labels for the word class of the reduplicated words, I assume
that reduplicated property items serve several of these functions. In addition, Migge’s
(2003) study of the syntactic properties of reduplicated words supports this idea.
In Guyanese Creole, compounding reduplication may change the lexical category
of the input form, i.e. nouns to adjective or adverbs as in (50a) and verbs to nouns as in
(50b) (Devonish, 2003).
(50) (a) piis ‘piece’ piispiis ‘in random pieces, piece by piece’
tap ‘top’ taptap ‘superficial in parts’
(b) bloo ‘breathe, bloobloo ‘whistle, flute’
lef ‘leave’ leflef ‘left-over food’
2.7 Chapter Summary

In this chapter I described four different processes of reduplication attested in Jamaican
Creole (JC). As discussed, the majority of processes in JC as in other CECs are primarily
full reduplication but there are also some cases of putative partial reduplication. The
reduplicated words serve a variety of functions in the different languages, which can be
grouped together as the functions of ‘property items’. In addition, CECs express common
meanings in reduplication, though not all the processes are reported for all the languages.
- 60 -
For example, JC does not appear to have a distinct class of stative reduplication. In fact the
semantic types found in JC reduplication are a subset of those found across the CECs in
general. I also noted that in some cases, JC reduplication created forms which are
impossible to distinguish on the basis of their segmental content alone. Similar cases were
reported for other creole languages as well, including Sranan and Guyanese Creole. Two
of these processes in JC, on which I focus the remainder of this dissertation, are distributive
reduplication and intensive reduplication. As I have suggested in this chapter, I propose
that these reduplication processes are only superficially similar. Specifically, based on
stress facts about the language discussed in Chapters 4 and 6, I show that distributive and
intensive reduplication are in fact different at the prosodic level. Further, I present phonetic
evidence of this difference in Chapter 6 and show how these phonetic differences are
implemented into the phonological grammar of the language in Chapter 7.
- 61 -
CHAPTER 3
JAMAICAN CREOLE IN ITS SOCIOCULTURAL CONTEXT
3.1 Introduction
In this chapter, I discuss aspects of the Jamaican language situation which are relevant to
the present study. The most important issue, for our purposes, is that there are three
distinguishable language varieties spoken in Jamaica. These include standard Jamaican
English, referred to as the acrolect, a conservative (rural) creole, referred to as the
basilect, which is the focus of this dissertation, and an intermediate variety, referred to as
the mesolect. Together these comprise the Jamaican Creole (JC) continuum. The
interaction among these varieties leads to significant variation in the phonological system
and in the grammar as a whole.
I provide a general overview of two aspects of the linguistic situation: the
contemporary sociolinguistic situation and a sociolinguistic profile of the speech
community from which I collected production data. I also discuss issues related to
degrees of regional and social variation which have been argued to be effects of the
-62-
putative creole continuum. This discussion will serve to locate the Top Alston
community chosen for this study, as well as the basilectal vernacular spoken by its
citizens, within the context of the general sociolinguistic situation.
In section 3.2 I give a broad overview of the sociohistorical background of the
language and discuss issues related to the impact of language contact on the phonological
properties of the language. Section 3.3 gives an overview of the contemporary linguistic
situation. In section 3.4 I present a sociolinguistic profile of the Top Alston community.
In 3.5 I present sociodemographic information on the informants interviewed for this
research and discuss some issues related to data collection. Section 3.6 summarizes the
main points of the chapter.
3.2 Sociohistorical Background
What is known as Jamaican Creole today is a variety forged out of contact between West
African languages on the one hand and varieties of the British Isles on the other (Lalla
and D’Costa, 1990). This is illustrated in Map 3.1 which shows the sources of the
Jamaican population between 1500 and 1700. As is confirmed by the linguistic
topography and history of Jamaica, the map also indicates that there are some minor
influences from Spanish, which resulted from earlier Spanish colonization (1505 – 1655),
Portuguese and Taino (Lalla and D’Costa, ibid). The main influence from these sources
is seen in a few lexical items (Cassidy and LePage 1967/1980).
-63-
-64-
Map 3.1. Sources of Jamaican population (Lalla and D’Costa 1990:4)
Demographic data on the origins and ethnolinguistic composition of slaves
brought to Jamaica during the British plantation period (1655 -1808) shows that there
were a variety of West African languages involved in contact with English, most of
which were typologically similar (LePage, 1960; Arends, 1995). Among these were
languages of the New Kwa family, including the Gbe and Akan clusters of languages;
languages of the Benue-Congo family, and the Bantu family, particularly the Kikongo
group of languages (Bendor-Samuels, 1988 in Migge, 1998; Arends, 1995). The
speakers of the Gbe varieties are distributed over four modern West African states:
Ghana, Togo, Benin and Nigeria (Migge, 1998: 129). The influence of West African
languages on the vocabulary, syntax and segmental phonology of CECs has been
convincingly demonstrated in various studies (cf. Smith, 1987; Migge, 1998; Lefebvre
1998 among others). However, little attention has been paid to other possible influences
on the prosody of CECs. Exploration of these influences on the prosody of JC is beyond
the scope of this present study. However, for purposes of this dissertation, the most
important aspect of this contact is that the prosodic system of these languages and that of
the English varieties are typologically dissimilar. On the one hand, the West African
languages had lexical tone systems and on the other hand the English varieties had stress-
accent systems. The precise implications of this for the emergence of Creole prosody are
still not well understood. As I discuss below, this has led to conflicting accounts of the
prosodic system of Jamaican Creole as well as other CEC varieties.
-65-
3.2.1 Impact of Language Contact on Word-Level Prosody
In Creole language contact situations, as in other cases of contact-induced change, a
recurrent question is whether the grammar of the emergent contact language consists of
separate/coexistent linguistic systems, a single merged system or a system in which there
is a superordinate/dominant linguistic system as well as a subordinate system (Weinreich,
1953). With regard to phonology, it is well known that intense contact can lead to
varying degrees of transfer from a source language to a recipient language, particularly in
cases of second language acquisition, of which creole formation may be viewed as a type.
This may result in a new system that is a compromise between the systems in contact
(Van Coetsem, 1996). According to Van Coetsem, the accent types in particular are
more readily transferred from a source language to a recipient language. In the case of
CECs, we do not know how much, if any, of the prosodic system of the input languages
was transferred to the languages and most importantly what aspects of the prosodic
systems have been retained in the synchronic phonologies of CECs. Harry and Devonish
(1998) argue that Caribbean Creoles may have developed prosodic systems that are: (a)
similar to those of the European languages in the contact situation; (b) similar to the West
African languages in the contact; (c) hybrid systems consisting of influences from more
than one West African language; (d) hybrid systems consisting of influences from both
European and West African languages; or (e) entirely original having no input from
either of the source languages. LePage (1960) mentions several areas of JC which show
some influence from West African languages, for example, in some aspects of the
-66-
phonemic pattern, the stress system and in some lexical items. Most interestingly for this
research, he notes that ‘tones’6 are often used to distinguish homophones.
3.2.2 Impact of Language Contact on Reduplication
West African influence on the Jamaican language and culture has long been recognized
in the literature. Some linguistic influences noted are: the predominance of CVCV
syllable structure, serial verb constructions, reduplication and certain aspects of the tense
aspect system (cf. Alleyne, 1980; Holm, 1988, among others). With regard to influence
on reduplication, the question is whether the patterns observed in JC can be traced to the
input languages, or whether the patterns are themselves innovations and therefore not
attributable to any source language. Since reduplication as a productive process is not
found in English, it would be necessary to look at reduplication processes in the
contributing West African languages to address the first part of this question. In this
connection, DeCamp (1974) notes that while reduplication is characteristic of all
European-based pidgins and creoles, the process is only marginal in the European source
languages and are of a different sort than is found in these contact varieties. Of JC,
DeCamp claims that many of the reduplications are clearly of African origin while others
are later internal developments (ibid.).
This dissertation does not concern itself with the issue of the origin of
reduplication processes in JC7 though it is certainly an important one. As noted above,

6
Based on data discussed in Chapters 4 and 6 I do not interpret this as meaning lexical tone but rather F0 differences
related to stress.
-67-
the related issue of the prosodic classification of the languages in contact is important as
well. However, the focus in this study is instead on the synchronic description of JC
reduplication and its interaction with JC word-level prosody.
3.3 Contemporary Language Situation
The language situation in Jamaica has been described as a language continuum (Reinecke
and Tokimasa, 1934), or in DeCamp’s (1971) terms, a post-creole continuum. This
describes a situation in which two distinct but lexically related polar language varieties
co-exist in the same speech community, with intermediate varieties in the ‘middle’ of the
continuum. DeCamp (ibid) for example, recognizes an acrolect, which is essentially the
local standard English variety; a mesolect, which includes a mixture of features from the
acrolect and the basilect. The basilect is the most conservative variety and is most
distinct from the acrolect, having most ‘creole features’.
The example in (51) below illustrates some possible variations for the statement
‘she was going to school’. As seen, the single statement shows a variety of
morphosyntactic forms. The examples in (52) show phonological variation in s-stop
clusters (Akers, 1981; Meade, 1995).
(51) im (b) en a go a skuul basilect

im did a go a skuul mesolect
im was going to skuul
shi was going to skuul acrolect
7
See Migge (2003) for a discussion of the origin of reduplication processes in Surinamese Creoles.
-68-
(52) // acrolect // basilect ‘stick’
// acrolect // basilect ‘spit’
I should note here that the term mesolect has been used both collectively and individually
to refer to the intermediate varieties in the same speech community. Some researchers
(cf. Patrick, 1999) see the mesolect as a variety distinct from both the acrolect and the
basilect.
The Jamaican speech community and other creole continua have been
characterized as being diglossic (McCain, 1996). Diglossia describes a language
situation in which there are two language varieties in the same speech community each
fulfilling distinct communicative/social roles (Ferguson, 1959). The situation may also
involve intermediate varieties; however, unlike creole continua situations, in diglossic
situations all speakers are competent in both the high and the low variety. Most
Jamaicans acquire a creole variety as their first language and learn Standard (Jamaican)
English at school as a second variety.
Jamaican speakers are often competent in one or more varieties with varying
degrees of proficiency in SJE. While it is generally possible to demarcate particular
linguistic features as being Creole forms, it is more difficult to identify discrete divisions
between basilect, mesolect and acrolect. Consequently, researchers like Wassink-
Beckford (1999) refer to speakers as being dominant in one or another. The different
degrees of competence in the different varieties, leads to social and stylistic variation that
constitutes the creole continuum.
-69-
3.3.1 Variation in Creole Continua
Variation is an intrinsic property of all languages. Even among so-called ‘standard’
varieties, one will find differences in the linguistic behavior of speakers. The recognition
that variation exists is even more striking for researchers of creole varieties. This is
intimately bound up with the very characterization of the creole language situation as a
speech continuum. Variation in creole continua, as in other languages, is nonetheless
systematic and the patterns can be shown to be operative at the level of the speech
community. In the first place, there is often a correlation between the use of a particular
language variety and the socio-economic status of the speaker. Two recent studies on
sociolinguistic variation in Jamaican speech communities (Patrick, 1999; Wassink-
Beckford, 1999) showed that linguistic variation is observable between the acrolectal end
of the continuum and the basilectal end and further that the variation can be attributed to
speakers’ social class. In the Jamaican setting, socio-economic status is also bound up
with the speaker’s level of education and occupation (cf. Meade, 2001). Other factors
such as age and sex have been shown to correlate with linguistic variation as well (see for
example, Irvine, 1994; Wassink-Beckford, 2001).
Geographical location also plays a role in linguistic variation. For example,
DeCamp (1961) maintains that the most conservative forms of Jamaican Creole are to be
found in historically isolated communities in the mountainous regions of the island, i.e.
the historical Maroon communities, for example, Accompong. In addition, there are
differences to be noted between rural and urban speakers and even among speakers
within the same locale (cf. Patrick, 1999). The Top Alston community, which I describe
-70-
below, is a rural community located in the hills of the northwestern part of the parish of
Clarendon. I selected this community since it was likely to have speakers who primarily
use a basilectal variety of JC. This was important since this study is primarily concerned
with the most conservative variety of Jamaican Creole. This is on the assumption that it
best preserves conservative grammatical features of the language such as the patterns of
reduplication under investigation in this dissertation. The choice was thus based largely
on the relative isolation of the community, sociodemographic profile of the residents,
respondents’ opinions from a pilot study conducted in August 2000 and my own
familiarity with the community.
3.4 Community Profile
The parish of Clarendon has an area of approximately 1,167 sq km (467 sq miles). The
2001 census of the Statistical Institute of Jamaica (Statin) reports a population of
236,150, making it one of the most populous parishes in the island. The total population
reported for Alston was 1,662 persons with an average of 3.9 persons per household
(Statin, 2002). Map 3.2 is a map of Jamaica showing the parish of Clarendon, located in
the central part of the island. The approximate location of the community is indicated by
the star on the map.
-71-
Map 3.2. Partial Map of Jamaica Showing Clarendon and Nearby Parishes
Top Alston is a farming community located approximately 50 miles north of the
capital, Clarendon, May Pen, and about 25 miles from Mandeville, the other major town
(See Map 3.2). There are three smaller towns nearby; Spauldings, Frankfield and,
Christiana. For people in the community, these towns represent the major business
centers, providing services such as banking, police, hospitals, clinics, libraries, shopping
etc. Neighbouring communities are (Bottom) Alston and Bailleston, which are separated
from Top Alston by a river. Other nearby communities include Tweedside, Wild Cane,
-72-
Moravia and Silent Hill. All of these communities are within ‘walking distance’8 of each
other. Although there have been roads in this community for some time, until
approximately five years ago the majority of traffic was pedestrian traffic and people had
to travel to the ‘main road’ to gain access to public transportation.
The majority of residents of the Top Alston community are farmers (crops and
animal husbandry). Others are employed as seamstresses, electricians, tradesmen or
tradesmen apprentices (masonry, carpentry etc.), informal commercial traders (hagglers),
shopkeepers for small community shops and part-time taxi operators. With regard to
education, the majority of people aged 40 and above, have at least primary level
education, while some have no formal education outside of a church setting. For
example, one participant reported learning to read after he started attending church at age
56. Those between ages 25 and 40 have secondary level education and some of the
younger persons 16 - 25 are pursuing post-secondary level studies.
Many activities such as grave digging and land clearing are still a community
effort, though they may now involve some amount of remuneration. Many community-
based activities are centered around local churches, for example, rallies, fund raising
concerts, baptisms, funerals etc. The majority of participants selected for this study were
reasonably well involved in these and other community-based activities. This was
important since their level of affiliation with the community might be taken as a measure
of their participation in community norms for language use (cf. Milroy, 1980). We might
8
The average walking time is about 30 minutes for the closest point to about 60 minutes or more to the furthest point.
-73-
therefore expect that other members of the community share patterns of use for
reduplicated words similar to those used by the speakers in this sample.
3.5 Data Collection
3.5.1 Subjects9
The original subject pool for this study consisted of 40 native speakers of basilectal
Jamaican Creole from Top Alston. The acoustic data I report on in Chapters 4 and 6 is
taken from 7 of these speakers. These 7 speakers were chosen because they provided
productions of both unreduplicated words and reduplicated words in the desired contexts.
Further, these speakers participated in a small perception test that verified their
perception of the contrast between intensive and distributive reduplications. The subject
pool included both male and female subjects in two age groups to control for possible age
or sex differences. I assume that rural subjects in general use a more conservative form
of the Creole and are able to produce the contrasts clearly. As noted above, speakers in a
small, relatively isolated rural community are likely to be competent in a Creole variety
closer to what is characterizable as the basilect, and would therefore have more Creole
features in their speech (cf. Beckford-Wassink, 1999; Meade, 2001). Table 3.1 shows
demographic data for the participants whose production data are reported on in Chapters
4 and 6. The majority of these informants reported use of JC in most social settings such
as, church, home, among close friends, and English in a few settings, for example,
9
The author (SG) lived in this rural area for 20 years prior to moving temporarily to the capital for 3 years.
-74-
business. A sample of the informant data sheet used in the interviews to collect this
information is given in Appendix A.
Participant Age Sex Highest Level of Language Variety Used Most

Education Often
EW 56 F Primary Creole
SG 29 F Post-Secondary Creole (English sometimes)
HH 67 F Primary Creole
RP 32 M Secondary Creole (English sometimes)
WW 64 M Primary Creole
SP(m) 42 M Secondary Creole (English rarely)
HF 48 M Secondary Creole (English sometimes)
Table 3.1. Informant Demographic Data
3.5.2 Production Tasks
As a result of the fact that JC is not written, speakers tend to associate reading and formal
situations with English. Given that an interview introduces some level of formality to the
interaction between the researcher and the participant, I used several strategies to
minimize the level of formality, thereby minimizing the likelihood of influence from
English. First, for as much as possible, all interactions with the participants were done in
Jamaican Creole; a detailed illustration of a typical elicitation session is given in section
5.1.2.3. Second, I used a picture task to elicit the target words. The advantage of this is
three-fold. Alongside the contextual information provided, the pictures helped to cue the
desired meaning for each target word and facilitated comparison across productions and
across subjects. In addition, the picture task helped to minimize influence from English
-75-
which could be potentially introduced through a reading task. From another perspective,
since the subject pool included participants with different degrees of literacy, the picture
task avoided any potential difficulties with reading as well as removed the undesirable
effect of separating participants based on their ability to read10. Wassink-Beckford (2001)
and Gooden (2002) also successfully used picture tasks to elicit data for basilect-
dominant speakers in Jamaica and Belize respectively.
3.6. Chapter Summary
The discussion presented here raised several issues which are important for the discussion
and analyses in ensuing chapters. First, the linguistic history of Jamaican Creole suggests
that there is influence from West African languages and English on the contemporary
linguistic system. Strong influence from the former might have resulted in JC prosody
manifesting properties of a tone-based prosodic system, as reported for Creoles such as
Saramaccan. However, as I show in Chapters 4 and 6 where JC prosody is concerned,
the evidence favors a treatment in terms of stress rather than tone. Second, research on
any subsystem of creole language varieties has to be mindful of the variability
characteristic of creole continua as well as acknowledge that this variation is conditioned
by social, regional and situational factors. We saw, for example, that considerable
differences in the phonological properties of words were associated with different levels
of the continuum. Given all of this, the selection of appropriate speaker samples is vital
10
One participant at first refused to participate in the session because he thought the tasks involved reading.
-76-
to delimiting the variety being targeted for study. In this regard, the community chosen
for this research is reasonably well-defined as a community where basilectal speech is
used in everyday life by members of the community. This is based on the relative
isolation of the community, the demographic data collected for the speakers, as well as
the testimonials of the speakers themselves regarding their use of JC.
-77-
CHAPTER 4
WORD PROSODY: EVIDENCE FOR JC AS A STRESS-BASED SYSTEM
4.1. Introduction
The aim of this chapter is to provide an overview of the phonological properties of the
basilectal variety of Jamaican Creole (JC) and provide background information for the
phonetic and phonological analyses presented in following chapters. The chapter focuses in
particular on a characterization of JC word prosody. The discussion of JC word level
prosody presented here is by no means exhaustive but it is nonetheless an important
contribution to discussions on the prosodic identity of Caribbean English Creoles (CEC) as
well as other creole languages. The prosodic identity of CECs like JC has been subject to
much debate. The fact that these languages emerged out of contact between tone languages
and stress-accent languages is an important aspect of the characterization of their prosodic
systems. As discussed in Chapter 3, there are several possible scenarios for the effect of
language contact on the prosodic system of creoles.
- 78 -
While traditional typologies classify languages as tonal, stress-accent or pitch-
accent, several CECs have been classified as having both stress and tone operative in their
prosodic system, i.e. a merged system. For example, Devonish (1989; 2002; 2003) argues
that CECs like Guyanese, Ndjuka and Jamaican among others are restricted tone languages
having only an underlying H tone and a surface HL tone melody. In addition, these
languages have stress, the assignment of which is dependent on the location of the H tone
in words. Brousseau (2003) claims that the phonological systems of Haitian appear as a
compromise between those of its contributing languages and Good (2003) argues that some
words in Saramaccan have prominences related to stress and other words have prominences
related to tone. Similar arguments have been presented for other Caribbean creoles as well;
for example, Papiamentu (Rivera-Castillo, 1998) and Ndjuka (Huttar, 1988).
Despite this broad consensus, there is disagreement on exactly how the roles of
stress and tone are distributed. Remijsen (2001) points out that in order for more than one
system to be functional in any language, the phonetic encoding must be distinct. Devonish
(1989) points out that the problem is how to distinguish between pitches produced as a
result of the presence of stress and those produced as a result of the presence of lexical
tone. Carter (1980) argues that while all languages use pitch in a systematic way, they vary
according to the role and function of pitch in the linguistic system. As such, the presence
of pitch contrast in minimal pairs is not sufficient to classify a system as tonal. What is
needed is additional distributional evidence to form a characterisation of the prosodic
system as a whole. The idea of a ‘mixed’ prosodic system is not unique to Caribbean
Creoles. Ramijsen (2001) and deLacy (2002), among others, discuss a wide range of
- 79 -
languages which have both lexical tone and stress. In these languages, tone and stress
interact to optimize stress assignment or the location of tones in words.
In the discussion of the JC word level prosodic system presented in this chapter I
examine phonological data in support of the proposal that Jamaican Creole can be
classified as a stress-accent language. I show that JC has stress system in which syllable
prominence is signaled by an F0 fall within the word. This classification forms the basis of
the analysis of the prosodic differences in the segmentally identical reduplicated words in
following chapters.
The chapter is organized as follows. Sections 4.2 and 4.3 give background
information on JC concerning the phonemic inventory and presents assumptions regarding
syllable structure, foot structure and moraicity that are important for subsequent analyses.
Section 4.4 reviews different perspectives on the classification of JC word level prosody.
Here I lay out both phonological and phonetic criteria for the classification of the prosodic
system. Section 4.5 provides empirical evidence for the classification of JC prosody in
terms of stress. In section 4.6 I give an overview of the strategies used for discovering the
acoustic cues to stress and accent contrasts in several languages and describe how these
strategies are applied to the JC data. Section 4.7 provides phonetic data on the F0 pattern in
a variety of monomorphemic words and compounds as further illustration of the phonetic
realization of stress in JC. In the final section, I summarize the main points discussed in the
chapter and highlight the implications of stress for the analysis of the two processes of
reduplication that yield segmentally identical words.
- 80 -
4.2. The Phoneme Inventory
I adopt the basic phonemic inventories of Meade (2001) and Beckford-Wassink (2001)
with some modifications where necessary. The phonemic inventory of consonants is
shown in Table 4.1 and the inventory for vowels is shown in Table 4.2. I assume, based
on the basilectal data presented in Meade and Beckford-Wassink, that these inventories
are representative of the basilectal variety of JC spoken by the informants in this
research. This is based on the fact that the sociodemographic data provided for the
basilectal speakers in both studies closely matched that of the speakers in this study. The
speakers in Beckford-Wassink’s study were all lower working class persons who were
born and raised in a rural community in St. Thomas; those in Meade’s study were from a
rural community, Guys Hill, in St. Catherine and had no more than primary level
education and were typically unemployed or had manual-skilled jobs. The speakers
interviewed for this dissertation, are also from a relatively isolated rural community. The
majority of these speakers have primary level education and are farmers or hold other
manual-skilled jobs (refer to Chapter 3 for additional details). We can further assume
that the similarities among the speech of informants from these different rural
communities stem from the observation that speakers in smaller relatively isolated rural
communities are likely to speak a creole variety that is closer to a basilectal variety.
Table 4.1 shows the phonemic consonant inventory. In addition to those listed here,
Meade (1996; 2001) also lists gj and kj as palatalized allophones of g and k which occur
before the low back vowel /a/; Patrick (1999) discusses the palatalization of these sounds
in detail. It should also be noted that h occurs sporadically being typically absent in
- 81 -
words like ama ‘hammer’ and ed ‘head’, occurs in hypercorrected forms like hegg ‘egg’
and appears consistently in words like hafu ‘type of yam’ and haaspital ‘hospital’.
Bilabial Labiodental Alveolar Postalveolar Palatal Velar Glottal

Stop   
Fricative      
Affricate      
Nasal      
Approximant      
Table 4.1: JC Consonant Inventory
Table 4.2 shows the phonemic vowel inventory. As shown, basilectal JC speakers
distinguish between seven short vowels, three long vowels and three diphthongs.
Examples of the different vowel length and diphthong contrasts are shown in (53).
Front Central Back

High Tense   
Lax   
Mid Tense   
Lax   
Low   
Diphthongs   
Long vowels :  ::
Table 4.2: JC Vowel Inventory
(53)  ‘pierce ‘  ‘joke’

 ‘hit/lick’  ‘lake’
 ‘metal container’  ‘pineapple’
 ‘mucus from eyes’ : ‘mortar’
 ‘bitter’ : ‘beater’
 ‘full’ : ‘fool, simpleton’
- 82 -
Some amount of controversy has surrounded the JC phoneme inventory mainly with
regards to the long vowels, diphthongs and the palatal sounds /kj/, /gj/ and /nj/. Cassidy
and LePage (1967) and Akers (1981), for example, include the palatal phones creating a
total consonant inventory of 23 phones11. On the other hand, Whittle (1989), Devonish and
Seiler (1991) and Meade (1996; 2001) treat the palatal sounds as allophones of the
phonemes k, g and n so that the tally of consonants amounts to 20. With regard to vowels,
some authors list as few as 9 vowels in their inventory (Lawton, 1963; Lalla and D’Costa,
1990) while others list up to 16 vowels (Wells, 1973; Veatch, 1991 in Wassink-Beckford,
2001). There is some agreement that the basilectal variety has a basic 5-vowel system in
which length is contrastive. However, authors disagree on just how many long vowels
there are. For example, LePage (1960) lists only two long vowels, Lawton (1963) has
none, Wells (1973) and Meade (1996, 2001) list three and Akers (1981) lists five.
Wassink-Beckford (2001) argues that these disparate accounts of the segmental inventory
may stem partly from sampling errors. In particular, she argues that in some cases speaker
samples were not controlled for sociolinguistic factors such as socioeconomic class, speech
style and gender which are known to have effects on speech. In other cases, no distinction
is made among speakers from different levels of the speech continuum. That is, basilectal,
mesolectal and acrolectal varieties are pooled together so that a distorted picture of the
linguistic system is created. Irvine (1994), points out that in linguistic research on creole
continua, it is essential for the researcher to control for variation. The key point is that in
creole continua, social dialect differences as observed between the basilectal, mesolectal
11
Akers (1981) does not include /v/, so that the total number of consonants he reports is 22.
- 83 -
and acrolectal varieties in JC could translate into considerable phonological variation. As
discussed in Chapter 3, the selection criteria for the informants whose speech is described
in this dissertation ensured that the phonological descriptions given are representative of a
basilectal variety.
4.3. Prosodic Structure
4.3.1 Syllable Structure
The basic syllable structure which is assumed for the JC data described in this
dissertation is as shown in (54) below.
(54) (C(C)) V(V) (C (C) (C))
This structure reflects the fact that the syllable in JC obligatorily has a nucleus which
may be complex, being comprised of a long vowel or diphthong. In addition, syllables
may have a complex onset and up to three segments in the rhyme. Representative
examples of possible syllable structures are shown in (55);
(55)  ‘knock’ CVC

 ‘skate’ CCVVC
 ‘dog’ CVVC
  ‘clash’ CCVC
 ‘thanks’ CVCCC
: ‘type of fruit’ CVVCV
 ‘prickle, thorn’ CVCV
:. ‘trellis’ VVCV
- 84 -
4.3.2. Foot Structure and Moraicity
Based on the discussion of JC stress facts in section 4.4, we will see that JC forms
trochaic feet. We will also see that moraic structure is important in stress assignment as
the language is quantity sensitive. That is, stress is typically assigned to bimoraic, i.e.
heavy, syllables and it is only in the absence of a heavy syllable that a light syllable is
stressed. Following Hayes’ (1989) treatment of syllable weight and moraicity, I assume
that coda consonants in JC contribute to syllable weight. Heavy syllables contain
maximally two moras and therefore include syllables with VC and VV sequences which
may be comprised of either long vowels or diphthongs. In addition, the data shows that
consonants in final consonant clusters or coda consonants in syllables with long vowels
or diphthongs do not contribute an additional mora to syllable weight, effectively ruling
out super heavy syllables in JC. That is, CVCC and CVVC syllables are just as heavy as
a CVC syllable.
4.4. JC Word-Level Prosody
The primary goal of this section is to lay out criteria by which to arrive at an adequate
characterization of the synchronic word-level prosodic system of Jamaican Creole. There
have been several arguments for a characterization of the JC prosodic system as a stress
system, a tonal system and as a system that incorporates both stress and tone. Since terms
like stress, tone and accent have been used with a variety of meanings in the literature, I
will first clarify how they will be used in this and following chapters. I use the term stress
to refer to the abstract phonological category by which we can distinguish degrees of

- 85 -
emphasis or contrast in sentences or words (Crystal, 1997). I follow Hyman’s (1978) basic
treatment of accent as the marking of a syllable with greater salience than surrounding
syllables. However, the treatment of accent adopted here goes beyond word-level accent
and incorporates phrase level (intonational) accent as well. As we will see below, within
the category accent we can further distinguish between stress-accent and pitch-accent.
Finally, the term tone is used to refer to the distinctive pitch level of a particular syllable or
phrase edge. In section 4.5.1.2, I discuss details of two subtypes of tones, pitch accents and
boundary tones. Briefly, a pitch accent is a tone (or closely linked group of tones) that is
contrastively linked to a syllable, whereas the boundary tone is linked to the edge of a
phrase. This broad characterization of tone recognizes that it can be lexically assigned or
can be assigned by the intonation to a lexically predefined position in a word such as a
stressed syllable.
I examine both phonological and phonetic properties of accent and tonal systems as
I outline criteria for prosodic categorization. First, I review the arguments presented in
support of the different characterizations of JC prosody and then establish criteria by which
prosodic classification can be achieved. Based on these discussions, I present an analysis of
the JC word-level prosodic system in subsequent sections.
4.4.1. Previous Analyses of JC Prosody
There have been some proposals to characterize the JC word prosodic system as a tonal
system; each of these will be discussed in turn. Lawton (1963; 1968), provides perhaps
- 86 -
the earliest and so far only instrumental (phonetic) analysis of JC prosody. I give an
overview of his main points here and revisit them in detail in section 4.5.2 as a basis for
comparison with my own phonetic analysis of JC word-level prosody. Using both
phonological data and phonetic evidence from tonometer12 readings, Lawton argued that
JC has lexical tone. The tonometer readings involved listening to a tune played on the
tonometer and matching it with the perceived tune of the target utterance. Lawton’s
conclusions are based on analyses of monosyllabic, bisyllabic and trisyllabic words in
four different prosodic contexts, i.e. phrase medial, phrase final, phrase initial and
isolation. According to Lawton, the test words had the same ‘toneme pattern’ in isolation
as they did in the different sentence positions of the statement intonations in which they
were presented. Based on this, he identified ‘three relative tone levels (tonemes) that
function lexically and phonologically as a basic part of the vowel’, i.e. high-level, mid-
level and low-level tones. Examples of a small number of minimal pairs are given in
(56). In Lawton’s work, (  ) represents high-level tone; (  ) high-falling tone and (  )
low-level tone.
(56)  ‘mortar’  ‘matter, mucus’

  ‘feces’  ‘genitals’
 ‘beater’  ‘bitter’
12
The tonometer is an instrument used to control frequency variation in stimulus presentation or to register and
measure response. In terms of construction, one model consisted of a series of steel bars, suspended from strings at a
distance of 22.4% of the overall length from the ends. The rods vibrate transversely in their fundamental modes when
struck with a small hammer. For a set of round rods of constant diameter, the frequency is proportional to the inverse
square of the length. The longest bars were audible to everyone, but the shortest had a frequency of 32.8 kHz,
inaudible to the human ear.
(http://www2.kenyon.edu/depts/physics/EarlyApparatus/Rudolf_Koenig_Apparatus/Tonometer/Tonometer.html)
- 87 -
All the words with long vowels are said to differ from those with short vowels by having a
high-falling tone on the first syllable versus a mid-level tone. According to Lawton, there
are no phonemically long vowels in JC since all the phonetically long vowels carry high-
falling tone in all contexts such that their length is predictable from the tonal assignment.
Lawton also identified tonal differences in phrases like those in (57).
(57)  ‘Mary is brown’ HL H HL

 ‘Mary Brown’ HL L HL
 ‘can go’ L L

 ‘can’t go’ HL L
For Lawton stress in JC is not significant for distinguishing lexical items and is
predictable in terms of tone. He stated further that stress occurs initially on all two-
syllable free morpheme nominals regardless of whether the tone is high-falling or mid-
level and that final syllables are unstressed.
Sutcliffe (1986) also argues that JC is tonal, though not in the same way as
Lawton. He claims that an important part of the morphosyntactic system is expressed
through tone or significant pitch. The analysis is based on data from 45 male and female
speakers of Jamaican descent in Dudley, in the West Midlands of England. The speakers
were second generation Jamaicans ranging in age from 16 to 23 at the time of the
recording. Sutcliffe reports that the speakers were able to switch between a local variety
of British English and a basilectal Creole variety. He refers to this as ‘patois’ or British
JC and maintains that BJC is JC despite changes over the generation. Most importantly,
he claims that BJC shows the same patterns of tonality as JC that would be found in the
speech of Jamaican born speakers. It is not entirely clear what Sutcliffe means when he
- 88 -
says JC is ‘tonal’, though he seems to equate tone with pitch. On the one hand, he claims
that lexical tone, is either rare in JC or possibly non-existent. On the other hand, the
paper discusses how putative tone in BJC expresses grammatical relations, i.e.
grammatical tone. Alleyne (1980) has a similar view that ‘tone’ functions only at the
grammatical level in Jamaican. Sutcliffe claims that ‘nouns, adjectives and lexical verbs
alter tone according to their function as well as position in the sentence’. This suggest an
interpretation in terms of intonation patterns, that is, prominences associated with
different words based on sentence position and intended meaning. In fact, Sutcliffe
argues that ‘tonality is involved in the expression of sentence type and is affected in
actual realization by the rhetorical intent of the speaker’ e.g. a proceeding tone is realized
as H if its underlying tone is H and !H if the underlying is a L tone. At the same time,
Sutcliffe adopts Carter’s (1982) position that JC operates on a 2 tone system (L and H), to
yield upstepped, downstepped, LH and HL tones.
Carter’s research on JC prosody (1979, 1982, 1983, in Holm, 1988) has oscillated
between a characterization of JC as tonal and a characterization as pitch-accent or stress-
accent. In her 1979 work (cited in Carter, 1983), Carter argues that CECs including JC
exhibit some features typical of African tone-languages, for example, downdrift. The
main thesis of this earlier work was that the precise nature of these retentions of African
language prosodies was as yet undetermined and further that they may not be the same
across the CEC languages. In her 1982 paper, Carter concluded that tone-patterns in JC
were the property of the phrase or ‘tone-group’ (Holm, 1988:143). These tone-groups
were said to show patterns of pitch polarity, meaning high and low tone alternated so tone
- 89 -
depended on where a word fell in the alternation. This yielded differences in the
pronunciation of names such as Pául Nelsón compared to Kévin Nélson. This analysis
was based on the speech of a Jamaican-born student residing in Britain whose speech was
characterized as ‘educated Jamaican English’. Carter’s (1983) article is a critique of
evidence presented in support of the categorization of CECs as tonal, thereby reversing
her earlier classification. Most of the evidence evaluated is from Allsopp (1972 in Carter,
1983). She concludes that though there does not seem to be sufficient evidence for
classifying CECs as tonal, the evidence seems to point to a pitch-accent or intonational
interpretation. Carter points out that even the most convincing evidence for tonality in
CECs, minimal pairs, is inadequate since it is not clear whether the patterns are attested
only in citation form or whether they are maintained in contexts and if so which
contexts13. She argues that in the absence of distributional evidence one cannot properly
evaluate whether the distinction is tonal or further evaluate the contribution of context
information to the observed pitch patterns.
A second set of research characterizes JC prosody as a stress-based system.
Wells’ (1973) work, Jamaican Pronunciation in London is based on data from 36
Jamaican-born Londoners who had been in London between 4 and 14 years. The
informants ranged in age from 18-42 and were originally from 11 of the 14 parishes in
Jamaica. They were grouped by occupation as manual (roughly working class) and non-
manual (middle class and up). Wells carefully noted that his study was not a study of
Jamaican Creole but rather of the adapted form of speech used by Jamaicans in London
13
Carter (1987) essentially agreed with Allsopp’s (1972) findings that there are tonal contrasts in Guyanese Creole.
- 90 -
(pg. 30-31). The research focused on segmental phonology but Wells discussed prosody
in his background information on the language. Wells described JC prosody as a stress-
based system and clearly stated that stress is contrastive. He claimed that in disyllabic
words stress usually falls on the second syllable if it has a long vowel, diphthong or more
than one coda consonant, otherwise, stress falls on the first syllable. In the following
examples, stress is marked by (  ).
(58)  ‘allow’;  ‘owl’;  ‘comment’; kunu, kanu ‘boat’
Wells also identified exceptions such as, kobel ‘noise of a quarrel, din’; depan
‘keep on, be on’. He also noted that the word level stress is affected by a ‘characteristic
Jamaican intonation phenomenon which has the effect of shifting the surface stress by
one syllable’ (pg. 20), for example, the word  ‘kitchen’ has initial stress but in a
(declarative) sentence becomes . Like Lawton (1963), Wells reported that the
stressed syllable has a high falling pitch. He evaluated claims made by Bailey (1966 in
Wells, 1973) regarding the presence of a tonal contrast in kyan ‘can’ (low tone) and
kyaan ‘can’t’ (high tone). According to Wells, his Westmoreland informants accepted
identical intonation patterns for both words. This was also confirmed by observations of
the informants’ use of the words in discourse where both words occurred with a high
pitch. Wells hypothesized that the two words are homophonous and further that kyan
‘can’ was unaccented and kyaan ‘can’t’ was accented. The accented form, he claimed, is
used when the word must be made contrastive and always has a negative meaning and the
unaccented word a positive meaning. With regard to sentence intonation, he adopted
- 91 -
Bailey’s (ibid) three-way division of pitch contours, i.e. falling, rising and high-level.
According to Wells, the rising contour () is used in questions, interrogative and
imperative tags and non-final clauses, as in ju wa:nt it i:t it ‘if you want it, eat it’. In
addition, non-final main clauses may have a fall () as in eniwie mi wi i:t it wails mi
wa:n i ‘anyway I’ll eat it, provided that I want it’ (pg. 24). There is also a high-to-mid
fall () which may be a variant of the high-level contour and is seen in rhetorical
statements for example, no stju:pidnes dat ‘that’s ridiculous!’. Finally, Wells noted
that structural ambiguity may be resolved by ‘appropriate tonality as in standard accents’,
as seen from the following examples.
(59) : ‘don’t splash the water on me, boy!’

: ‘don’t splash the water on my boy!’
I should point out that Wells’ use of the term ‘tonality’ here refers to differences in the F0
contours associated with phrasing and not to lexical tone contrasts.
LePage (1960) and Cassidy and LePage (1967) also mention that JC has a stress
system but the matter is only dealt with cursorily. A more in-depth treatment is offered
by Alderete (1993) who assumes Cassidy and LePage’s (1967) classification of JC
prosody as stress-based. He presents a phonological analysis of the system. Alderete
argues that primary stress tends to fall on a heavy syllable and where there are no heavy
syllables, the first light syllable in the word is stressed as seen in the examples in (60).
- 92 -
(60). ‘sleep of children’
 . ‘something odd’
 .. ‘neglectful’
 .. ‘type of tree’
 .. ‘fan palm’
 ... ‘curdled milk’
Further, secondary stress falls on the third syllable of a trisyllabic word if it is heavy but
is not stressed if it is light. A quadrasyllabic word with all light syllables gets secondary
stress on the penultimate syllable as in (61).
(61) .b.  ‘rectangular platform’

..  ‘curdled milk’
Finally, in a later work, Allsopp (1996) argues that although general homogeneity
cannot be claimed for Caribbean English (CE) with respect to the prosodic systems, there is
a general sense in which a ‘West Indian accent’ is distinguishable for others. Allsopp
claims that the most distinguishing feature is the phrasal intonation. He claims that the
separation of syllabic pitch and stress in CE is a major factor of difference from other
Englishes like American English and British English. He notes (though incorrectly so) that
whereas stress tends to be coincident with high pitch in Standard English (SE) varieties,
like American English and British English, these are usually separated in general CE.
Notice that Allsopp assumes that the same syllable is stressed in SE as in CE. It appears
that for him both varieties are similar with regard to where stress is located but differ in
their phonetic manifestation of stress. Cassidy and LePage (1967) have a similar view of
- 93 -
stress in JC. They claim that stress and (high) pitch are not coincident but rather that high
pitch precedes the stressed syllable in JC.
A third characterization treats JC prosody as a mixed system. To my knowledge,
Devonish’s work (1989) is the only explicit treatment of JC prosody as a ‘mixed’ system.
Devonish (e.c.; 1989) claimed that the prosodic system of JC consists of a lexically
assigned HL melody, as well as an HL melody associated with word initial syllables in
addition to stress/segmental prominence associated with all syllables bearing surface HL
melody. Further, prominence is realised in stress-like ways in JC, mainly in terms of
extra length for prominent syllables and in the assignment of an HL melody. Whether the
first prominence in a word is located on the first or the second syllable is lexically
determined. In addition, whether or not the word has a second prominent syllable is also
lexically determined. These hypotheses are formalised in Devonish (2002) where he
argues that CECs with a basic 5-vowel system like JC are restricted tone languages with a
single underlying H tone and a surface HL tonal melody. He classifies these languages as
tone-driven stress systems, i.e. having prosodic systems in which both tone and stress
interact.
There are some difficulties with these earlier classifications of JC prosody. First,
some of the classifications are based on insufficient data or in some cases, claims are made
without supporting data. Another data problem is the failure to isolate the Creole varieties
in the analyses. In some cases it is not clear what variety of JC is being dealt with, in
other cases the analyses are based on Jamaican English, yet statements are made about JC
as a whole. As discussed in chapter 3, this is problematic since there are significant
- 94 -
phonological differences among the basilectal, mesolectal and acrolectal varieties in the
JC language continuum. Finally, with the exception of Lawton (1963) and Devonish
(2002) these studies are largely based only on impressionistic (auditory) analyses and
could benefit from instrumental analysis to support the claims made.
Second, some classifications are too global in that they group all CECs together.
Carter (1983) and Allsopp (1996) hinted that the prosodies may not be uniform across the
languages. Devonish (1989, 2002) presents sociohistorical evidence that there is a split in
the development of the synchronic prosodic phonologies of CECs14 which created, on the
one hand, languages like Guyanese, Ndjuka and JC with tone-driven stress systems (my
term) and, on the other, languages like Saramaccan and Berbice Dutch with stress-driven
tone systems (my term). My interpretation of the two types of CEC languages stems from
the following statements:
I suggest that two clear options faced the speakers of early Anglo-West African. The
first of these was to assume that the....HL melody was primary, with segmental
prominence being assigned to the syllable bearing the HL melody (2002: 52).
and in the second option;
Every word would have a HL melody assigned to it....automatically to the segmentally

prominent syllable.. (ibid: 52).
Third, some characterizations are one-sided in that they present only phonetic data
without phonological data or vice versa. This leads to a fourth problem which perhaps
stems from a failure to recognize that the use of pitch is not an exclusive property of tonal
systems as it is also used in stress-accent and pitch accent systems as well. Sutcliffe
14
Other researchers (cf. Alleyne, 1980) have noted the independent development of Caribbean Creole varieties.
- 95 -
(1986), for example, seems to equate the presence of pitch differences between words
with tonality. Alleyne (1980) notes that in some ‘Afro-American dialects, there is very
little agreement as to the analysis of suprasegmental features such as pitch, stress and
length.’ He attributes lack of agreement to ‘the fact that all three (features) are often
present and the selection of one rather than another as a distinctive feature depends on the
perspective of the analyst.’ One caveat is that at the time Lawton conducted his study,
researchers tended to equate ‘stress’ with intensity, loudness or increased effort on the
part of the speaker. Furthermore, for Lawton tone meant a ‘contrastive use of pitch’ so
that ‘any (my emphasis) language that has pitch (tone) contrasts..’ was classified as a
tonal language (pg. 18). However, Scott (1939), Fry (1955, 1958), Lehiste (1961),
among others, showed that for languages such as English and Serbo-Croatian the
phonological categories such as stress and accent are related to acoustic cues other than
intensity, i.e. fundamental frequency, duration, spectral characteristics of the vowel.
More recent studies replicated these results in a wide range of languages and, most
importantly, showed varied results for the intensity cue, indicating that intensity was not
as reliable as other cues to stress and accent. Still, it is significant that most of the later
studies on JC prosody appear not to have made this adjustment in their analysis of the
prosodic system. An adequate study of JC prosody needs to reflect scholarship on the
subject that will help to establish criteria for deciding along what parameters systems
differ when they do not have lexically contrastive tone specifications. As I suggest
below, this comes from an examination of both the associated phonological and phonetic
properties.
- 96 -
4.4.2. Criteria for Prosodic Classification
There are several ways in which stress differs from both tone and pitch accent and thus
there are several factors which can be used as a basis for classifying the prosodic system
of a language as one of these systems. I first discuss phonological factors and then
phonetic factors.
First, stress is culminative in the sense that each content word has a metrically
strong syllable that bears main stress (Hayes, 1995; Kager, 1995). Second, stress is
alternating by nature. This has to do with rhythmic distribution by which adjacent
stresses are avoided (Kenstowicz, 1994; Kager, 1995), e.g. racoon + coat --->
racooncoat. In this example, the stress on the syllable ‘coon’ would be adjacent to the
stress on the word ‘coat’ in the compound, i.e. *racooncoat. As a result, the stress on
‘coon’ is shifted one syllable to the left so that the stress pattern on the compound is a
rhythmic sequence of stressed-unstressed-stressed syllables. So, syllables bearing stress
tend to occur equally spaced apart, and further, unlike tones, stress does not assimilate, as
this would produce sequences of stressed syllables (i.e. clashes). Third, stress is
hierarchical; most of the languages surveyed tend to have different levels of stress, i.e.
primary, secondary, tertiary etc. (Hayes, 1995; Kager, 1995). Fourth, stress contrasts
tend to be manifested segmentally, for example, by vowel or consonant lengthening
(Kenstowicz, 1994; Kager, 1995) and conversely the lack of stress may be manifested in
systematic vowel reduction.
With regard to the phonetic properties, it has been noted that stress is perhaps the
most ‘elusive’ phonological entity (Kenstowicz, 1994) and that there are no invariant
- 97 -
cues. Beckman (1986) also mentions that it is expected that languages will vary in their
phonetic manifestation of stress, just as with the phonetic implementation of other
phonological categories like voice. Cross-linguistic research on the phonetic properties
of stress has shown however, that stress is realized through phonetic properties such as
fundamental frequency (F0), duration, vowel duration and intensity. Prominent syllables
potentially have longer durations, higher intensities and specific pitch movements
associated with the perception of prominence. According to Remijsen (2001), we can
distinguish three word-level prosodic categories, on the basis of both phonological
characteristics and phonetic encoding. That is, while pitch-accent is encoded by the F0
pattern, stress can be realized by means of other acoustic parameters such as vowel
quality, duration and intensity as well as by the association to pitch accent or tone.
Beckman (1984, 1986) makes a distinction between stress-accent and non-stress-accent
languages based on the degree to which they utilize phonetic properties other than the
tone (F0) pattern as cues to prominence. Further, in both pitch accent and stress accent
languages, there are differences in the F0 pattern of words. Whereas in pitch-accent
languages the shape of the F0 pattern of words is fixed at the lexical level, it is variable in
stress languages where the F0 contour of the word is determined at the phrasal level by
intonational patterns with a pragmatic meaning (Ladd, 1996). Furthermore, since the
phonological categories ‘pitch accent’ and ‘stress’ are independent we can observe
different types of interactions across languages, yielding at least four different groupings.
That is, languages can have both stress and lexically contrastive pitch accents (e.g. Serbo-
Croatian, Swedish); or stress with intonationally contrasting pitch accents (e.g. Greek,
- 98 -
Jordanian Arabic, English); lexically contrasting pitch accents without stress (e.g.
Japanese) or finally neither stress nor pitch accents (e.g. Korean) (cf. DeLacy, 2002)15.
In contrast to stress, lexical tone is contrastive in a paradigmatic way while both
lexical pitch-accent and lexical stress are syntagmatically contrastive, if at all (Beckman,
1984; 1986; Remijsen, 2001). Phonetically, lexical tone and lexical pitch-accent are
similar in the sense that both are encoded by means of F0. Stated in another way, pitch
accent languages are tonal at the lexical level, so they must satisfy the criterion of having
invariant tonal contours on tone-bearing syllables since the tone is a lexical property
(Hayes, 1995). Conversely, the tonal contour of accented syllables in stress languages
can vary freely as they need not be a part of the lexicon but may be assigned by the
intonation.
Beckman (1986) and Kenstowicz (1994) also mention other properties of stress.
First, the perception/judgement of the native speakers is different with regard to stress
and tone. The perception of tonal contrasts tends to be inextricably linked to segmental
contrasts whereas the perception of accentual contrasts is not linked to segmental
specifications. Second, tone and accent have different historical developments.
According to Beckman (1986) tonal oppositions function like segmental oppositions (i.e.
paradigmatically) because they develop historically from segmental oppositions, whereas
accentual contrast are removed from segmental oppositions since they develop from
culminative reanalyses of other suprasegmental patterns (pgs. 35-36).
15
I consider DeLacy’s discussion of tone-driven stress systems and stress-driven tone systems to be similar in spirit to
the classifications I make here.
- 99 -
The hypothesis which I propose here, is that JC falls into the category of stress-
accent languages since tones are associated to prominent syllables and are necessarily
accompanied by other phonetic cues to prominence. One indicator of syllable prominence
identified in this research is the alignment of the F0 contour with the word. As in other
stress systems, we will see that stress in JC is lexical, predictable and it alternates.
Furthermore, the JC stress system is largely of the variable, quantity-sensitive and
bounded variety.
4.5. Evidence for Stress in JC
Supporting evidence for categorizing JC as a stress system comes from two main sources,
strong native speaker intuitions from four informants (including the author) and the
phonetic data I discuss further below. I used the Dictionary of Jamaican English (DJE)
(Cassidy and LePage, 1967) as the initial source for stress assignment in JC. This was
bolstered by native speaker intuitions on syllable prominence in each word. Each of the
words taken from the DJE was elicited from the informants to cross-check for
consistency in the DJE notations for syllable prominence. In cases where there were
differences, I relied on native speaker intuitions. The informants were also asked to
judge the perceived strength of syllables in polysyllabic words as well as the
appropriateness of pronunciation for words with prominences in varied locations e.g.
baRAmpa vs. BArampa vs baramPA ‘rattooned ginger root’. In the sections below I
show phonetic evidence in support of the hypothesis that JC has a stress system. As
demonstrated, the phonetic difference between lexically contrastive words is attributable

- 100 -
to differences in the alignment of the F0 pattern with the word rather than to different tone
shapes. In section 4.5.1 I give a brief overview of the theoretical assumptions and the
elicitation techniques (section 4.6.2) used for data collection. I will motivate these
further below in section 4.6.
4.5.1. Theoretical Assumptions
4.5.1.1. Prosody
The view of prosody adopted here is directly in line with the phonological properties of
stress outlined in the previous chapter as well as with the metrical theory of stress
introduced there. Following the assumptions of metrical theory, I treat the prosody of an
utterance as a hierarchically organized structure of phonologically defined constituents
such as the syllable, foot, prosodic word etc. That is, prosody is crucially not synonymous
with suprasegmental features but is rather an organizational structure (Beckman, 1996).
From this perspective, we can view pitch accents as components of prosodic constituents
such as the syllable and we can view different levels of stress, i.e. primary, secondary etc.
as differentials in the levels of prominence contrast for these prosodic constituents.
Potential ambiguities such as that observed for intensive and distributive reduplication, can
therefore be resolved by reference to the prosodic organization of prominent syllables in
the word. As noted above, we can define a pitch accent as a tone (or closely linked group
of tones) that is contrastively linked to a syllable (Beckman, 1996: 30). Since in a stress
language, pitch accents are anchored to metrically strong syllables, if these syllables are
- 101 -
organized differently in different words then the alignment of the associated F0 will
necessarily differ.
In the following section, I discuss some phonetic evidence supporting the
hypothesis that the word-level prosodic system of JC is a stress-accent system. I argue that
JC has a typologically similar intonation system to other stress-accent languages like
Russian, German and English. As such, we can predict that the F0 pattern associated with
stressed syllables will vary depending on the prosodic context and the intonation. Further,
this typological classification predicts that JC has no lexically contrastive tones (pitch
accents) which can contribute a fixed shape to the F0 contour. I assume that JC has a
grammar of pragmatic tone morphemes, pitch accents, that are aligned with the head
syllables of prominent feet. A complete analysis of the intonational system of JC is beyond
the scope of this dissertation. However, for purposes of the ensuing discussion I assume,
based on the data presented, that there are at least two pitch accents, H* and H+L* and two
boundary tones H% and L%. I describe these in more detail just below.
4.5.1.2. Notations
The analysis of word-level prosody presented in this chapter follows in general the
autosegmental approach to intonational phonology (Beckman and Pierrehumbert, 1986).
One of the central tenets of this theory is that in English and other typologically similar
languages, pitch accents (tones) are associated with words in a specific way, the main
condition being that the pitch accents are aligned with prominent, i.e. metrically strong
- 102 -
syllables. As we will see in Chapter 5, where JC is concerned this is typically a heavy
syllable, though there are also stressed light syllables. The tones associated with stressed
syllables are the ‘starred’ (*) tones and as such we can refer to the associated syllable as
accented. Pitch accents can be monotonal (e.g. H*) or bitonal (e.g. H+L*), where the
asterisk (*) marks the tone associated with the stressed syllable. Other tones like the
boundary tone are aligned with the edge of phrases. For example, the low pitch contour at
the end of statements is marked with L% and the high pitch contour at the end of
questions is marked with H%.
4.5.1.3. F0 Plots
Figure 4.1 is a graph plot of fundamental frequency (vertical axis) over time (horizontal
axis). As shown here, statements in Jamaican Creole of the sort analysed in this chapter
typically show movement in pitch from a high point to a low point (dotted line), whereas
questions typically have a pitch movement from a low pitch to a high pitch (solid line).
- 103 -
200
170
140
buriburi
110
80
50
0.16 2.04
Time (s)
Figure 4.1: F0 contour of the statement Mieri want wan buriburi wan ‘Mary wants one
that has burrs all over’ (dotted line) and the related question Mieri want wan buriburi wan
Does Mary want one that has burrs all over? (solid line).
The graphs discussed in this and ensuing chapters were made using PRAAT 4.026 and are
annotated as illustrated in Figure 4.2.

buridnf DF
200
150
100
50
4000
2000
0
tones H+L* L%
phones b    b   
words mieri want wan buriburi wan
0.4 0.8 1.2 1.6 2

Time (seconds)
Figure 4.2: Annotation conventions

- 104 -
In each graph, the portion of the F0 associated with each target word is marked off by
vertical bars on the F0 traces (top row). A spectrogram (middle row) shows word
segmentation for the entire utterance and phone segmentation for the target word; these
are identified by vertical bars. The bottom row has transcription information on different
tiers; pitch accents (tone tier), phones in broad phonemic transcription (phones tier) and
words in a slightly modified version of the Cassidy and LePage (1967) writing system
(words tier). With regard to the tones tier, the notation H* and H+L* are used to denote
the two types of accents illustrated in the data. These are placed at the end of the syllable
on which the accent is located. The boundary tones, L% and H% are placed at the end of
the utterance.
4.5.2. Tone or Stress: F0 Pattern in Bisyllabic Words
The set of words I examine in this section have been classified as having contrasting
tonal patterns and which I have analysed as having a strong-weak versus weak-strong
stress pattern, for example,  ‘father’ contrasts with  ‘male religious leader’.
As I discuss in Chapter 5, the strong-weak stress pattern in these words is consistent with
the stress pattern in other bisyllabic words with an initial heavy syllable, e.g.  ‘type
of yam’ or with two light syllables, e.g.  ‘prickle, thorn’. In both cases, stress is
assigned to the initial syllable. The contrasting cases with a weak-strong stress pattern
are analysed as having lexically specified stress patterns since they do not reflect the
regular pattern. The main issue which will be addressed here concerns the shape and the
alignment of the F0 contour associated with these words. I illustrate using the pair of
- 105 -
words  ‘mother’ and  ‘female religious leader’ since the initial segment is a
sonorant and does not cause perturbation of the F0 contour as would the initial obstruent
of  ‘father’ and  ‘male religious leader’.
madais2 SG
250
200
150
100
4000
2000
0
tones H+L*
hones m a d a
words mada
0.4 0.6
Time (seconds)
Figure 4.3: Fundamental frequency contour (upper panel) and spectrogram (lower panel)
of the word  ‘mother’ produced in isolation. This word has primary stress on the
initial syllable.
As we see in Figure 4.3 there is a transition from a high pitch to a low pitch on the initial
stressed syllable of the word. As we see below, this F0 pattern is maintained in both final
and non-final position in statements (Figure 4.4).
- 106 -
madadf SG madadnf2 SG
250
200
150
100
4000
2000
0
tones H+L* L% H+L* L%
phones m a d a m a d a
words som yam fi mi mada mi mada yam dem
0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1

Time (seconds) Time (seconds)
Figure 4.4: Fundamental frequency contours (upper panels) and spectrograms (lower
panels) of the word  ‘mother’ produced in statement intonation in final position
som yam fi mi mada ‘Some yams for my mother’ (left graph) and in non-final position
mi mada yam dem ‘My mother’s yams’(right graph).
As with the citation form shown in Figure 4.5, when the word is final in the sentence
(nuclear position), there is a fall in pitch on the initial stressed syllable (left graph).
When the word is not final in the sentence (prenuclear), the same fall in pitch is observed
on the initial syllable and the low pitch is extended over the words that come after it
(right graph). Notice also that in all cases the high portion of the F0 pattern is within the
same syllable as the fall. I transcribe this pattern as H+L*, the starred tone indicating that
it is associated with the stressed first syllable.
When we examine the word in question intonation some differences are
observable. Figure 4.5 shows the word in final and non-final position.
- 107 -
madaynf SG madaynnf2 SG
400
300
200
100
4000
2000
0
tones H* H% H* H%
words a fi mi mada a fi mi mada yam
0.2 0.4 0.6 0.2 0.4 0.6 0.8 1

panel) of the word  ‘mother’ produced in question intonation in final position a fi
mi mada ‘Is it my mother’s?’ (left graph) and in non-final position a fi mi mada yam ‘Is
it my mother’s yam?’(right graph).
Producing sentences in question intonation involves raising the terminal F0 contour
to a high point (an upturn), i.e. a high boundary tone (H%). As we can see, the fall in pitch
onto the stressed syllable which was observed in the other contexts is absent here. Instead,
there is a high pitch associated with the stressed syllable which is then raised even higher
for the high final contour of the question intonation. (I treat this as a high pitch because
they correspond to the high pitch points in this speaker’s speech in statement intonation,
which both cases are at or just above 200 Hz. This shows that the utterance in question
intonation is produced in a higher overall pitch range than the one in statement intonation).
A similar pattern is observed when the word occurs in non-final position with question
intonation and in this case the high pitch is extended to the following words before the final
rise of the question intonation.
- 108 -
In the next set of examples, I examine the word ‘female religious leader’, which has
a weak-strong stress pattern, in these same contexts. Figure 4.6 shows the word in citation
form.
madarlis SG
250
200
150
100
4000
2000
0
tones H+L* L%
phones m a d a
words mada
0.4 0.5 0.6 0.7

Time (seconds)
of the word‘female religious leader’ produced in citation form. This word has
primary stress on the second syllable.
As we can see there is a fall in pitch into the stressed second syllable. In this case, the
high portion of the F0 is in the preceding unstressed syllable. I analyse this as a H+L*
pitch accent, the low portion of which is anchored to the stressed second syllable. A
similar pattern is seen when the word occurs in final and non-final position in statement
intonation as seen in Figure 4.7.
- 109 -
madarldf SG madarldnf SG
250
200
150
100
4000
2000
0
tones H+L*L% H+L* L%
words som yam fi mi mada mi mada yam dem
0.2 0.4 0.6 0.8 1 1.2 1.4 0.2 0.4 0.6 0.8 1
panels) of the word  ‘female religious leader’ produced in statement intonation in
final position, som yam fi mi mada ‘Some yams for my female religious leader’ (left
graph), and in non-final position, mi mada yam dem ‘My female religious leader’s
yams’(right graph).
In question intonation, the pitch remains relatively high throughout the word and in this
case I transcribe the pitch accent on the stressed syllable as a H* (Figure 4.8).
- 110 -
madarlynf SG madarlynnf SG
400
300
200
100
4000
2000
0
tones H* H% H* H%
words a fi mi mada a fi mi mada yam
0.4 0.6 0.8 1 0.4 0.6 0.8 1

panels) of the word  ‘female religious leader’ produced in question intonation in
final position a fi mi mada ‘Is it my female religious leader’s?’ (left graph) and in non-
final position a fi mi mada yam ‘Is it my female religious leader’s yam?’(right graph).
To summarize, for both words there are differences based on whether the word
occurs in statement or question intonation. In statement intonation and in the citation
form a H+L* was associated with the stressed syllable and in question intonation a H*
was associated with these syllables.
The observed patterns are generally consistent with the pattern seen for other
speakers. As I show however, some speakers maintain the F0 contrast between these
words even in question intonation, while some speakers do not. Figure 4.9 through
Figure 4.12 show representative examples of ‘mother’ (left graphs) and 
‘female religious leader’ (right graphs) in the different contexts produced by RP a 32 year
old male.
- 111 -
madadf1 RP madarldf1 RP
150
100
50
4000
2000
0
words som yam fi mi mada som yam fi mada
0.4 0.8 0.4 0.8

panels) of the word  ‘mother’ produced in statement intonation in final position,
som yam fi mi mada ‘some yams for my mother’ (left graph), and  ‘female
religious leader’ produced in statement intonation in final position, som yam fi mi mada
‘Some yams for my female religious leader’ (right graph).
As seen in the example above, the H+L* accent is aligned with the stressed syllable in
each word. A similar pattern is seen when the word occurs in non-final position in
statement intonation, as illustrated in Figure 4.10, and in final position in question
intonation as shown in Figure 4.11.
- 112 -
madadnf3 RP madarldnf3 RP
150
100
50
4000
2000
0
words mi mada want som yam mai mada want som yam
0.4 0.8 1.2 0.4 0.8

panels) of the word word  ‘mother’ produced in statement intonation in non-final
position, mi mada want som yam ‘my mother wants some yams’ (left graph), and 
‘female religious leader’ produced in statement intonation in non-final position, mi mada
want som yam ‘My female religious leader wants some yams’(right graph).
madaynf2 RP madarlynf RP
250
200
150
100
50
4000
2000
0
tones H+L* H% H+L* H%
words a som yam fi mi mada som yam fi mi mada
0.4 0.8 0.4 0.8

panels) of the word  ‘mother’ produced in question intonation in final position, a
som yam fi mi mada ‘Are these some yams for my mother?’ (left graph), and 
‘female religious leader’ produced in statement intonation in final position, som yam fi
mi mada ‘Are these some yams for my female religious leader?’ (right graph).
- 113 -
Notice that these patterns seen in question intonation for RP are different from the
productions of SG shown just above. That is, whereas the H+L* accent is maintained in
question intonation in RP’s productions, we observed a H* accent in SG’s productions.
Interestingly the H* accent appears in one production of  when it appears in
prenuclear (nonfinal) position in question intonation as shown in Figure 4.12. I take this
as evidence that the pitch contrast between the words can be neutralized in question
intonation.
madaynnf1 RP madarlynnf5 RP
200
150
100
50
4000
2000
0
tones H* H% H+L* H%
words mi mada wan som yam mi mada wan som yam
0.4 0.8 0.4 0.8

panels) of the word word  ‘mother’ produced in question intonation in non-final
position, mi mada want som yam ‘Does my mother wants some yams?’ (left graph), and
 ‘female religious leader’ produced in statement intonation in non-final position, mi
mada want som yam ‘Does my female religious leader wants some yams?’(right graph).
- 114 -
4.5.2.1. Summary and Discussion
To summarize, the difference between the words ‘mother’ and  ‘female
religious leader’, can be captured in terms of differences in the alignment of a HL F0
pattern with the word16. For ‘mother’, there is a low F0 target on the first syllable.
We see this manifested in the F0 contours as a fall in pitch into the stressed initial
syllable. For  ‘female religious leader’, there is also a low tonal target but on the
second syllable. In this case we saw a fall in pitch onto the stressed second syllable. In
both instances I analysed the pattern as a H+L* pitch accent. It was also noted that in
question intonation, the contrast is potentially neutralized since some speakers do not
maintain the H+L* accent. These patterns are summarized in Table 4.3.
Prosodic Condition SW WS
Citation H+L* (L%)17 H+L* (L%)
Statement final H+L* (L%) H+L* (L%)
Statement non-final H+L* (L%) H+L* (L%)
Yes-No question final H* (H%) H+L* ~ H*(H%)
Yes-No question non-final H* (H%) H+L* ~ H* (H%)
Table 4.3: Summary of F0 Patterns
For mada ‘mother’, the H+L* pitch accent is seen in citation form and in statement
intonation only, while for mada ‘female religious leader’, the pitch accent may be
16
One speaker (84 yr.old female) produced  ‘female religious leader’ with a L+H* pitch accent on the stressed
second syllable and  ‘mother’ with a L* accent on the stressed initial syllable.
17
The tones in parenthesis are boundary tones associated with the remainder of the phrase; L% for a statement and H%
for a question. Where the target word occurs in non-final position, only the pitch accent on the target word and the
final boundary tone are transcribed.
- 115 -
maintained in all contexts. I should also note that the pitch accents are in general not
affected by the position of the word in the sentence, i.e. final versus non-final position.
There are thus two types of differences between words with a stress-weak stress pattern
and those with a weak-strong stress pattern. First, there is a difference in the alignment
of the same pitch accent shape within the word in statement intonation and in citation
form. Second, there is a difference in the type of pitch accent they permit in question
intonation.
Given the data discussed in this section, the idea that there is a lexical tone
contrast between words like ‘mother’ and  ‘female religious leader’, is
untenable. If there were underlying tones, we would expect to find different F0 shapes for
the tone associated with each word. Further, we would expect to observe each contrasting
tone shape/tonal melody in the different prosodic contexts. In fact, we saw that in both
cases we have the same HL melody and what differs is the alignment of the pitch fall
with the stressed syllable of each word. Based on these data, it is reasonable to expect
other lexically contrastive words to exhibit similar differences in the alignment of the F0
contour with the word. These data also show that phonetically, the stressed syllable in JC
is associated with a pitch fall onto that syllable. Since there is no single location for main
stress in words, we can predict that the location of the fall will vary depending on the
position of the stressed syllable within the word. This is illustrated in the following
sections where I examine the F0 patterns of trisyllabic and quadrasyllabic
monomorphemic words and compound words.
- 116 -
4.6. Discovering Contrasting Cues to Word Prosody: Techniques
In chapter 2 we saw many cases in which reduplication created segmentally identical
words which were said to have contrasting prosodies. At the same time, we saw that
researchers had different views on what prosodic property was being dealt with. In the
previous section I presented some evidence that the relevant prosodic property for JC is
stress. In this section, I elaborate on the type of methodology undertaken in arriving at
this conclusion. First, I review general strategies used to discover contrasting acoustic
cues to word prosody in several languages and then in section 4.6.2 I describe the
methods of data collection for the JC data discussed herein.
Two forms of stress tend to occur in languages like English, lexical stress and
sentence level stress. Lexical stress is concerned with the emphasis of individual
syllables in polysyllabic words. Sentence level stress on the other hand is concerned with
the stress placed on words in order to indicate (or contrast) their importance/prominence
in relation to other words in a sentence. Since my primary concern is with word-level or
lexical stress, I focus mainly on strategies used for discovering acoustic cues to stress,
accent and intonation as these are directly relevant to the study of JC prosody given that it
is a stressed-based system.
In section 4.6.1 I discuss strategies used to discover cues to stress and accent
contrast in a variety of languages including, English, Serbo-Croatian, Japanese, Arabic
and Greek. In section 4.6.1.1. I discuss the general findings of these studies and then
discuss how they apply to the study of JC reduplication in section 4.6.2.
- 117 -
4.6.1. Strategies
Several early experiments were conducted to investigate the phonetic properties of stress
contrasts in segmentally identical English words. Among these are Scott (1939), Fry
(1955, 1958, 1968) and Huss (1978) among others. One recurring strategy used by these
studies is variation of the intonation contour in which the test material was presented
while keeping the segmental content constant. The F0 contour associated with the words
could be varied by prosodic context since its shape was not a property of the word as in
tonal systems. In this way while the F0 contour of the sentence frame is varied one can
observe whether and how the F0 contour associated with the target words change.
Scott 1939 for example, conducted an experiment using pairs of English words
with contrastive stress e.g. IMport (noun), imPORT (verb); INcrease (noun), inCREASE
(verb). Scott noted that when these words are spoken in isolation, intonation gives an
added clue to their identity but that in some instances intonation does not help in the
distinction. Eleven subjects where asked to write down sentences they heard. These
sentences had the noun form of the test items in a frame like ‘Are you sure wood-imports
would?’ and sentences with verb stress like ‘Are you sure wood imports would?’ Scott
found that in general listeners were able to get the meaning intended by the speaker. On
another occasion, one subject had only 15 out of 24 responses corresponding with the
speaker’s intention. Scott concluded that unaided, stress is not very efficient as a
distinguishing feature in English (pg. 45).
Fry (1955, 1958, 1965) conducted a series of more elaborate experiments to test
the effect of changes in duration, F0, intensity and vowel quality on English listeners
- 118 -
perception of stress. The stimuli for these experiments were synthesized tokens of the
type seen in Scott’s experiment; object, subject, digest, contract, permit. In a preliminary
experiment, Fry (1955) tested the effect of duration and intensity on the perception of
stress and concluded based on his results that duration was a more effective cue to stress
than intensity. The second experiment, Fry (1958) tested the effect of F0 variation in
conditions where the influence of sentence intonation was reduced to a minimum. The
experiment had synthesized sequences in which F0 remained constant throughout the
syllable and change in F0 was instead made between syllables. The results showed an
effect of changing duration similar to that found in the initial experiment. There was also
an effect of F0 change such that a step-up change of F0 change caused a smaller
percentage of noun responses and a step-down change caused a greater percentage of
noun responses. This result supported Fry’s hypothesis that if two syllables differ in F0,
the syllable with the higher F0 was more likely to be perceived as stressed. While a step
change in F0 affected stress judgments, as long as the results were perceptible as stressed,
the magnitude of the pitch change did not affect the perception of stress (pg. 144).
Huss (1978), conducted similar experiments with the same type of noun
verb contrast but using natural speech, e.g. import, decrease, increase, insult. Like Fry,
Huss tested for effects of F0, formant frequency, intensity and duration on subjects’
perception of stress. The sentential contexts were segmentally identical and were of the
type, ‘the German’s imPORT sinks’ vs. ‘the Germans’ IMport sinks’. These utterances
were placed into larger contexts in one of the following positions (a) post-nuclear, falling
pitch (b) post-nuclear, rising pitch; (c) nuclear, falling pitch, (d) nuclear, rising pitch.
- 119 -
That is, two positions for declarative sentences (falling pitch) and two for questions
(rising pitch). Huss found that there was a preference for the noun pattern in the post-
nuclear position regardless of pitch, confirming the hypothesis that the lexical stress
differences were neutralized in this context. For example, for declarative sentences
(falling pitch), nouns had higher 1st: 2nd syllable F0 ratios given that their first syllable F0
values were higher. In post-nuclear position however, the first and second syllables of
nouns and verbs were at approximately the same F0 level.
Other acoustic studies were conducted to characterize the attributes of speaker F0
(Lieberman, 1960; Lehiste, 1961; Bruce, 1977; Beckman, 1986; among others).
Lieberman (1960), for example, investigated the importance of changes in F0, peak
envelope amplitude and duration to the recognition of stressed syllables in English. As in
previous studies, the test items were contrasting noun-verb pairs like contract, digest,
conflict. Lieberman compared the acoustic values for F0, peak envelope amplitude and
duration for a stressed syllable to those for an unstressed syllable within a word e.g. con-
vs. -tract in contract, and across words e.g. the first syllable of conTRACT with the first
syllable of CONtract. This was done for all the words which had clearly perceived stress
patterns. Lieberman found that the stressed syllables had a higher F0 than the unstressed
syllables in the same word in 90% of the cases; they had higher peak envelope amplitude
in 87% and a longer duration in 66%. When he compared syllables across words,
stressed syllables had higher F0 than their unstressed counterpart in 72% of the cases;
higher peak envelope amplitude in 90% and longer duration in 70%. Based on these
- 120 -
results, Lieberman concluded that F0 and envelope amplitude are the most relevant
acoustic correlates of stress.
Comparable studies were also conducted with ‘pitch-accent’ languages such as
Serbo-Croatian and Swedish. Lehiste (1961) provided an acoustic phonetic description of
four accent types in Serbo-Croatian (Belgrade dialect); short-falling, long-falling; short-
rising, long-rising and provided an acoustic-phonetic description of the accent types.
Words with the different accent types were pronounced in a sentence frame that allowed
words from the same category with different inflectional forms to be alternated in the
same position. This allowed for cross comparison of the different words which occurred
in that sentence position. The results showed no difference in F0 patterns for disyllabic
words with short-falling and short-rising accents. There was however a difference in
three syllable words. For example, for the falling pattern in disyllabic words, the F0 peak
was reached towards the end of the stressed vowel (approx. 66% to 75% of the duration),
the fall took place on the second syllable. For the short rising pattern, the F0 peak was
also late in the first syllable but slightly later than in short-falling pattern, i.e. 86% of
duration vs. 70%. In three syllable words on the other hand, the F0 on the second syllable
in the short-rise pattern started higher than the peak in the first syllable and the third
syllable had low frequency.
Bruce (1977) conducted experiments in Swedish to determine how F0 is used to
signal word accent, sentence accent and terminal juncture. Swedish has two word
accents, accent I (acute) and accent II (grave), which are preserved in stressed syllables.
The test material consisted of sentence frames in which the following were varied (a)
- 121 -
word accent in different sentence positions (b) sentence accent location and the domain
of focus (c) number of syllables (d) sentence position for the word (final and non-final)
and (e) syntactic structure of part of the sentence containing a compound word or two
words. He concluded that the precise timing of the F0 fall associated with the two accents
determined how they are identified; in accent I, the fall occurs earlier than in accent II.
Beckman’s (1986) study compared accent patterns in English and Japanese
focusing on the acoustic correlates of accent in the two languages. For the production
study, the Japanese words were six minimal pairs. These were placed in 3 random orders
and presented to subjects who read them in a given sentence frame. The English data had
five minimal pairs (the same as was used in Fry’s studies). Beckman’s results indicated
that in Japanese there was a difference in F0 patterns that contrasted simple rises with
simple falls on contrasting test items; the English pattern was similar. In general, words
had a simple rising intonational pattern for nouns and a falling pattern for verbs.
Beckman interpreted this as evidence that accentual contrasts are realized as differences
in F0 patterns.
More recently, De Jong and Zawaydeh (1999) studied word-level prosody in
Jordanian Arabic. One of their aims was to examine the phonetic correlates of stress,
which was felt to be a marker of prominence at the word-level. De Jong and Zawaydeh
predicted that there would be three direct correlates of stress; increased duration,
increased intensity and more extreme formant values. They also predicted that there
would be one indirect correlate of stress, fundamental frequency patterns. More
specifically, if the prosodic system of Arabic was like that of English, the F0 associated
- 122 -
with stressed syllables was expected to vary depending on discourse (pg. 5). Their test
material consisted of 10 types of words spoken in 5 different prosodic conditions for both
statements and questions in both final and non final position, i.e. (a) isolation (b) phrase
final (statement) (c) non phrase final (statement) (d) phrase final (question) (e) non
phrase final (question). They recorded four female subjects. For the recording, each
word was spoken in a block consisting of five repetitions of a word in each of the 5
conditions in the order a ! b! d ! c !e. The word blocks were randomized.
Measurements included vowel durations, an estimate of the F1 value and the F0 at the
midpoint of each vowel. With regard to F0, stressed syllables in the target words in
statements showed an increased F0 value.
Arvanti (1992) discussed perceptual and acoustic properties of secondary stress in
Modern Greek. The research addressed the debates surrounding the presence and nature
of rhythmic stress and stress induced by the stress well-formedness condition (SWFC) by
which a lexical stress is allowed only on the last 3 syllable of a word. Arvanti questioned
(a) whether the stress added to a host-and-clitic group because of a SWFC violation is the
most prominent stress in the group as has been claimed in the literature, (b) whether this
added stress is perceptually distinct from lexical stress and if so, (c) whether secondary
stress and rhythmic stress were perceptually and acoustically identical and finally
whether there was any evidence for rhythmic stress. Two experiments were conducted.
The test material for the first experiment consisted of segmentally identical phrases with
different word boundaries. One member of the test pair had a lexical stress and a SWFC-
induced stress, e.g. ’ari ’sta su ‘class mark’ (su is a clitic). The other member had two
- 123 -
words with lexical stress on both syllables, e.g. ’ari ’stasu ‘Ari stop’. Two distracter
phrases were designed similar to the test phrases. All the phrases were incorporated into
larger sentences and introduced to speakers in a randomized list who produced 6
repetitions of each. For the perception task, the distracter phrases and test phrases were
extracted from the sentences and repeated twice. The stimuli were presented to 18
subjects who were asked to give both possible interpretations of the test phrases.
Identification rate for the test phrases was 49.66% and for the distracters 99.1%, showing
that contrary to previous findings, the SWFC-induced stress was the most prominent in
the host-and-clitic group and that the original stress of the host weakened.
The second experiment directly addressed Arvaniti’s third query about the
perceptual and acoustic similarity of secondary stress in the host of host-and-clitic group.
Four pairs of segmentally identical words with different spelling and different stress
patterns were chosen. One member of the pair had antepenultimate stress, the other had
stress on the last syllable. These were incorporated into segmentally identical sentences
(orthographically different). The word pairs in this set differed in the position of the
primary stress e.g. a’poxi ‘hunting net’ vs. apo’xi ‘abstention’; there were four pairs of
distracters. As in the first experiment, six repetitions of each sentence were recorded
from a randomized list. The stimuli were prepared as before and the same 18 subjects
made judgements; the test procedure was the same. The results indicated a 97.2%
identification rate for test words and 98.2% identification rate for distracters, showing
that rhythmic stress and secondary stress are distinguished and there are thus not the
same.
- 124 -
For the acoustic analysis, duration, amplitude (peak, root mean square, amplitude
integral) and F0 measurements were taken for vowel in a putative secondary stressed
syllable and compared with the same vowel in a putative rhythmic stressed syllable (pg.
408). The results showed that the F0 contours showed significant differences between SS
and RS words; the SS words had high F0 on the antepenultimate syllables while RS words
had very low F0.
In general cues to stress/accent contrasts include alignment of the F0 contour with
the words; changes in the F0 pattern associated with the word; changes in vowel quality
and vowel duration and changes in intensity, though different languages may rely to
different degrees on these acoustic cues to syllable prominence.
4.6.1..1. Discussion
The methodologies employed in the studies discussed in this section suggest ways in
which one might find out what the acoustic cues to the stress contrast in intensive and
distributive reduplication are. In order to find out what the acoustic cues to stress or
accent contrasts are in the putatively contrastive words a combination of tasks was used.
First, comparable sentence frames were used for each target word such that they could be
commuted in the same syntactic position without altering the grammaticality of the
sentences. Second, target words are examined in several different prosodic contexts and
in several different intonations to see how and if the phonetic properties differ in the
different contexts. The phonetic analyses presented in this dissertation closely mimic the
strategies outlined here. Recall from earlier discussions that Lawton (1963) also
- 125 -
examined several monomorphemic words in different prosodic contexts, i.e. phrase
initial, phrase medial, phrase final and citation. According to Lawton, the pitch pattern of
the target words did not change from that shown in citation form. Lawton’s analysis has
an unfortunate gap in that no description is given for some of the target words in citation
form as a means of verifying the proposed pitch pattern in context. In addition, the data
was only examined in declarative intonation. Crucially different from Lawton’s analysis,
this analysis is accompanied by F0 traces of a sample of words in all the appropriate
contexts and examines the words in varying intonational contours.
In the phonetic study of reduplicated words I discuss in Chapter 6, I focus
primarily on the alignment of the F0 contour with the stressed syllables in the words and
on the realization of (pitch) accents on these syllables. The study of JC word-level
prosody in this dissertation is thus primarily a production study since we first need to
have a clear understanding of what the prosodic properties are before we can elaborate on
how these properties are exploited by listeners. The data I present suggest that the
percept of stress contrast may be elicited by differences in the location of the F0 fall in
words.
4.6.2. Method for Data Collection
4.6.2.1. Speakers
The production data reported on in this section are from six (6) native speakers from the
Top Alston community in addition to the author (SG). There are three females and four
males ranging in age from 29 to 87 at the time of the recording. These participants were
- 126 -
compensated for their time. Data for reduplicated words and non-reduplicated words are
reported on for all of these speakers (RP, EW, WW, HF, HH, SP(m), SG). All the
speakers for whom reduplicated data are described have a clear perception of the contrast
between intensive and distributive reduplications. This is based on the results of a small
listening task in which the speakers were presented with pictures depicting an intensive
meaning or a distributive meaning. They listened to two repetitions of a reduplicated
word (9 x 2 = 18) and were asked to choose the picture which best fit the meaning
expressed by the word they heard. Following this, the participants were asked to explain
the choice they made. With regard to SG, a listening task was administered with the help
of a phonetically trained graduate student at OSU. SG listened to two randomized lists of
28 reduplicated words (28 x 2 = 56) and identified the type of reduplication as the word
was heard. The words were sorted in two groups, citation forms and statements in one
group and questions in the other group.
4.6.2.2. The Data
The analysis is based on several recordings of monomorphemic words with different
stress patterns, bisyllabic and quadrasyllabic nominal compounds and reduplicated words
with intensive and distributive meanings as shown in Table 4.4. The bisyllabic and
trisyllabic words varied according to the position of the stressed syllable in the word.
Stressed syllables are marked with (  )
- 127 -
bisyllabic  ‘mother’
 ‘female religious leader’
trisyllabic  ‘enamel’
 ‘undershirt’
quadrasyllabic  ‘worthless person’
 ‘alligator’
 ‘type of wood’
compounds  ‘pregnant woman’
reduplicated  ‘very yellow; scattered yellow’
 ‘very green; scattered green’
 ‘very sticky; sticky all over’
 ‘very black; black all over’
 ‘very long’
 ‘very low’
 ‘very near’
Table 4.4: Elicited Words used in Analysis
All the target words were produced in carrier sentences such that they were direct objects
of main verbs or copula constructions. The words were produced in the five different
prosodic conditions listed in Table 4.5.
Prosodic Condition Pragmatic Context
(a) Isolation (c) Non-final (prenuclear) Statement

(b) Final (nuclear)
(d) question - final (e) Non-final (prenuclear) Yes-No Question
Table 4.5: Prosodic Conditions for Target Words
These conditions placed the target words in final and non final position in the context of a
broad focus statement, a pragmatically neutral yes-no question, and in citation form. The
- 128 -
final position in the sentence corresponds to having the word in nuclear position, for
example, im want wan we blakblak ‘he wants one that is very black’, and the non-final
position to having the word in prenuclear position, for example, im want di blakblak wan
an moni ‘he wants the very black one and (some) money’. The nuclear position (location
of the nuclear accent) is generally defined as the location for the last accent in an
intermediate phrase and prenuclear position as the location for all the accents appearing
before the nuclear accent. As discussed above, the literature on the acoustic properties of
stress contrasts in a variety of languages shows that these properties are not uniform
across all prosodic contexts in that some contexts better preserve the contrasts than
others. Based on this, it reasonable to expect the production of JC speakers to be
similarly heterogeneous across the contexts in which the test items were elicited.
4.6.2.3. Elicitation Technique
All interactions between the participants and the researcher were done mostly in JC. All
the utterances for all the subjects except SG were elicited using a set of pictures and
contextual information to cue the desired intonation and remind the speaker of the carrier
sentence. The researcher first discussed the different types of sentences being targeted
with each participant. That is, one with the target word at the end (final) and one with
other words after the target word (non-final), for both a statement and a question. Each
participant was then presented with a picture depicting the target word and was asked
questions to elicit the target word. Following that, questions were asked to elicit the
different utterances types. In cases where the target word was not known, no further
- 129 -
attempts were made to elicit productions for that word. The example in (62) illustrates
part of a typical elicitation session.
(62) Researcher shows picture of a pregnant woman

[In JC]: Yu nuo wan ada niem fi dis person (pointing to picture)
Do you know another name for this person?
Informant: beliuman ‘pregnant woman’ lit .belly woman (citation)
Researcher gives context:
Wan komyuniti dakta want fi chekup pan dem kaina pipla ya an yu nuo, so gi
smady di infamieshan dat im want som a dem pipl ya.
A community doctor wants to check up on these kinds of people and you know
this, give someone this information that he wants some of these people (shown in
the picture).
Informant: Im wan som beliuman (statement final)

He wants some pregnant women.
Researcher: Dis dakta greedy tu and im want som yam so tel smady se im want
som a dem pipl ya and som yam.
This doctor is also greedy and he wants some yams, so tell someone
that he wants some of these people (shown in the picture) and some
yams.
Informant: Im wan som beliuman an som yam (statement non-final)

He wants some pregnant women and some yams.
Researcher: Yu si di dakta an yu nuo we dem kaina pipl ya liv so ask im if im

want som a dem.
You see the doctor and you know where these kinds of people
(shown
in the picture) live so ask him if he wants some of them
Informant: Yu want som beliuman? (question final)

Do you want some pregnant women?
Researcher: Sins as yu plant yam tu yu kyan giv di dakta som so ask im if im
want som a dem pipl ya and som yam.
Since you plant yams as well you can give the doctor some, so ask
- 130 -
him if he wants some of these people (shown in the picture) and some
yams.
Informant: Yu want som beliuman an som yam? (question non-final)
Do you want some pregnant women and some yams?
The interaction with each informant lasted for an average of 45 minutes. The elicitation
sessions for the reduplicated words was done 1 month prior to that for the unreduplicated
words and compounds.
The recordings for SG were done in a sound attenuated booth in the phonetics lab
in the Department of Linguistics, OSU, to a Sony DTC-790 digital audio tape deck
recorder. All the other recordings were done in a quiet room in a home in Top Alston, to
a Sony PCM-M1 digital audio portable recorder. All the subjects wore a Shure SM-10a
head-worn microphone and were instructed to speak at a normal rate and as naturally as
possible. Both sets of data were digitized at a sampling rate of 22050 Hz, 16 bit using
Creative Wave Studio (v. 4.50.08). F0 calculation was done using the F0 tracking utility18
in PRAAT 4.0.26. For the purposes of getting an uninterrupted pitch track, to the extent
possible, the words and sentences were made up of sonorants.
18
The online PRAAT manual indicates that this is an autocorrelation method as described in Paul Boersma. (1993).
Accurate short-term analysis of the fundamental frequency and the harmonics-to-noise ratio of a sampled sound. In
Proceedings of the Institute of Phonetic Science of the University of Amsterdam. Vol. 17: 97-110.
- 131 -
4.7. F0 Pattern in other Monomorphemic Words and Compounds
4.7.1. F0 Pattern in Trisyllabic Words
In this section I examine the F0 contours of trisyllabic words. I look at two representative
words with primary stress in different locations, i.e. penultimate  ‘men’s
undershirt’, and final  ‘enamel’.
mirina SG
300
250
200
150
100
4000
2000
0
tones H+L* L%
phones m   i n a
words miriina
1.2 1.4 1.6

Time (seconds)
Figure 4.13: Fundamental frequency contour (upper panel) and spectrogram (lower
panel) of  ‘men’s undershirt’ produced in citation form. This word has primary
stress on the penultimate syllable.
As shown in Figure 4.13, there is a HL F0 pattern associated with the penultimate stressed
syllable of the word  ‘men’s undershirt’ and both the low portion and the high
portion of the F0 are located within the stressed syllable. I transcribe this a H+L* pitch
accent reflecting the fact that there is a low tonal target on the stress syllable and a
preceding high.
- 132 -
In Figure 4.14, the word occurs in final and non-final position in statement
intonation and the H+L* pitch accent is maintained.
mirinadf1 SG mirinadnf1 SG
300
250
200
150
100
4000
2000
0
phones m   i n a m   i n a
words im want wan miriina im want wan an som yam
0.8 1.2 1.6 0.8 1.2 1.6 2

panels) of  ‘men’s undershirt’ produced in statement intonation final position, im
want wan miriina ‘He wants a men’s undershirt’ (left graph) and non-final position, im
want wan miriina an som yam ‘He wants a men’s undershirt and some yams’ (right
graph).
Figure 4.15 shows the word in final and non-final position in question intonation. In
these examples, the H+L* is also maintained in both contexts.
- 133 -
mirinaynf1 SG mirinaynnf1 SG
400
300
200
100
4000
2000
0
phones m   i n a m   i n a
words im want wan miriina im want wan miriina an som yam
0.4 0.8 1.2 0.4 0.8 1.2 1.6

panels) of ‘men’s undershirt’ produced in question intonation final position, im
want wan miriina ‘Does he want a men’s undershirt?’ (left graph) and non-final position,
im want wan miriina an som yam ‘Does he want a men’s undershirt and some yams?’
(right graph).
The examples in Figure 4.16 through Figure 4.18 show  ‘enamel’ which has
primary stress on the final syllable. As illustrated by the citation form shown in Figure
4.16, we see the now familiar fall in pitch onto the stressed syllable. As before, I analyse
this as a H+L* pitch accent on the stressed final syllable.
- 134 -
inamelis1 SG
300
250
200
150
100
4000
2000
0
tones H+L* L%
phones  n a m ! l
words inamel
0.2 0.4 0.6

Time (seconds)
panel) of the word  ‘enamel’ produced in citation. This word has primary stress
on the final syllable.
When the word occurs in statement intonation as in Figure 4.17, we see the same H+L*
accent associated with the stressed final syllable. Notice however, that in these examples,
the high portion of the accent is in the preceding syllable. Perhaps it is this kind of
observation which led some researchers like Cassidy and LePage (1967) to conclude that
high pitch always precedes the stressed syllable in JC.
- 135 -
inameldf2 SG inameldnf1 SG
300
250
200
150
100
4000
2000
0
phones  n a m ! l  n a m ! l
words im want wan we a im want wan wan an som yam
0.4 0.8 1.2 1.6 0.8 1.2 1.6 2

panels) of the word ‘enamel’ produced in statement intonation final position, im
wan we a inamel ‘He wants one that is (made of) enamel’ (left graph) and non-final
position, im wan we a inamel an som yam ‘He wants one that is (made of) enamel and
some yams.’ (right graph).
Figure 4.18 shows the word in question intonation and again we see a H+L* accent with
the low portion within the stressed syllable and the high portion preceding it. The pitch
then rises to the high final contour of the question intonation. That is the F0 contour
associated with these words shows a fall-rise melody in question intonation.
- 136 -
inamelynf3 SG inamelynnf3 SG
400
300
200
100
4000
2000
0
phones  n a m ! l  n a m ! l
words im want wan we a inamel im want wan inamel wan an som yam
0.4 0.8 1.2 0.4 0.8 1.2 1.6

panels) of the word  ‘enamel’ produced in question intonation final position, im
wan we a inamel ‘Does he wants one that is (made of) enamel?’ (left graph) and non-final
position, im wan we a inamel an som yam ‘Does he wants one that is (made of) enamel
and some yams?’ (right graph).
4.7.1.1. Interim Summary
To summarize, in trisyllabic words there is a low tonal target on the stressed syllable of
words which is preceded by a high. This high can either be within the same syllable or
on the preceding syllable. In both cases, this is transcribed as a H+L* pitch accent.
Further, this pattern is maintained in both prosodic contexts considered: final and non-
final in both statement and question intonation. The main differences observed have to
do with how the H+L* accent is aligned with the words and also the type of intonation
with which they are produced.
- 137 -
4.7.2. F0 Pattern in Quadrasyllabic words
In this section, I examine the F0 contour of quadrasyllabic words. Recall that the
predominant stress pattern on these words is primary stress on the penultimate syllable.
The graphs shown are for a word with all light syllables  ‘alligator’ and for a
word with a heavy penultimate syllable,  ‘type of wood’. Figure 4.19 through
Figure 4.21 show utterances produced by SG and Figure 4.22 through Figure 4.24 show
utterances produced by HH, a 64 year old female.
As shown in the citation form of  ‘type of wood’ in Figure 4.19 there
is a fall in pitch onto the stressed penultimate syllable of the word. As before, I analyse
this as a H+L* accent. Notice also that the high portion of the F0 is in the preceding
syllable.
gyalimentais SG
250
200
150
100
4000
2000
0
tones H+L* L%
phones g a l  m ! n t a
words gyalimenta
0.2 0.4 0.6 0.8

Time (seconds)
panel)of  ‘type of wood’ produced in citation. This word has primary stress
on the penultimate syllable.
- 138 -
When the word appears in final and non-final position in statement intonation, as shown
in Figure 4.20, we also observe a fall in pitch onto the stressed penultimate syllable. As
in the citation form, the high portion of the F0 is located in the preceding syllable. In
both of the examples shown in the pattern is transcribed as a H+L* pitch accent.
gyalimentadf SG gyalimentadnf SG
250
200
150
100
4000
2000
0
phones g a l  m ! n t a g a l m ! n t a
words im want wan outa gyalimenta im want wan outa gyalimenta an mahuo
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2

panels) of  ‘type of wood’ produced in statement intonation in final position,
im want wan outa gyalimenta ‘He wants one of galimeta wood’ (left graph) and in non-
final position, im want wan outa gyalimenta an mahuo ‘He wants one of galimeta wood
and mahoe’ (right graph).
A similar pattern is seen when the word occurs in question intonation, as shown in Figure
4.21. Notice that when the word occurs in non-final position, where another accented
word ‘mahoe’ follows, the H+L* pitch accent is still realized on the stressed syllable.
- 139 -
gyalimentaynf SG gyalimentaynnf SG
300
200
100
4000
2000
0
phones g a l  m ! n t a g a l  m ! n t a
words yu want wan outa gyalimenta yu want wan outa gyalimenta an mahuo
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2

panels) of  ‘type of wood’ produced in question intonation in final position,
im want wan outa gyalimenta ‘Does he want one of galimeta wood?’ (left graph) and in
non-final position, im want wan outa gyalimenta an mahuo ‘Does he want one of
galimeta wood and mahoe?’ (right graph).
The patterns seen in the data described above are comparable to those produced by other
speakers. The graphs below, (Figure 4.22 and 4.23), show representative F0 contours of
 ‘alligator’ produced by HH. In these examples, there is a consistent fall in pitch
on the stressed penultimate syllable of the word as we saw in the previous examples.
This high-low melody is analysed as a H+L* pitch accent.
- 140 -
aligetais HH
250
200
150
100
4000
2000
0
tones H+L*
phones a l  g ! t a
words aligeta
0.4 0.6 0.8

Time (seconds)
panel) of  ‘alligator’ produced in citation. This word has primary stress on the
penultimate syllable.
aligetadf2 HH aligetadnf2 HH
250
200
150
100
4000
2000
0
phones a l  g ! t a a l  g ! t a
words im want wan aligeta im want som aligeta an som yam
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2 2.4 2.8
panels) of  ‘alligator’ produced in statement intonation in final position, im want
wan aligeta ‘He wants an alligator’ (left graph) and in non-final position, im want wan
aligeta an som yam ‘He wants an alligator and some yams’ (right graph).
- 141 -
aligetaynf4 HH aligetaynnf1 HH
300
250
200
150
100
4000
2000
0
phones a l  g ! t a a l  g ! t a
words yu want wan aligeta yu want wan aligeta som yam
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2

panels) of  ‘alligator’ produced in question intonation in final position, im want
wan aligeta ‘Does he want an alligator?’ (left graph) and in non-final position, im want
wan aligeta an som yam ‘Does he want an alligator and some yams?’ (right graph).
As we can see from these representative quadrasyllabic words, there is a consistent fall in
pitch onto the penultimate stressed syllable which is maintained in all contexts. I analysed
this as a H+L* pitch accent to reflect the observation that there is a low tonal target on
the stressed penultimate syllable. As with trisyllabic words, being in final or non-final
position does not affect the type of pitch accent associated with the stressed syllable in
quadrasyllabic words.
- 142 -
4.7.3. F0 Pattern in Compound Words
In this section I examine the F0 pattern of quadrasyllabic compound words. I focus on
these types since they provide a source of comparison with monomorphemic words and
reduplicated words with similar syllable structures. I include the comparison with
monomorphemic words in this section and leave the comparison with reduplicated words
for Chapter 6 where I discuss the F0 pattern of reduplicated words.
beliumanis SG
250
200
150
100
4000
2000
0
tones H+L*
phones b ! l  u m a n
words beliuman
0.4 0.6 0.8 1

Time (seconds)
panel) of ‘pregnant woman’ produced in citation.
As shown in Figure 4.25 there is a fall in pitch onto the penultimate syllable of the word,
 ‘pregnant woman’, which is indicative of a H+L* pitch accent, as seen for the
quadrasyllabic words in section 4.7.2.
Figure 4.26 and Figure 4.27 show the word in statement intonation and question
intonation respectively. As we has seen for quadrasyllabic words, in this case the H+L*
- 143 -
pitch accent seen in citation form is maintained in both intonations in final and non-final
position in the utterances.
beliumandf1 SG beliumandnf1 SG
300
250
200
150
100
4000
2000
0
phones b ! l  u m a n b ! l  u m a n
words im want wan beliuman im want wan beliuman an som yam
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2

panels) of  ‘pregnant woman’ produced in statement intonation final position,
im want wan beliuman ‘He wants a pregnant woman’ and in non-final position, im want
wan beliuman an som yam ‘He wants a pregnant woman and some yams.’
- 144 -
beliumanynf1 SG beliumanynnf1 SG
400
300
200
100
4000
2000
0
phones b ! l  u m a n b ! l  um a n
words yu want wan beliuman yu want wan beliuman an som yam
0.4 0.8 1.2 0.4 0.8 1.2 1.6

panels) of  ‘pregnant woman’ produced in question intonation final position, im
want wan beliuman ‘Does he want a pregnant woman?’ and in non-final position, im
want wan beliuman an som yam ‘Does he want a pregnant woman and some yams?’
The patterns seen here are comparable to those produced by other speakers. The graphs
below show representative F0 contours of  ‘pregnant woman’ produced by SP
(Figure 4.28).
- 145 -
beliumandf2 SP(m) beliumandnf1 SP(m)
150
100
50
4000
2000
0
words want wan beliuman want wan beliuman an som yam
0.4 0.8 1.2 0.8 1.2 1.6 2 2.4

panels) of  ‘pregnant woman’ produced in statement intonation final position,
(im) want wan beliuman ‘He wants a pregnant woman’ and in non-final position, (im)
want wan beliuman an som yam ‘He wants a pregnant woman and some yams.’
As in the SG productions, there is a fall in pitch onto the stressed penultimate syllable of
the compound, which is transcribed as a H+L* pitch accent. The same pitch accent is
seen in the examples below in Figure 4.29 when the word is produced in question
intonation.
- 146 -
beliumanynf4 SP(m) beliumanynnf2 SP(m)
150
100
50
4000
2000
0
words yu want wan beliuman yu want wan beliuman an som yam
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2

panels) of  ‘pregnant woman’ produced in question intonation final position yu
want wan beliuman ‘Do you want a pregnant woman?’ and in non-final position yu want
wan beliuman an som yam ‘Do you want a pregnant woman and some yams?’
In summary, in quadrasyllabic compounds there is a consistent fall in pitch onto the
stressed penultimate syllable which is maintained in all contexts. As before, we can
analyse this as a H+L* pitch accent to reflect the observation that there is a low tonal
target on the stressed penultimate syllable. As seen for trisyllabic and quadrasyllabic
words, being in final or non-final position does not affect the type of pitch accent
associated with the stressed syllable.
- 147 -
4.7.4. Section Summary
The main points from this section are summarized as follows. First, there is a pitch fall
onto the stressed syllable which appears on the words in both nuclear and prenuclear
positions in sentences and in citation form. This high-low melody can be analysed as a
H+L* pitch accent in which the low is associated with the primarily stressed syllable of
the word. The high portion of the F0 may be in the same syllable as the low or in the
preceding syllable. Second, we also saw that a H* pitch accent may also be associated
with the stressed syllable in bisyllabic words occurring in question intonation. Another
relevant observation is that only the most prominent syllable in the word is assigned a
pitch accent so, syllables with secondary stresses an unstressed syllables are considered
unaccented. For example, while  ‘enamel’ has primary stress on the final syllable
and secondary stress on the initial syllable a pitch accent is only associated with the final
syllable. This falls out naturally from the prosodic structure of the word as depicted by
the metrical grid representation in (63).
(63)
x Prwd accented
x x Foot stressed
x x x

Metrical grid representation of  ‘enamel’.
Only the final syllable in enamel gets accented because it is in a metrically stronger
position than other syllables.
- 148 -
Third, the difference between lexically contrastive words can be accounted for in
terms of differences in the alignment of the pitch accent with the word. All of these
observations support the proposal of JC word-level prosody as a stress-based system
rather than a tone-based system.
Table 4.6 summarizes the type of pitch accents on trisyllabic words,
quadrasyllabic words and compounds.
Prosodic Condition
Citation H+L* (L%)
Statement final H+L* (L%)
Statement non-final H+L* (L%)
Yes-No question final H+L* (H%)
Yes-No question non-final H+L* (H%)
Table 4.6: Summary of Pitch Accents in Trisyllabic, Quadrasyllabic and Compound

Words
As discussed above, several researchers have made statements regarding the F0 pattern
associated with stressed syllables in JC. All of these views are in essence rather similar.
Lawton (1963), for example, stated that stressed syllables have a high-falling tone while
Wells (1973) claimed that stressed syllables in JC have a high-falling pitch. Cassidy and
LePage (1967) reported that high pitch preceded the stressed syllable and Devonish (e.c.;
1998; 2003) suggests that stressed syllables have a high-low melody. Based on the data
presented in this chapter, it is clear that these observations are essentially correct. That is,
- 149 -
when we examine the test words in statement intonation and in citation form, we saw that
the stressed syllable in JC has HL F0 pattern, the low portion of which is associated with
the stressed syllable. Notice, however, that a high pitch does not always ‘precede’ the
stressed syllable as Cassidy and LePage suggest. For example, in bisyllabic words with a
strong-weak pattern, the high pitch peak was contained in the same syllable as the fall,
whereas in words with a weak-strong pattern, the high preceded the stressed syllable.
The crucial observation is that the shape of the F0 is not contrastive. Rather, lexical
contrasts are signaled by differences in the alignment of the same pitch accent shape with
the word rather than by different tone patterns. I propose therefore that native speakers
determine which syllable is stressed based on where in a word the fall in pitch occurs.
The only exception to this generalization was with bisyllabic words in question intonation
where we saw that in some cases there was no fall in pitch onto the stressed syllable. In
these cases, I would suggest that the listener would have to rely on some other acoustic
cue for the percept of stress such as vowel duration or differences in vowel quality.
The observation that lexical contrasts in JC are signaled differences in F0
alignment allowed for a characterization of the JC prosodic system as a stress system.
This places JC with language with typologically similar systems such as Russian,
German and English. The phonological grammar of stress languages like JC indicates
that the tones of a pitch accent are aligned with stressed syllables. The specific question
which will be addressed in chapter 6 is whether the segmentally identical reduplicated
words are also differentiated by timing differences of a pitch fall onto stressed syllables
or timing differences in the alignment of the F0 peak with the word.
- 150 -
CHAPTER 5
WORD LEVEL STRESS IN JAMAICAN CREOLE
5.1 Introduction
The data presented in the previous chapter supports a classification of the JC word-level
prosodic system as a stress-based system. This is a significant claim, especially since
researchers have characterized JC prosody differently or have otherwise failed to
characterize it at all. Recall for example, that researchers like Lawton (1963, 1968) and
Akers (1981) claimed that stress is not distinctive in JC. The data discussed here
illustrates the word level stress patterns in the language and further demonstrates that
stress distinguishes between words.
In this section, I give some background information on the data taken from
Cassidy and LePage (1967). In the following sections, I discuss the stress patterns on
monomorphemic words and compound words. I also present data for words which have
stress patterns that fall outside of the more generally observed pattern. The discussion is
based on a sample of 200 monomorphemic content words and 84 compound words taken
primarily from the Dictionary of Jamaican English (DJE) (Cassidy and LePage, 1967),
along with some elicitations from native speakers. I used the DJE as the primary source
- 151 -
for primary stress assignment in the words, supplemented by native speaker intuitions on
syllable prominence. Cassidy and LePage mark syllables with a high rising pitch with an
acute accent. Although they do not explicitly say that this notation should be interpreted
as primary stress in JC, when crosschecked with native speaker intuitions, in the majority
of cases, the syllable marked with the acute accent coincided with native speaker
judgments regarding which syllable was prominent. In cases where they did not match, I
relied on the native speaker’s judgment.
The discussion of stress presented here shares similarities with Alderete’s (1993)
previous analysis of JC stress but differs in some important respects which I discuss in
detail just below and in section 5.4.
5.2 Stress Patterns
Stress assignment in JC makes reference to several phonological parameters such as
syllable weight and syllable position. In the data shown, the syllables bearing stress are
marked with ( " ) for primary stress and ( Æ ) for secondary stress and syllable boundaries
are indicated by a period. I assume the maximal onset approach to syllabification and use
the notations L and H to refer to light and heavy syllables respectively.
5.2.1 Bisyllabic Words
As seen in the examples below, bisyllabic words typically have initial stress when both
syllables are light as in (64a). However, if the word has a heavy syllable, as in (64b) and
(64c), then the heavy syllable is stressed (see also Wells (1973) and Alderete (1993)).
-152-
Further, the stress pattern on HH words in (64d) shows that when there are two heavy
syllables the initial syllable gets stress. There is no secondary stress on bisyllabic words.
(64) (a) "LL "ka.sha ‘kind of tree (with thorns)’

"ma.ka ‘prickle’
"krE.bE ‘legacy, worn out or useless item’
"ha.fu ‘type of yam’
"nu.ku ‘amputated’
(b) L"H a."kam ‘keloid on ear due to piercing; wild yam’
a."rIndZ ‘orange’
go."yaak ‘a type of banana’
gI."laant ‘to go about idly or frivolously’
a."noNs ‘to announce’
pa."laav ‘to lay idle and prostrate’
bI."fuo ‘before’
su."puoz ‘suppose’
dI."said ‘decide’
a."gEn ‘also, in addition, anymore, any longer’
(c) "HL "uo.pm` ‘to open’
"fai.fa ‘trumpeter fish’
"pai.zn` ‘to poison’
"maa.mI ‘type of fruit’
"aa.da ‘to order’
"baa.bI ‘a type of yam’
"gyan.da ‘gander’
"gan.dZa ‘hemp, marijuana’
"gyan.zI ‘t-shirt’
(d) "HH "dZii.zas ‘Jesus’
"gaa.lIn ‘egret’
"iiv.nIn ~ "iiv.lIn ‘evening’
"duo.duo ‘dough, unbaked bread/cake’
"nuo.tIs ‘to notice, stare at’
"braad.kyaas ‘to broadcast’
The general pattern seen has primary stress on the leftmost heavy syllable. For
words with an initial heavy syllable, stress is thus on the first syllable, e.g. "gaalin ‘egret’,
and for words with a final heavy syllable, stress is on the final syllable, e.g. bI"fuo
-153-
‘before’. In cases where there are no heavy syllables, stress falls on the penultimate
syllable as in "hafu ‘type of yam’.
In addition to this pattern, some bisyllabic words exhibit an alternate stress
pattern. The data in (65) and (66) show bisyllabic words that have a contrast between a
weak-strong (WS) and a strong-weak (SW) stress pattern. The words in (65), have
semantic differences between words in the same word class whereas the words in (66)
have differences between words of different lexical categories.
(65) Contrastive Pattern - Same Word Class
(a) HL kon"trI ‘nickname’ (b) "kontrI ‘place’

faa"da ‘religious leader, male’ "faada ‘parent, male’
sIs"ta ‘religious affiliate, fem.; "sIsta ‘sibling, fem. nurse’
specially ranked nurse’
LL ma"da ‘religious leader, fem.’ "mada ‘parent, fem.’
brE"da ‘religious affiliate, male’ "brEda ‘sibling, male’
The words in (65a) have primary stress on the final light syllable contrary to the expected
pattern. The data presented earlier showed that in the regular pattern bisyllabic LL and
HL words have primary stress on the initial syllable, which is what we see in the words in
(65b). This unexpected weak-strong stress pattern is also seen in the adjectives in (66a)
which contrast with the nouns in (66b) which have the predicted strong-weak pattern.
(66) Contrastive Pattern - Different Classes
Adjective Noun
(a) LL bI"zI ‘fruit - poison antidote’ (b) "bIzI ‘busy’
og"lI ‘ugly’ "oglI ‘fruit’
-154-
We can compare these patterns to that seen in the words in (67). These words are
different from those seen in (65) and (66) in that the stress pattern varies between a weak-
strong (WS) and a strong-weak (SW) pattern even in the same speaker’s speech without
affecting the meaning. Thus only one variant conforms to the expected stress pattern.
(67) Variable Pattern
HL "stuo. rI ~ stuo."rI ‘story’

"pie.pa ~ pie."pa ‘paper’
"pii.pl ~ pii"pl ‘people’
"tie.bl ~ tie."bl ‘table’
LL "pE.pa ~ pE."pa ‘pepper’
The stress patterns in the words in (68) also differ from the regular pattern
observed in (64). Based on the patterns seen in (64b), the forms in (68) would be
expected to have stress on the final heavy syllable, as in go"yaak ‘type of banana’.
However, contrary to our expectations, these words have stress on the initial light
syllable. 5.6% of the bisyllabic words in the dataset have this pattern.
(68) "LH "ba.fan ‘clumsy person’

"va.mIt ‘to vomit’
"u.man ‘woman’
"wI.kId ‘wicked’
As I discuss further below, I treat words with these alternate stress patterns as lexically
marked since their stress patterns differ from the regular pattern.
-155-
5.2.2 Trisyllabic Words
Trisyllabic words with heavy syllables may have primary stress on the initial syllable as
in (69), the penultimate syllable as in (70), or the final syllable as in (71). In these
examples, we see that once again it is the heavy syllable which receives primary stress in
each case, as it is the only heavy syllable in the words. Secondary stress occurs on the
initial syllable of the word as in (71) but there is no secondary stress on the words in (69)
and (70). Some evidence for the absence of stress on word final light syllables can be
gleaned from vowel deletion. In some words, vowels in unstressed final syllables tend to
be deleted whereas those in stressed syllables tend not to be deleted, e.g. nI"glEkful Æ
nIglEkfl` ‘neglectful’; "baksaidin Æ "baksaidn` ‘severe beating’ but *"dZImbIÆln` Æ
dZImbIÆlIn ‘type of tree’.
(69) "HLL "niiz.bE.rI ‘naseberry’

"baa.kI.nI ‘a type of ring game’
"gyaa.sha.nI ‘folk tale bull’
"nuo.ba.dI ‘nobody’
"boN.go.bI ‘type of fish’
"mas.kI.ta ‘mosquito’
"gaN.ga.lu ‘bully, ruffian’
"kom.bo.lo ‘companion, friend’
"san.ta.pi ‘centipede’
"saN.ku.ku ‘to stoop’
(70) L"HL ma."rii.na ‘sleeveless undershirt’

ma."fii.na ‘wretched, paltry, unfortunate’
ha."naa.sI ‘folktale character (spider); general term for spider’
tu."maa.lI ‘crab dish’
ba."faa.nI ‘a child that doesn’t learn to walk on time’
ka."saa.da ‘cassava’
ta."kaa.rI ‘East Indian dish of vegetable and curry’
gI."zaa.da ‘open tart filled with grated sweetened coconut’
a."nan.sI ‘folktale character (spider); general term for spider’
-156-
su."sum.ba ‘type of bean’
ku."kum.ba ‘cucumber’
ba."ram.pa ‘fallow land; part of ginger plant ’
pa."laN.ka ‘fishing device’
pI."tSie.rI ‘type of bird’
ba."rau.tn` ‘without’
pI."tie.ta ‘potato’
tSi."kie.nI ‘trickery’
sa."puo.ta ‘a type of fruit’
a."riel.ya ‘type of flowering plant’
ko."maan.da ‘ variety of cocoa’
(71) ÆLL"H Æba.na."klaa ‘sour milk’

Æba.ra."rap ‘imitation of something falling or colliding’
Æbra.ga."dap ‘fish fritter; flour and saltfish sauce; a sudden stop’
Æpa.ra."kiit ‘parakeet’
Æka.la."ban ‘a box-like trap made of sticks for catching birds’
ÆI.na."mEl ‘enamel’
In the following examples in (72) and (73), there is more than one heavy syllable
in a word and in these cases, the leftmost heavy syllable gets primary stress. Secondary
stress falls on the final heavy syllable in (73) but not on the final light syllable in (72).

(72) "HHL "bak.sai.dn` ‘severe flogging’
"kyaa.pIn.ta ‘carpenter’
(73) "HLÆH "baa.bI.Ækyuu ‘concrete patio for drying food’

"baN.ga.ÆraN ‘useless odds and ends; noise’
"dZIm.bI.ÆlIn ‘type of tree’
"hIg.na.Ærant ‘quick to anger, arrogant, ill-mannered’
"dan.dI.Æmait ‘dynamite’
"kul.tI.Æviet ‘to plant crops’
"skia.fI.Ælas ‘type of skin disease; eczema’
"kruo.ko.Ædail ‘crocodile’
"haas.pI.Ætal ‘hospital’
"maa.Sa.Ælaa ‘public quarrel’
-157-
"waa.rI.Æfai ‘extremely angry, warlike’
The stress pattern in the forms shown in (74) through (75) is also consistent with the
observation made above. In the examples in (74), the penultimate (heavy) syllable is
stressed rather than the final (heavy) syllable since it is the leftmost heavy syllable; there
is no secondary stress in these words.
(74) L"HH tSa."paa.lIn ‘tarpaulin’

ku."muu.dZIn ‘two-faced; deceitful’
gI."laan.tIn ‘to fool around instead of doing duties’
dZu."mie.kan ‘Jamaican’
ma."Sie.tId ‘dilapidated, of ill appearance’
nI."glEk.ful ‘neglectful’
The example in (75) has all heavy syllables and as expected the initial syllable is stressed;
secondary stress falls on the final syllable.
(75) "HHÆH "kaN.kan.Ætar ‘type of top’
Based on trisyllabic words which contain at least one heavy syllable, the stress
pattern can therefore be described as follows: primary stress falls on the leftmost heavy
syllable and secondary stress falls on the second syllable from the primarily stressed
syllable, if there is one. However, word final light syllables are not stressed.
The examples below show that when there are no heavy syllables there are several
possibilities for the stress pattern in trisyllabic words. Primary stress can fall on any of
the syllables in the word and secondary stress may fall on the initial syllable. In the
examples in (76), primary stress falls on the penultimate syllable and there is no
secondary stress.
-158-
(76) L"LL ma."sa.la ‘East Indian dish’
ma."ka.ka ‘type of wood boring beetle’
pa."ka.sa ‘coconut milk and flour sauce’
mo."zE.la ‘type of yam’
ka."ra.to ‘species of Bromelia, wild pine’
tu."ba.ka ‘tobacco’
su."ma.dI ‘somebody’
a."na.ta ‘annatto’
a."nE.da ‘another’
ka."ta.ri ‘East Indian dish’
In the examples in (77), primary stress occurs on the initial syllable and as seen
before there is no secondary stress on the word final light syllable.
(77) "LLL "be.ya.ka ‘type of tree’

"ta.la.wa ‘strong, sturdy’
"mo.gI.kl` ‘kind of mug’
"ma.la.ta ‘mulatto’
"ku.mI.na ‘type of religious dance ceremony’
"ba.la.ho ‘small piper; to flatter’
"ba.chI.la ‘bachelor’
"ko.ta.ku ‘type of basket’
"du.ku.nu ‘boiled pudding’
"ka.ka.tI ‘snobbish, hoity-toity’
"kra.ka.sa ‘type of palm tree’
The third pattern illustrated by the data in (78) places primary stress on the final syllable
of the word and secondary stress on the initial syllable.
(78) ÆLL"LÆkI.tI."bu ‘firefly; meddlesome person’

Æba.ga."bu ‘generic term for various insects, worms’
Æba.la."hu ‘boisterous person’
Æa.na."bo ‘type of tree’
Either the "LLL pattern or the pattern L"LL might be considered as the general pattern as
these are the more common patterns in the data. The ÆLL"L words represent a mere 3.6%
-159-
of the trisyllabic words. The "LLL pattern could be derived by assigning initial syllable
primary stress when there are no heavy syllables. On the other hand, consistent with
Alderete’s (1993) analysis and with observations made above, the L"LL pattern could be
derived by assigning primary stress to the penultimate syllable as in bisyllabic "LL words.
In both cases the penultimate syllable is stressed since there are no heavy syllables. In
fact, a consideration of the stress pattern on quadrasyllabic words, which I discuss further
below, shows that the second analysis is preferable. That is, primary stress generally falls
on the penultimate syllable in both L"LL words and quadrasyllabic ÆLL"LL words as it
does in bisyllabic "LL words, since there are no heavy syllables. This approach thus
lends itself to a unified account of bisyllabic, trisyllabic and quadrasyllabic words with
penultimate primary stress.
Further evidence for this proposal comes from the stress pattern on apparent
exceptional cases. These trisyllabic words show alternate stress patterns which are not
expected based on the stress patterns of other trisyllabic words with heavy syllables. The
words in (79) and (80) are expected to have primary stress on the leftmost heavy syllable
as in "kyaapinta ‘carpenter’ or ni"glEkful ‘neglectful’. Instead, primary stress is assigned
to the second heavy syllable in each word, which is precisely the pattern seen in L"LL
words. That is, primary stress is assigned to the penultimate syllable. As we saw in L"LL
words, there is no secondary stress in these words either.
(79) H"HL fan."daN.gl` ‘elaborate or fussy decoration or ornament’

man."tSEs.ta ‘Manchester’
mam."paa.la ‘common cock; impotent, effeminate man’
njuz."moN.ga ‘type of bug associated with spreading of
news/gossip; person who spreads news/gossip’
-160-
raa."baa.ba ‘coarse person’
(80) H"HH taa."paa.lIn ‘tarpaulin’
ad."van.tIdZ ‘to take advantage of; cheat’
kan."som.San ‘heavy cough associated with consumption disease’
A similar pattern is seen in LLH words shown in (81). These words are expected to have
stress on the only heavy syllable in the word as seen in words like, Ækala"ban ‘box-like
trap’, however, they have penultimate stress like the words in (79) and (80).
(81) L"LH to."ma.tis ‘kind of tomato’

di."la.rans ‘Chicago publisher on occults, banned from Jamaica;
witchcraft’
pro."vi.dZan ‘generic term for yam, cocoa, dasheen, potatoes’
The final set of trisyllabic words shown in (82) are also expected to have primary stress
on the only heavy syllable in the word and secondary stress on the initial syllable as was
seen in other LLH words. As seen, primary stress instead occurs on the initial syllable of
the word and secondary stress occurs on the final heavy syllable. 8.3% of the trisyllabic
words in this dataset have this stress pattern.
(82) "LLÆH "ba.bI.Ælan ‘Rastafarian term for non-believer, white man or

policeman’
"ba.ka.Ælau ‘codfish’
"bo.na.ÆbIs ‘type of bean/broke-pot bean’
"ku.chu.ÆmEnt ‘odd and ends’
"dI.pI.ÆdEns ‘coconut rundown (sauce)’
"a.frI.Ækan ‘African’
"a.ni.Æsou ‘type of bush medicine for colds’
"gya.la.Æwas ‘type of lizard’
"fru.ta.ÆpaN ‘breadfruit’
-161-
To summarize, primary stress typically falls on the leftmost heavy syllable of a
trisyllabic word, e.g. "kyaapinta ‘carpenter’, "dZimbiÆlin ‘type of tree’, ku"muudZin
‘deceitful’, or on the only heavy syllable as in Ækala"ban ‘box-like trap’. When there are
no heavy syllables in the word, primary stress generally falls on the penultimate syllable
as in mo"zEla ‘type of yam’. Secondary stress is expected to fall two syllables away from
the syllable with primary stress, e.g. "baNgaÆraN ‘useless odds and ends’; Ækala"ban ‘box-
like trap’. Of course, when primary stress is on the penultimate syllable, secondary stress
does not occur, as in ma"SietId ‘dilapidated’. In words with an alternate stress pattern,
primary stress falls predominantly on the penultimate syllable whether or not there is a
heavy syllable elsewhere in the word. In a few words, primary stress falls on the initial
syllable with no secondary stress or primary stress occurs on the final syllable and
secondary stress occurs on the initial syllable.
5.2.3 Quadrasyllabic Words
As shown in (83), in quadrasyllabic words when there are no heavy syllables, primary
stress falls on the penultimate syllable as was seen in bisyllabic and trisyllabic words and
secondary stress falls two syllables away from the primarily stressed syllable, i.e. on the
initial syllable.
-162-
(83) ÆLL"LL Æba.ra."ku.ta ‘barracuda’
Æbra.zI."lI.ta ‘species of dye-wood’
Æsa.pa."dI.la ‘sapodilla (type of fruit)’
Ækja.ta."pI.la ‘caterpillar’
Æa.lI."gE.ta ‘alligator’
Æba.nI."klE.va ‘sour/curdled milk’
Æa.lI."ka.sha ‘plant resembling a carrot’
Æa.lI."bo.tn` ‘one who works without receiving payment; worthless
person’
ÆI.nI."kwI.tI ‘evil, obeah, small dumpling’
As with other words seen prior, primary stress falls on the leftmost or only heavy syllable
in the word in (84); secondary stress falls on the initial syllable.
(84) ÆLL"HH Æba.da."rie.San ‘thing or person that is a source of annoyance’

ÆLL"HL Æba.nI."klie.va ‘sour/curdled milk’
Æka.lI."fie.va ‘a type of mullet’
Æko.rI."fie.va ‘to play favorite’
Æpo.ko."mie.nja ‘type of revivalist cult’
Ædu.fI."dai.a ‘dwarfish, stunted person’
ÆtSE.rI."miil.ja ‘a type of fruit similar to a cherry’
Ægja.lI."mEn.ta ‘Galimenta-wood’
Æna.ja."bIN.gI ‘subsection of the Rastafarian cult; meeting of
Rastafarians’
5.2.4 Interim Summary
The data discussed above shows that the leftmost or only heavy syllable in a word
receives primary stress in JC. If there are no heavy syllables, then the typical pattern is
for the penultimate syllable to get primary stress. Secondary stress falls two syllables
away from the syllable with primary stress but not on a word final light syllable. This
generalization is consistent with the data seen in bisyllabic, trisyllabic and quadrasyllabic
-163-
words. Table 5.1 illustrates the parsing of different types of syllables for the general
pattern for stress assignment.
"LL "ma.ka ‘thorn’

L"H pa."laav ‘to lie about idly’
"HL "pai.zn` ‘poison’
"HH "gaa.lIn ‘egret’
L"LL su."ma.dI ‘somebody’
L"HL nI."glEk.ful ‘neglectful’
L"HH gI."laan.tin ‘to fool around’
"HLL "niiz.bE.rI ‘type of fruit’
"HLÆH "dZIm.bI.ÆlIn ‘type of tree’
"HHL "bak.sai.dn` ‘severe beating’
"HHÆH "kaN.kan.Ætar ‘type of top’
ÆLL"HH Æba.da."rie.San ‘thing or person that is a source of annoyance’
ÆLL"HL Æpo.ko."mienya type of revivalist cult’
Table 5.1: General Stress Pattern in Monomorphemic Words of Different Syllable Types
We also saw that there are also alternate patterns. In bisyllabic words, primary stress
may fall on the final light syllable even the initial syllable is heavy. In trisyllabic words
primary stress may fall on the penultimate syllable with no secondary stress. In addition,
a few words have final primary stress and initial secondary stress or initial primary stress
and no secondary stress. There are no alternate stress patterns in the quadrasyllabic
words in this dataset. I will show that these alternate patterns may be accounted for in a
principled manner within an Optimality Theoretic framework.
-164-
5.2.5 Compound Words
The data in this section illustrate the stress pattern on nominal compound words. I include
these primarily as a source of comparison with reduplicated words since they are another
productive word formation process in JC as in many creole languages. As I show in
Chapter 6, this is a worthwhile exercise since the stress pattern on distributive reduplication
reflects the pattern seen in compounds words, i.e. the regular pattern. Intensive
reduplications on the other hand have a different pitch accent pattern than compounds and
distributives. I argue that this reflects differences in the number of primary stresses
permitted in intensive reduplications as opposed to compounds and distributive
reduplications. In addition, in other English-based creoles like Krio and Guyanese,
compounding has been shown to share similarities with reduplication processes. Nylander
(1999) (Krio) and Devonish (2003) (Guyanese), for example, argue that reduplication
processes exhibit similar prosodic properties to compounding.
Bisyllabic compounds are shown in (85), trisyllabic compounds in (86),
quadrasyllabic compounds in (87) and longer ones in (88). The generalizations
regarding the stress pattern on monomorphemic words can be extended to the stress
patterns on each part of the compound word. The stress pattern on compound words is
thus directly related to the stress pattern of the input word19. In the bisyllabic compounds
19
Lawton (1963) reported a different pattern for some compound words in addition to the patterns seen here.
According to Lawton, the ‘tonal’ specification of some input words is altered in the compound. The transcriptions
below are the same as those used in Lawton’s work, ( ¤ ) represents high-level tone;
( Ÿ ) high-falling tone and ( ! ) low-level tone.
bu@laŸ ‘type of cake; insult’ bu¤la¤kie⁄k ‘type of cake’
gua!baŸ ‘guava’ gua!ba¤boŸd ‘guava bud’
pa⁄snŸ ‘parson’ pa⁄sn¤brau⁄n ‘type of orange’
From the point of view of stress, all these words have identical stress patterns to the trisyllabic compound words
comprised of a bisyllabic word + monosyllabic word. That is, they have a SWS stress pattern as in Æbula"kiek.
-165-
in (85) for example, each input word has primary stress and the resulting compound has
two stresses. Notice however that the primary stress on the initial word is realized as a
secondary stress in the compound word, e.g. "bod ‘bird’ + "mout ‘mouth’ --->
Æbod."mout ‘pecked by a bird’.
(85) ÆH"H ÆbIg."man "bIg ‘big’+ "man ‘man’ ‘adult male human’
Æbod."mout "bod ‘bird’ + "mout ‘mouth’ ‘pecked by a bird’
Æblak."mInt "blak ‘black’ + "mInt ‘mint’ ‘variety of mint’
Æblak."krab "blak ‘black’ + "krab ‘crab’ ‘variety of crab’
Ækrab."tatS "krab ‘crab’ + "tatS ‘thatch’ ‘variety of thatch’
Ækrab."yaaz "krab ‘crab’ + "yaaz ‘yaws’ ‘type of foot disease’
Æhaas."baat "haas ‘horse + "baat ‘bath’ ‘weed used to scrub horses’
Æbiif."biin "biif ‘beef’ + "biin ‘bean’ ‘brown beans eaten with beef’
In trisyllabic compounds in (86) the location of the stresses also reflects the pattern on the
input words, e.g. "baa.bl` ‘barble’ + "dov ‘dove’ --->Æbaa.bl`."dov ‘type of dove’. As is
expected, based on the pattern in bisyllabic compounds, the primary stress of the input
word "baa.bl` ‘barble’ is realized as a secondary stress in the compound word.
(86) ÆLL"H ÆbE.lI."hat "bE.lI ‘belly’ + "hat ‘hurt’ ‘stomach ache’

ÆbE.lI."gad "bE.lI ‘belly’ + "gad ‘god’ ‘glutton’
Ædo.tI "gjal "do.tI ‘dirty’ + "gjal ‘girl’ ‘type of laundry soap’
Æku.lI "plom "ku.lI ‘Indian’ + "plom ‘plum’ ‘type of plum’
ÆbIg. "sm`a.dI "bIg ‘big’ + su."ma.dI ‘adult person’20
‘somebody’
ÆHL"H Æbaa.bl`."dov "baa.bl` ‘barble’ + "dov ‘dove’ ‘type of dove’
Æbaa.bl`."dZak "baa.bl` ‘barble’ + "dZak ‘Jack’ ‘large male donkey’
ÆdaN.kI "ruop "daN.kI ‘donkey’ + "ruop ‘rope’ ‘tobacco leaves rolled into a
rope’
ÆdaN.kI."wiid "daN.kI ‘donkey’ + "wiid ‘weed’ ‘plant that looks like thyme’
ÆH"HL ÆbIg "fIN.ga "bIg ‘big’ + "fIN.ga ‘finger’ ‘thumb’
20
The initial unstressed syllable of su."ma.dI ‘somebody’ is often deleted as it is in the formation of the compound in
this example.
-166-
Quadrasyllabic compounds shown in (87) generally have primary stress on the
penultimate syllable and secondary stress on the second syllable preceding the
penultimate syllable. In these cases the input words are two bisyllabic words each with
primary stress on the initial syllable, e.g. "ba.mI ‘bammy’ + "prE.sa presser
ÆÆ ba.mI "prE.sa ‘tool for flattening type of cassava bread’; "baa.buun ‘baboon’ +
"kuo.ko ‘cocoa’ Æ Æbaa.buun."kuo.ko ‘type of cocoa’. As before, the primary stress on
the first input word is realized as secondary stress in the compound. Notice that though
the stress pattern on ka.Æsaa.va "puon ‘cassava pudding’ is different, it is consistent with
the generalizations made above. That is, the observed stress pattern is reflective of the
stress pattern on the input words, ka."saa.va ‘cassava and "puon ‘pudding’.
(87) ÆLL"LH Æba.kra "kya.bIdZ "ba.kra ‘Caucasian’ + "kya.bIdZ ‘cabbage’ ‘type of cabbage’
ÆbE.lI."u.man "bE.lI ‘belly’ + "u.man ‘woman’ ‘pregnant woman’
ÆLL"LL Æba.mI "prE.sa "ba.mI ‘bammy’ + "prE.sa ‘presser’ ‘tool for flattening
bammies’
Æku.ku."ma.ka "ku.ku ‘coco’ + "ma.ka ‘thorn’ ‘heavy stick with
thorns’
Æku.lI "pE.pa "ku.lI ‘Indian’ + "pE.pa ‘pepper’ ‘type of pepper’
ÆHH"HL Æbaa.buun."kuo.ko "baa.buun ‘baboon’ + "kuo.ko ‘cocoa’ ‘type of cocoa’
Æbaas.tad."sii.da "baas.tad ‘bastard’ + "sii.da ‘cedar’ ‘wild cedar’
ÆHH"LH Æbaas.tad."o.kro "baas.tad ‘bastard’ + "o.kro ‘okra ‘wild okra’
ÆLL"HL ÆlI.kl`."fIN.ga "lI.kl` ‘little’ + "fIN.ga ‘finger’ ‘little finger’
Ædo.tI."boN.gl` "do.tI ‘dirty’ + "boN.gl` ‘bundle’ ‘untidy person’
Æti.ti."bai.ta "ti.ti ‘breast’ + "bai.ta ‘biter’ ‘insect’
ÆHL"LH ÆtSai.nI "kja.bIdZ "tSai.nI ‘Chinese’ + "kya.bIdZ ‘cabbage’ ‘type of cabbage’
ÆHL"HL ÆtSuo.tSo."aa.ba "tSuo.tSo ‘squash’+ "aa.ba ‘arbor’ ‘trellis for type of
squash plant’
ÆHL"HL Æaa.tSI."bom.bo "aa.tSI ‘saltfish fritters’ + "bombo ‘obscene name for
‘expletive’ saltfish fritters’
ÆLH"LH Ækro.kIn."lI.zad "kro.kIn ‘croaking’ + "lI.zad ‘lizard’ ‘type of lizard’
ÆHH"LL Ækro.kIn."dro.ma "kro.kIn ‘croaking’ + "dro.ma ‘drummer’ ‘type of fish’
LÆHL"H ka.Æsaa.va "puon ka.Æsaa.va ‘cassava’ + "puon ‘pudding’ ‘cassava pudding’
-167-
The stress pattern on longer compounds shown in (88) similarly reflects the stress
patterns of the input words; the primary stress on the first input word is again realized as
secondary stress.
(88) ÆLLL"HL ÆbI.ta.ka."saa.va "bI.ta ‘bitter + ka.Æsaa.va ‘cassava’ ‘type of cassava’

LÆHL"HL ka.Æsaa.va "flau.wa ka.Æsaa.va ‘cassava’ + "flau.wa ‘flour’ ‘cassava flour’
To summarize, the stress pattern in compounds depends on the stress pattern of the input
words. Primary stress on the first member of the compound is realized as secondary stress
in the compound word. In addition, the rightmost member of all the compounds is the
strongest, i.e. has main stress. For example, Æhaas"baat ‘weed used to scrub horses’;
ÆbElI"gad ‘glutton’; Ækuku"maka ‘heavy stick with thorns’.
5.3 Summary
As we have seen, in monomorphemic words primary stress generally falls on the leftmost
heavy syllable. If there is no heavy syllable, primary stress falls on the penultimate
syllable of a word. Secondary stress falls two syllables away from the syllable with
primary stress, if there is such a syllable except in the case of word final light syllables
which do not bear stress. In the majority of cases, stress falls on the initial or penultimate
syllable of a word and falls only rarely on the final syllable. Where secondary stress falls
on a word final syllable it is typically a heavy syllable. These patterns are reflective of
the general pattern for 90% of the words in the database.
-168-
In the section below, I account for the general and alternate stress patterns within an
Optimality Theoretic framework.
5.4 Predicting Stress Assignment in JC - An Optimality Theoretic Account
5.4.1 General Pattern
In this section, I present an analysis of stress formulated within an Optimality Theoretic
(OT) framework. As is outlined in Chapter 1, this is a constraint-based approach to
phonological well-formedness. Based on the observations on stress assignment discussed
in the preceding sections, the following general statements can be made regarding the
predominant pattern for stress assignment in monomorphemic words in JC:
(i) primary stress falls on the leftmost heavy syllable otherwise;

(ii) primary stress falls on the penultimate syllable
(iii) secondary stress falls on the syllable two syllable away from that with
primary stress, if there is one, but not on a word final light syllable.
In deriving the stress assignment on compounds, I assume that each monomorphemic
input word receives stress as outlined in (i) through (iii). However, the additional
statements in (iv) and (v) are also required for compounds:
(iv) the rightmost member of the compound has primary stress;

(v) primary stress on the input word of the leftmost member is realized as
secondary stress.
The first type of data I discuss are non-compound words with a sequence of two
light syllables, e.g. "kaSa ‘type of thorny tree’. Words like these were shown to have
primary stress on the initial syllable. I refer to the constraints given in (89) in accounting
for the stress pattern on these words.
-169-
(89) Parse Syllable (PARSESYL): Syllables are parsed into feet (Prince and Smolensky,
1993)
Foot Type (FTTYPE[TROCH]): Feet are trochaic (ibid, 1993)
The constraint PARSESYL requires syllables within a given word to be grouped into feet.
The FTTYPE constraint requires prosodic feet to be a sequence of LL or HL syllables or a
single H or L syllable which are all types of trochaic feet. Prince (1990), proposes the
following trochaic rhythmic harmony scale in (90):
(90) (LL, H) › HL › L
Informally, this suggests that trochaic feet comprised of a sequence of LL syllables or a
single H syllable are preferable to those comprised of a sequence of HL syllables which
are in turn preferable to feet with a single L syllable21.
Tableau 5.1 shows the evaluation of the bisyllabic "LL word "kaSa ‘type of thorny
tree’. In this and other tableaux, foot constituency is indicated by parentheses.
21
Kager (1999:172) and Prince and Smolensky (1993) suggest that LH and HH feet with initial stress are also trochaic
feet but they violate the constraint Weight-to- Stress-Principle (WSP), which requires heavy syllables to be stressed.
-170-
FTTYPE (TROCH) PARSESYL
) a. ("kaSa)
b. ("ka)Sa *!
c. (ka"Sa) *!
d. kaSa *!*
Tableau 5.1: ["kaSa ] ‘type of thorny tree’
Candidate (a) is the winning candidate since both of its syllables are parsed into feet
forming a trochaic (LL) foot. This candidate satisfies both of the constraints considered.
One of the syllables in candidate (b) is not parsed into a foot causing the candidate to
incur a fatal violation of PARSESYL, which rules it out. Both syllables in candidate (c) are
parsed into feet but main stress is on the second syllable. This does not form a trochaic
foot, since it is right-headed (LW"LS) instead of left-headed ("LSLW). This candidate incurs
a fatal violation of FTTYPE and is ruled out as well. None of the syllables in candidate (d)
are parsed into feet, consequently it fails on the PARSESYL constraint.
The constraint in (91) prohibits a sequence of two stresses in a word. This is
instrumental in ruling out the illformed candidate (d) in Tableau 5.2, which has secondary
stress on the final syllable.
(91) * CLASH No stressed syllables are adjacent (Kager, 1999)
-171-
* CLASH FTTYPE (TROCH) PARSESYL
) a. ("kaSa)
b. ("ka)Sa *!
c. (ka"Sa) *!
d. ("ka)(ÆSa) *!
Tableau 5.2: ["kaSa ] ‘type of thorny tree’
This predicts that there is no secondary stress in bisyllabic words since this would
produce a sequence of two stresses. I will motivate the ranking of * CLASH, FTTYPE, and
PARSESYL further below.
The selection of a trisyllabic L"LL word highlights the importance of having binary
feet. The constraint which requires binary feet is given in (92).
(92) Foot Binarity (FTBIN): Feet are binary under a moraic or syllabic analysis
(Prince and Smolensky, 1993; MCarthy and Prince, 1986, 1993)
In Tableau 5.3, candidate (b) is selected as optimal. Candidate (a) has one
unparsed syllable as does candidate (b) but it incurs a fatal violation of FTTYPE which
rules it out. Candidate (c) is ruled out by FTBIN and candidate (d) is ruled out by * CLASH.
Candidate (e) has a single violation as does the winning candidate (b). The crucial
ranking of FTBIN above PARSESYL rules out the suboptimal candidate (e).
-172-
* CLASH FTBIN PARSESYL FTTYPE (TROCH)
a.(mo"zE)la * *!
) b. mo ("zEla) *
c.("mo)(zEÆla) *! *
d. ("mo)(ÆzEla) *! *
e. ("mozE)(Æla) *!
Tableau 5.3: [mo"zEla] ‘type of yam’
This tableau correctly predicts the general stress pattern on trisyllabic LLL words with
main stress on the penultimate syllable. I discuss alternate patterns in section 5.4.2.
With the constraints thus far we are also able to predict stress assignment in
quadrasyllabic ÆLL"LL words as is illustrated in Tableau 5.4.
*CLASH FTBIN PARSESYL FTTYPE (TROCH)

) a. (ÆalI)("gEta)
b. (ÆalI) ("gE)ta *! *
c. (ÆalI) gE ("ta) *! *
d. (Æa)(lI"gE)ta *! * *
e. (aÆlI) ("gEta) *! *
Tableau 5.4: [ÆalI"gEta] ‘alligator’
As shown here, the optimal candidate is (a), which has no violations of the constraints
considered. Candidates (b) and (c) lose since they fatally violate FTBIN and PARSESYL.
Candidate (d) is the worst; it violates three of the constraints considered. Candidate (e) is
ruled out since it has a sequence of two adjacent stresses which causes a fatal violation of
*CLASH.
-173-
Notice however, that so far no distinction has been made regarding which syllable
has primary stress and which one has secondary stress; the constraints simply predict
which syllables are prominent. Consider candidates (a) and (b) in Tableau 5.5 below.
Both candidates satisfy all of the constraints considered having stresses on the initial and
on the penultimate syllable. As was noted above, in words with all light syllables,
primary stress falls on the penultimate syllable and secondary stress falls on alternating
syllables starting from the syllable with main stress as in candidate (a).
*CLASH FTBIN PARSESYL FTTYPE

(TROCH)
) a. (ÆalI)("gEta)
) b. ("alI) (ÆgEta)
c. (ÆalI) ("gE)ta *! *
d. (ÆalI) gE ("ta) *! *
In order to differentiate between the primarily and secondarily stressed syllables, two
additional constraints are needed. The specific constraints are given in (93). The
constraint ALIGNHD refers specifically to the foot with main stress, i.e. the prosodic head.
It places the foot with main stress at the right edge of the word. ALIGNFT places a foot at
the right edge of a prosodic word.
-174-
(93) Align (Head, Right; Prosodic word, Right) (ALIGNHD(R)): The right edge of a
prosodic head foot is aligned to the right edge of a prosodic word. (Hammond,
1999; Kager, 1999)
Align (Foot, Right; Prosodic word, Right) (ALIGNFT(R)): The right edge of a
prosodic foot is aligned to the right edge of a prosodic word. (Kager, ibid)
In Tableau 5.6, as in subsequent ones, prosodic word constituency is indicated by square
brackets [ ]. As shown, ALIGNHD is crucial in ruling out the illformed candidate (b),
which incurs two fatal violations of the constraint. Violations of ALIGN are evaluated
based on the number of syllables between any foot or head foot and the edge of the word,
in this case the right edge. Candidate (a) has one foot that is two syllables removed from
the right edge of the word and so it incurs two violations of ALIGNFT. Its head foot is
flush with the right edge of the word completely satisfying ALIGNHD. Candidate (b) also
has two violations of ALIGNFT. However, its head foot is two syllables away from the
right edge of the word (word initially) so it incurs two fatal violations of ALIGNHD which
rules it out. Both candidates (c) and (d) are ruled out by FTBIN. This tableau correctly
predicts primary stress on the penultimate syllable and secondary stress on the initial
syllable. At this point there is no crucial ranking among ALIGNHD(R), ALIGNFT(R) and
FTTYPE.
-175-
*CLASH FTBIN PARSE FTTYPE ALIGN ALIGN
SYL (TROCH) HD(R) FT(R)
) a. [(ÆalI)("gEta)] **
b.[("alI)(ÆgEta)] **! **
c.[(ÆalI) ("gE)ta] *! * * ***
d.[(ÆalI)gE ("ta)] *! * **
The next examples which I discuss are those for which syllable weight is important to
stress assignment. Recall that primary stress is assigned to the leftmost or only heavy
syllable in a word. For example, a bisyllabic L"H word such as a"kam ‘keloid’ has stress
on the leftmost heavy syllable which in this case is the final syllable. Tableau 5.7
illustrates that the constraints established thus far are not sufficient to predict stress
assignment in these words. In this example, the wrong surface form is predicted; this is
indicated by the symbol 0.
*CLASH FTBIN PARSE FTTYPE ALIGN ALIGN

SYL (TROCH) HD(R) FT(R)
a. [ a("kam)] *!
b. [ (a"kam)] *!
c. [(Æa)("kam)] *! * *
0 d. [("akam)]
e. [ akam] *!*
f. [("a) kam] *! * *
Tableau 5.7: [a"kam ] ‘keloid’
Primary stress should be assigned to the heavy syllable, however, none of the constraints
thus far refer to syllable weight. The constraint Weight-to-Stress (WSP), given in (94),
-176-
requires heavy syllables to be stressed and also reflects the tendency for heavy syllables
to be stressed in JC. As discussed earlier, for syllables to be heavy, coda consonants,
long vowels and diphthongs are assumed to contribute to syllable weight22.
(94) WEIGHT-TO-STRESS (WSP): Heavy syllables are stressed (Hammond, 1999).
Tableau 5.8 reevaluates the candidates in Tableau 5.7 with the inclusion of WSP.
*CLASH FTBIN WSP FTTYPE PARSE ALIGN ALIGN

(TROCH) SYL HD(R) FT(R)
) a. [ a("kam)] *
b. [(a"kam)] *!
c. [(Æa)("kam)] *! * *
d. [("akam)] *!
e. [ akam] *! **
f. [("a) kam] *! * * *
Tableau 5.8: [a"kam ] ‘keloid’
With the inclusion of WSP, candidate (a) is correctly predicted as the optimal candidate.
This establishes that WSP is crucially ranked above PARSESYL or else we would predict
the illformed candidate (d) as the surface form since it has only one violation of the
constraints considered as does candidate (a). This shows that it is more important for a
heavy syllable to be stressed than for all syllables to be parsed into feet. In addition,
although candidate (b) is phonetically identical to candidate (a), it has a right-headed
(iambic) foot and is therefore dispreferred. This also establishes a crucial ranking of
22
I assume that the constraint WEIGHT-BY-POSITION (WP) is also at work here (Kager, 1999). WP captures the
observation that coda consonants are moraic. Thus, they contribute to syllable weight.
-177-
FTTYPE above PARSESYL or else candidate (b) would be incorrectly predicted as the
surface form. Notice also that consistent with these observations, bisyllabic LH words
are not predicted to have secondary stress since this would cause a violation of *CLASH.
The inclusion of WSP also correctly predicts that "HL words like "gandZa ‘marijuana’
will have stress on the first syllable since that syllable is heavy. Tableau 5.9 shows that
we are able to predict the correct surface form with the constraints introduced thus far. I
have excluded the Align constraints from further evaluation of bisyllabic words since
these words are predicted not to have secondary stress.
*CLASH FTBIN WSP FTTYPE PARSE

(TROCH) SYL
a. [ ("gan)dZa ] *!
) b. [ ("gandZa)]
c. [ (gan"dZa)] *!
d. [ (Ægan)("dZa)] *! *
e. [ gan("dZa) ] *! * *
Tableau 5.9: ["gandZa] ‘marijuana’
Candidate (b) wins since it does not have any violations of the constraints considered.
Candidate (a) fails on the PARSESYL constraint and candidate (c) fails since it has an
iambic foot. Candidate (d) has two adjacent stressed syllables which causes a fatal
violation of *CLASH. Candidate (e) has a stressed light syllable which results in a fatal
violation of FTBIN; it violates other constraints as well. All of these latter candidates are
thus ruled out.
-178-
These constraints trivially account for trisyllabic "HLL words like "gangalu ‘ruffian,
bully’; L"HL words, e.g. su"sumba ‘type of wild bean’ and LL"H, e.g. Ækala"ban ‘box-like
trap’. This is illustrated in Tableau 5.10 through Tableau 5.12. In Tableau 5.10,
candidate (d) is selected as optimal since it has only minimal violations of the lower
ranked constraints and crucially does not violate any of the higher ranked constraints, as
is the case with the losing candidates.

a. [("gaNga)(Ælu)] *! * *
b. [("gaN)galu] **! ** **
c. [("gaN)(Ægalu)] *! ** **
) d.[("gaNga) lu] * * *
e. [(gaN"ga)lu ] *! * * * *
f. [(ÆgaNga)("lu)] *! *
Tableau 5.10: ["gaNgalu] ‘ruffian, bully’
This correctly predicts that there is no secondary stress on a word final light syllable. In
this case candidate (d) is selected as optimal. Candidate (a) closely competes with
candidate (d), however, it has a fatal violation of FTBIN which eliminates it. Candidate (c)
has another stress adjacent to the stress on the initial syllable which causes it to violate
the highly ranked *CLASH constraint. Candidate (e) fails on several of the constraints,
FTTYPE, WSP and PARSESYL and is ruled out. Candidate (f), which primary stress on the
word final syllable and secondary stress on the initial syllable, is ruled out by FTBIN. The
preference of candidate (d) over candidate (a) in this tableau shows that it is not as
-179-
important to have exhaustive footing as it is to have binary feet. This correctly predicts
that there is no secondary stress on word final light syllables. Notice also that the
prohibition on footed word final light syllables is also extended to candidate (f) which as
noted above fails on FTBIN.
Tableau 5.11 illustrates the evaluation of a L"HL word, su"sumba ‘wild bean’, which
has primary stress on the penultimate syllable.
*CLASH FT WSP FTTYPE PARSE ALIGN ALIGN

BIN (TROCH) SYL HD(R) FT(R)
a. [(su"sum)ba] *! * * *
) b.[ su("sumba)] *
c. [su("sum)ba] **! * *
d. [("susum)ba] *! * * *
e. [("susum)(Æba)] *! * * *
Tableau 5.11: [su"sumba] ‘wild bean’
The correct output form is candidate (b) which only has violations of the lowly ranked
PARSESYL constraint. The closest competing candidate, (a), which also has penultimate
stress fails on FTTYPE since it creates an iambic foot (LWHS) instead of a trochaic foot.
Given the constraints posited thus far, two types of stresses are predicted to occur
in trisyllabic LLH words such as Ækala"ban ‘box like trap’, shown in Tableau 5.12. The
word has primary stress on the final syllable and secondary stress on the initial syllable.
-180-
a. [ka (la"ban)] *! *
b. [kala ("ban)] *!*
) c. [(Ækala)("ban)] *
d. [(kaÆla)("ban)] *! * *
e.[("kala)(Æban)] *! *
Tableau 5.12: [Ækala"ban] ‘box like trap’
As seen, candidate (c) wins since it has a single violation of the lowly ranked ALIGNFT
constraint. Candidate (a) is ruled out by FTTYPE. Candidate (b) which does not have
secondary stress fatally violates PARSESYL and is ruled out. Candidate (d) fatally violates
*clash and candidate (e), the closest competing candidate, which has primary stress on
the initial syllable and secondary stress on the final syllable, fails on ALIGNHD. This
establishes a crucial ranking of ALIGNHD above ALIGNFT. Thus far there is no crucial
ranking between WSP and FTTYPE.
Tableau 5.13 illustrates that the stress pattern on ÆLL"HL quadrasyllabic words is
also predictable given the constraint ranking established thus far.

) a. [(ÆgjalI)("mEnta)] **
b. [(ÆgjalI) ("mEn)ta] *! * **, *
c.[(Ægja)(lI"mEn)ta ] *! * * * ***, *
d.[(ÆgjalI)mEn("ta)] *! * * **
e. [(gjaÆlI) ("mEnta)] *! * **
f. [("gjalI)(ÆmEnta)] **! **
Tableau 5.13: [ÆgjalI"mEnta] ‘type of wood’
-181-
Candidate (a) wins since it only violates the lowly ranked ALIGNFT constraint. The
closest competing candidate, (f), is eliminated due to fatal violations of the ALIGNHD
constraint.
All the examples I have discussed thus far have just one heavy syllable. The
following examples illustrate how stress is predicted in words with more than one heavy
syllable. Recall that these words have primary stress on the leftmost heavy syllable and
secondary stress two syllables away from the main stress. In the bisyllabic word in
Tableau 5.14 there are two heavy syllables, so we need to be able to predict which one
will be assigned stress.

a. [("gaa)lin] * *! * *
) b. [("gaalin)] *
c.[("gaa)(Ælin)] *! * *
d.[gaa("lin)] * *!
Tableau 5.14: ["gaalin] ‘egret’
This tableau shows that the constraint ranking thus far is able to predict the stress pattern
on words with more than one heavy syllable. The correct surface form, candidate (b),
with stress on the initial syllable is selected as optimal since its only violation is of the
WSP constraint. The closest competing candidate in (d), which also has a violation of WSP,
is ruled out by PARSESYL.
-182-
Accounting for trisyllabic words with more than one heavy syllable is a bit more
involved. In particular, the observation that the leftmost heavy syllable is stressed
conflicts with the need to have the head foot at the right edge of the word. This is
illustrated in Tableau 5.15 with the word "baNgaÆraN ‘useless odds and ends’. As seen
here, the correct surface form, candidate (a) with primary stress on the initial syllable is
ruled out by ALIGNHD(R), which requires that the head foot be at the right edge of the
word. The favored candidate is the illformed candidate (b) with initial secondary stress
and final primary stress.

a. [("baNga)(ÆraN)] *! *
0 b.[(ÆbaNga)("raN)] *
c.[("baN)ga(ÆraN)] *! ** **
Tableau 5.15: ["baNgaÆraN] ‘useless odds and ends’
To resolve this issue, I propose the constraint ALIGNL in (95), which specifically requires
the bimoraic head foot to be leftmost in the word. As formulated, ALIGNL thus forces the
heavy syllable with main stress in Tableau 5.15 to be in word initial position, where ϕ =
prosodic head.
(95) Align (Heavy Head (ϕµµ), Left; Prosodic word, Left) (ALIGNHD(ϕµµ) (L)):
The left edge of a heavy prosodic head is aligned to the left edge of a
prosodic word.
-183-
Tableau 5.16 demonstrates the function of this constraint with the reevaluation of
the candidates in Tableau 5.15.
*CLASH FT WSP FTTYPE PARSE ALIGNHD ALIGN ALIGN

BIN (TROCH) SYL (ϕµµ) (L) HD(R) FT(R)
) a. [("baNga)(ÆraN)] * *
b.[(ÆbaNga)("raN)] *!* *
c.[("baN)ga(ÆraN)] *! ** **
Tableau 5.16: ["baNgaÆraN] ‘useless odds and ends’
In this case candidate (a) is correctly selected as optimal. The initial heavy syllable has
primary stress and the final syllable has secondary stress. The illformed candidate (b)
fatally violates ALIGNHD(L) and is ruled out. Candidate (c) satisfies ALIGNHD(L) but is
eliminated by PARSESYL. This shows that in trisyllabic words with more than one heavy
syllable, having the heavy foot with main stress leftmost in the word takes precedence
over aligning prosodic feet with the right edge of the word.
Tableaux 5.17 and 5.18 show that these constraints trivially account for trisyllabic
L"HH and "HHÆH words. In Tableau 5.17, candidate (a) is chosen as optimal as it has the
least number of constraint violations. The crucial ranking of *CLASH above ALIGNHD(L)
rules out candidate (c). Likewise, the ranking of FTBIN above ALIGNHD(L) rules out
candidate (e). Candidate (b) is ruled out by PARSEYL, candidate (d) by ALIGNHD(L) and
candidate (f) by FTTYPE.
-184-
ALIGNHD
(ϕµµ) (L)
(TROCH)
*CLASH
FTTYPE
ALIGN
ALIGN
PARSE
HD(R)
FTBIN
FT(R)
WSP
SYL
) a.[ma ("SietId)] * * *
b.[ma("Sie)tId] * * **! * *
c.[(ma"Sie)(ÆtId)] *! * * *
d.[(ÆmaSie)("tId)] **! * *
e.[("ma)Sie(ÆtId)] *! * * ** **
f. [(ma"Sie) tId * * *! * * *
Tableau 5.17: [ma"SietId] ‘dilapidated, of ill appearance’
The example in Tableau 5.18 shows the selection of a trisyllabic "HHÆH word. In
this case primary stress is predicted to occur on the initial heavy syllable and secondary
stress on the final syllable.

ALIGNHD
(ϕµµ) (L)
(TROCH)
*CLASH
FTTYPE
ALIGN
ALIGN
PARSE
HD(R)
FTBIN
FT(R)
WSP
SYL
) a. ("kaNkan)(Ætar) * * *
b. ("kaN)kan(Ætar) * *! ** **
c. (ÆkaNkan)("tar) *!* * *
d. ("kaN)(Ækan)tar *! * * ** **,*
e. (ÆkaN)kan("tar) *! * * **
Tableau 5.18: ["kaNkanÆtar] ‘type of top’
Candidate (a) is correctly chosen as the optimal candidate; it satisfies ALIGNHD(L) at the
expense of a violation of ALIGNHD(R). The closest competing candidate, (c), satisfies
ALIGNHD(R) but incurs two fatal violations of ALIGNHD(L) and is eliminated. Candidate
-185-
(b) also satisfies ALIGNHD(L) but fails on PARSESYL and (d) is ruled out since it incurs a
fatal violation of *CLASH. In this case a violation of ALIGNHD(R) is less critical than a
violation of the highly ranked ALIGNHD(L) constraint.
The constraint hierarchy thus far is as shown in the diagram in (96). The commas
between the constraints represent non-crucial rankings and the connecting lines represent
the crucial dominance relationships among the constraints.
(96) Constraint hierarchy for JC General Stress Pattern
*CLASH, FTBIN
ALIGNHD (ϕµµ)(L), WSP, FTTYPE (TROCH)
PARSESYL, ALIGNHD(R)
ALIGNFT(R)
With this constraint hierarchy we are able to account for the general stress pattern
observed in the data. Informally, this means that words with wellformed stress patterns
cannot have adjacent stresses (*CLASH) and must have binary feet (FTBIN). In addition,
the stressed syllables must be heavy, must be trochaic feet and be initial in the word
(WSP, FTTYPE, ALIGNHD(L)). Of less importance is the need for syllables to be parsed into
feet and for the main stress to be at the right edge of the word. These requirements are
together more important than the general need to have prosodic feet at the right edge of
the word (ALIGNFT(R)).
-186-
5.4.2 Alternate Patterns
I now focus the discussion on several words with stress patterns different from those
predicted for the general pattern described above. With regard to bisyllabic words with
contrastive stress, only those forms that have a different stress pattern from the general
pattern are problematic. Among these are H"L and L"L words with second syllable stress,
e.g. faa"da ‘male religious leader’; ma"da ‘female religious leader’. All the other forms
can be accounted for given the constraint hierarchy in (96). For example, initial syllable
stress is correctly predicted to occur in bisyllabic "HL and "LL words, e.g. "faada ‘father’,
"mada ‘mother’. For words with the alternate pattern, the regular SW stress pattern is
reversed to a WS pattern. Some additional evidence for this proposal comes from the
variation in stress pattern observed in some words, e.g. "piepa~pie"pa ‘paper’.
Interestingly, none of the words with a WS stress pattern have the canonical iambic
pattern, i.e. L"H. I assume that words with a WS stress pattern are lexically marked and
are assigned stress in the input.
There are also words which pattern as if they are comprised of only light
syllables. These are "LH, H"HL, H"HH and L"LH words, which all have stress on the
penultimate syllable instead of the predicted leftmost heavy syllable. I assume that in
these cases, the mora of coda consonants, which would normally contribute to syllable
weight, is unparsed. Thus, coda consonants in these words are assumed to have the
structure shown in (97)23.
23
The reader can compare this structure to the one in (11b) shown in Chapter 1, in which the mora of the coda
consonant is parsed.
-187-
(97) σ
µ µ
V C (Prince and Smolensky 1993: 70)
Following Prince and Smolensky (1993), the final consonants in these words can be
analysed as directly adjoining the syllable. As such, the potentially contributing mora is
not ‘seen’ by the syllable and cannot contribute to syllable weight, i.e. it is extraprosodic.
Given their structure, these words necessarily violate the weight-by-position principle
given in footnote 22 (repeated in (98)).
(98) WEIGHT-BY-POSITION(WP): Coda consonants are moraic. (Kager, 1999)
In the regular cases, the WP constraint is assumed to be undominated since heavy
syllables are invariably stressed. As I show just below, to account for cases with the
alternate stress pattern, we can posit a constraint hierarchy similar to that in (96), but
crucially with the WP constraint ranked low.
Tableau 5.19 shows how the stress pattern on "LH words is predicted. Recall that
these pattern like bisyllabic "LL words, which have primary stress on the penultimate
syllable. In this and subsequent tableaux, the form with the nonmoraic coda consonant is
marked with ( @ ).
-188-
(TROCH)
*CLASH
ALIGN
ALIGN
PARSE
FTYPE
HD(R)
FTBIN
FT(R)
WSP
SYL
WP
) a.[("bafan @ )] *
b.[("bafan)] *!
c. [ba ("fan)] *!
d. [("ba) (Æfan)] *! * * *
Tableau 5.19: ["bafan ] ‘clumsy person’
Candidate (a) with the extraprosodic coda consonant is selected as optimal since its only
violation is of the lowly ranked WP constraint. The competing candidate in (b) has a
moraic coda consonant so WP is satisfied. However, this heavy syllable is not stressed,
which results in a fatal violation of the highly ranked WSP constraint. Candidate (c),
which has the regular pattern, is ruled out by PARSESYL.
Tableau 5.20 illustrates the selection of a H"HL word. Candidate (a) is selected as
optimal since it has the least number of violations of the constraints considered.
Candidate (d) has initial primary stress, which is the regular stress pattern. It satisfies WP
but is eliminated by WSP since it has a heavy syllable that is not stressed.
-189-
ALIGNHD
(ϕµµ) (L)
(TROCH)
*CLASH
ALIGN
ALIGN
PARSE
FTYPE
HD(R)
FTBIN
FT(R)
WSP
SYL
WP
) a.[fan @ ("daN @gl`)] * **
b.[fan ("daNgl`)] *! * *
c.[(Æfan)("daN) gl`] *! * * * **,*
d.[("fandaN)gl`] *! * * *
e.[("fan)daN (Ægl`)] *! * * ** **
Tableau 5.20:[fan"daNgl` ] ‘elaborate ornament’
As illustrated in Tableau 5.21, penultimate syllable stress is also correctly
predicted for trisyllabic H"HH words. Candidate (a) is selected as optimal since it has the
least number of violations of the constraints considered. Candidate (d) reflects the regular
stress pattern but is eliminated by WSP since it has a stressless heavy syllable.
ALIGNHD
(ϕµµ) (L)
(TROCH)
*CLASH
ALIGN
ALIGN
PARSE
FTYPE
HD(R)
FTBIN
FT(R)
WSP
SYL
WP
) a. ad @ ("van @tidZ @) * ***
b. ad("vantidZ ) *! ** *
c. ad("van) tidZ *! ** ** * *
d. ("ad)van(ÆtidZ) *! * ** **
e. (Æad)van("tidZ) **! * * **
Tableau 5.21:[ad"vantidZ ] ‘to take advantage of, cheat’
-190-
Tableau 5.22 shows the evaluation of L"LH words. As shown, the predicted winner is
candidate (a) and the regular form, candidate (d) is eliminated by ALIGNHD(ϕµµ) (L) .
ALIGNHD
(ϕµµ) (L)
(TROCH)
*CLASH
ALIGN
ALIGN
PARSE
FTYPE
HD(R)
FTBIN
FT(R)
WSP
SYL
WP
) a. pro("vIdZan @) *
b. pro("vIdZan ) *! *
c. (pro"vi) dZan *! * * * *
d. (ÆprovI)("dZan) **!
Tableau 5.22: [pro"vIdZan] ‘generic term for yam, potatoes etc.’
I now discuss "LLÆH and "LLL words, which have the prosodic head at the left edge
of the word instead of at the right edge as in regular cases. To account for the stress
pattern on these words, I assume the constraints in (99), which require the head foot to be
word initial. These constraints are similar to the align constraints posited in the account
of words with a regular stress pattern (see 93). Further, I assume that these constraints
are relevant only for the evaluation of lexically marked words such as "LLÆH and "LLL
words.
(99) Align (Head, Left; Prosodic word, Right) (ALIGNHD(L)): The left edge of a
prosodic head foot is aligned to the left edge of a prosodic word.
Align (Foot, Left; Prosodic word, Left) (ALIGNFT(L)): The left edge of a
prosodic foot is aligned to the left edge of a prosodic word.

-191-
Tableau 5.23 shows the selection of the trisyllabic "LLL word "kumina ‘type of
religious dance ceremony’. This word has initial primary stress and no secondary stress.
As demonstrated, the crucial ranking of ALIGNHD(L) and ALIGNHD(L) above
ALIGNHD(R)and ALIGNHD(R) allows candidate (a) to be selected as the winner. Candidate
(c), which has the regular stress pattern in eliminated by ALIGNHD(L).
ALIGNHD
(ϕµµ) (L)
(TROCH)
*CLASH
ALIGN
ALIGN
ALIGN
ALIGN
PARSE
FTYPE
HD(R)
HD(L)
FTBIN
FT(R)
FT(L)
WSP
SYL
) a.("kumi)na * * *
b. ("ku) mina *! **
c. ku ("mina) * *! *
d. ("kumi)(Æna) *! * *
Tableau 5.23: ["kumina] ‘type of religious dance ceremony’
Tableau 5.24 shows that the "LLÆH words are accounted for in a similar way. In these
cases, primary stress is assigned to the initial light syllable in spite of the fact that there is
a heavy syllable which is predicted to get primary stress.
-192-
ALIGNHD
(ϕµµ) (L)
(TROCH)
*CLASH
ALIGN
ALIGN
ALIGN
ALIGN
PARSE
FTYPE
HD(R)
HD(L)
FTBIN
FT(R)
FT(L)
WSP
SYL
a.("babi) lan *! * * *
) b.("babi) (Ælan) * *
c. (Æbabi) ("lan) **! ** **
Tableau 5.24: ["babiÆlan] ‘rastafarian term for non-believer or policeman’
The predicted surface form is candidate (b), which has violations only of the lowly
ranked ALIGN(R) constraints. Candidate (c) which has the stress pattern predicted for
regular cases, is ruled out by ALIGNHD(ϕµµ) (L). Thus, except for the ranking ALIGNHD(L)
and ALIGNHD(L) above ALIGNHD(R)and ALIGNHD(R), we can predict the correct surface
form for these words with a constraint ranking similar to that assumed for words with a
regular stress pattern.
The final type of exceptional words I discuss here are ÆLL"L words with final
syllable stress. This pattern is counter to the predicted pattern of penultimate syllable
stress. I assume that in these cases, the words are lexically marked for primary stress on
the final syllable. To account for these, the constraint FTBIN, which penalizes degenerate
feet, has to be ranked low since a light syllable is in the metrically strong position
bearing, main stress (Tableau 5.25).
-193-
FTTYPE PARSESYL FTBIN ALIGNHD(R)
(TROCH)
) a.[(Æbaga)("bu)] *
b.[(Æba)(ga"bu)] *! *
c. [ba("gabu)] *!
Tableau 5.25: [Æbaga"bu] ‘insects, worms’
Candidate (b) is ruled out since it incurs a fatal violation of FTTYPE. Candidate (c) which
reflects the regular stress pattern is ruled out in this case due to the crucial ranking of
PARSESYL above FTBIN. Candidate (a) is thus selected as optimal.
5.4.3 Section Summary
In the preceding sections, I provided phonological evidence supporting the
characterization of the JC word prosodic system in terms of stress. JC forms trochaic feet
from right-to-left and assigns primary stress to the heavy syllable nearest the left edge of
the word. When there are no heavy syllables, the penultimate syllable is stressed.
Secondary stress is predicted to occur on the final heavy syllable of a trisyllabic word
syllable but not on a word final light syllable. In quadrasyllabic words, secondary stress
is predicted to fall on the initial syllable and primary stress on the penultimate syllable.
Secondary is not predicted in bisyllabic words due to the pressure to avoid adjacent
stresses.
We also saw that alternate stress patterns could be accounted for by the reranking
of constraints as well as constraints specific to lexically marked words. For example, the
ranking of PARSESYL above FTBIN forced primary stress assignment on a word final light
-194-
syllable in ÆLL"L words instead of on the penultimate syllable. This reflects the fact that
the patterns are lexically marked and need to be accounted for separately.
5.5 Chapter Summary
The data presented in this chapter provided evidence for JC as a language with a weight-
sensitive stress system. In the following chapters, I focus on the contrastive uses of stress
in JC and pay particular attention to cases where stress distinguishes between contrasting
pairs of reduplicated words. The aim is to interpret the prosodic differences in JC
reduplicated words in the context of the prosodic system. Since JC has a stressed-based
system, this means that the discussion of the prosodic differences between reduplicated
words should be expressed in terms of differences in stress. As discussed above, it is
possible that the phonological properties of compounds are mirrored in reduplicated
words. The main questions are, how is stress manifested in reduplicated words?; what
kind of stress differences are there between intensive and distributive reduplication?; are
there any similarities between the stress pattern on reduplicated words and compound
words?; and, are there any similarities between the stress pattern on reduplicated words
and monomorphemic words?
-195-
CHAPTER 6
PHONETIC STUDY OF THE PROSODY OF REDUPLICATED WORDS
6.1 Introduction
This chapter presents the results of a phonetic analysis of word-level prosody in Jamaican
Creole (JC). The goal is to establish a basis for the proposed stress differences between
distributive and intensive reduplicated words in terms of the metrical structure and the
associated intonational phonology. In other words, the goal is to determine the
organization of strong and weak syllables in these words and how they are associated with
given intonations. The hypothesis being tested is that intensive and distributive
reduplicated words have different prosodic structures and will therefore be differentiated in
their alignment with the F0 contour. The discussion is based on an instrumental analysis of
recorded utterances of 7 native speakers including myself. The target words occur in
different intonations including broad-focus statements, yes-no questions and citation form
in two different prosodic positions, non-final (prenuclear) and final (nuclear) in a statement
or question, as was done for unreduplicated words in chapter 4. The selection of utterance
types analysed in this chapter is in keeping with the chapter’s descriptive style. The results
-196-
of the analyses in this chapter provide a basis for future quantitative analyses of specific
aspects of the prosodic system. Earlier, I argued that the alignment of the F0 pattern is
contrastive in JC and can account for differences in prosody when segmental similarity
between words would otherwise camouflage the contrast. In this case as well, ambiguity at
the segmental level is resolved by reference to higher units of prosody.
In chapters 4 and 5 I reviewed phonetic as well as phonological evidence for
characterising JC as a stress-accent language. We saw that JC has a prominence driven
stress system in which heavy syllables are stressed. In cases where there are no heavy
syllables, stress falls on the penultimate syllable of the word. In addition, we saw that
there are no lexically contrastive pitch accent shapes that contribute a fixed shape to the F0
contour. As demonstrated earlier, lexical contrasts are signaled instead by differences in
the alignment of the F0 contour with the word. I illustrate here that the stress contrast
between distributive and intensive reduplications is also a difference in the alignment of the
F0 contour with the word. I argue that distributive reduplications have a single accentable
syllable and thus one pitch accent, while intensive reduplications have two accentable
syllables and thus two pitch accents. Phonetically, the F0 contour of both words is a high-
low melody which is aligned differently in each word. I analyse this high-low melody as a
sequence of a HL pitch accent, the low part of which is anchored to the prosodic head of
the word. I also show that the type of stress pattern seen in distributive reduplication is also
seen in other words in the language, i.e. monomorphemic words and compounds.
-197-
The analysis is based on several recordings of reduplicated words with intensive
and distributive meanings24 as shown in Table 6.1.
 ‘scattered yellow’

 ‘very yellow’
 ‘sticky all over’
 ‘very sticky’
 ‘black all over’
 ‘very black’
 ‘sort of long; longish’
 ‘very long’
 ‘low here and there; sort of low’
 ‘very low’
 ‘sort of near’
 ‘very near’
 ‘scattered green’
 ‘very green’
Table 6.1: Elicited Reduplicated Words used in Analysis
The chapter is organized as follows. In section 6.2, I examine the F0 pattern on
reduplicated words and make comparisons with non reduplicated words, compounds and
narrow focus phrases. In section 6.3, I present a summary of the findings and discuss the
implications for the phonological analysis of reduplication in Chapter 7. Section 6.4 is the
chapter summary.
24
As discussed in chapter 2, the category distributive in JC includes both the semantic notion of attributive and
distributive whereas these are treated separately in other CECs.
-198-
6.2 F0 Pattern of Reduplicated Words
In this section I examine the prosodic properties of intensive and distributive reduplicated
words. Earlier, I alluded to a stress contrast between these segmentally identical
reduplicated words. In this section, I focus on defining the contrasting metrical structures
associated with the different intonational contours for each type of reduplication. As was
done for the non-reduplicated words and compounds, we need to find out: (a) if the F0
contours for the two meanings differ in the different prosodic positions and under varying
intonation contours; (b) how the F0 contours for the two meanings differ in the different
prosodic positions and in the different intonation contours.
6.2.1. Monosyllabic Base Words
The reduplicated words I examined here are formed from monosyllabic words. In some
cases the words have a distributive meaning and in other cases they have an intensive
meaning, e.g.  ‘very green/green all over’. I look at each type in turn.
6.2.1.1. Distributive Reduplication
Recall that in distributive reduplication, the meaning of the reduplicated word has the
sense of ‘scattered; all over the place; here and there; occasionally’. For example, 
‘green’ reduplicates as  ‘green all over’. In the example below (Figure 6.1),
there is a fall from a high pitch on the first syllable onto the stressed second syllable of
the word.
-199-
greenDISis3 SG
250
200
150
100
4000
2000
0
tones H+L*
phones g  i n g  i n
words griingriin
0.2 0.4 0.6 0.8

Time (seconds)
of the reduplicated word ‘scattered green’ produced in citation form by speaker
SG. This reduplicated word has primary stress on the second syllable.
Based on our observations regarding unreduplicated words in the previous sections, we
know this is characteristic of syllables which bear primary stress. I therefore analyse this
pattern as a H+L* pitch accent to capture the fact that there is a low tonal target on the
stressed second syllable of the word. Notice that this stress pattern is similar to that seen
in bisyllabic compound words such as  ‘type of bean’. Figure 6.2 shows the word
in final and non-final position in a statement. In both cases we can still observe the
H+L* pitch accent.
-200-
greenDISdf1 SG greenDISdnf1 SG
300
250
200
150
100
4000
2000
0
phones g  i n g  i n g  i ng  i n
words im want wan we griingriin im want wan griingriin wan an som yam
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2

panels) of the reduplicated word ‘scattered green’ produced in statement
intonation in final position in the sentence, im want wan we griingriin ‘He wants one that
is scattered green’ (left graph) and in non-final position, in the sentence, im want wan
griingriin wan an som yam ‘He wants a scattered green one and some yams’ (right
graph).
The next set of examples in Figure 6.3 shows the word ‘scattered green’ in final
and non-final position in a question.
-201-
greenDISynf1 SG greenDISynnf2 SG
400
300
200
100
4000
2000
0
phones g  i n g  i n g  i ng  i n
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2

panels) of the reduplicated word ‘scattered green’ produced in question
intonation in final position in the sentence, im want wan we griingriin ‘Does he want one
that is scattered green?’ (left graph) and in non-final position in the question, im want
wan griingriin wan an som yam ‘Does he want a scattered green one and some yams?’
(right graph).
This figure shows that the fall on the stressed syllable is still present just before the
phrase-final rise to the H% of the question intonation, i.e. a fall-rise pattern. Notice that
this is similar to the pattern we saw for compounds and quadrasyllabic words where the
H+L* pitch accent was also maintained in all the prosodic contexts considered and also
had a fall-rise F0 pattern in question intonation. I now compare these patterns to those for
the intensive reduplication.
-202-
6.2.1.2. Intensive Reduplication
In intensive reduplication, the meaning of the reduplicated word has more intensity or
emphasis than in the unreduplicated word; for example,  ‘green’ reduplicates as
 ‘very green’. In the example of intensive reduplication shown in Figure 6.4 we
can see that there is a kind of ‘hat-like’ F0 pattern. That is, there is an initial rise to a high
on the initial stressed syllable, which is maintained into the second syllable after which
the pitch falls to a low on the stressed second syllable. Notice that both the low and the
preceding high are both located within the second syllable. I analyse this pattern as a
H+L* pitch accent to capture the idea that there is a low tonal target on the stressed
second syllable of the word. In addition, I analyse the rise on the initial syllable as a
prenuclear rise to a H* pitch accent. (I refer to the accent as prenuclear since it appears
before the main pitch accent (H+L*) in the word.) This shows that both stressed
syllables in the intensive reduplicated word are accented.
-203-
greenINis2 RP
250
200
150
100
4000
2000
0
tones H* H+L* L%
words griingriin
0.2 0.4 0.6 0.8

Time (seconds)
of the reduplicated word ‘very green’ produced in citation form by speaker RP
As we can see in Figure 6.5, this is not the only strategy for producing a contrast with
distributives. Speakers can also produce the word with a rise to a high F0 peak on the
first syllable and a fall onto the second syllable. Notice however, that in both Figure 6.4
and Figure 6.5 there is a late fall in pitch on the second syllable. This is crucially
different from the distributive reduplications where the fall in pitch on the second syllable
is much earlier.
-204-
greenINis1 SG
250
200
150
100
4000
2000
0
tones H+L*
words griingriin
0.2 0.4 0.6 0.8

Time (seconds)
of the reduplicated word ‘very green’ produced in citation form by speaker SG.
Both of these patterns can also be observed in the speech of a single speaker.
Figure 6.6 shows the monosyllabic base reduplicated word  ‘very low’ produced in
citation form by speaker HF. The pitch is slightly higher in the graph on the left than in
the one on the right but the location of the pitch peak and fall on the second syllable
remains the same for both productions.
-205-
lowINis1 HF lowINis2 HF
200
150
100
50
4000
2000
0
tones H* H+L* H* H+L*
phones l uo l uo l uo l uo
words luoluo luoluo
0.2 0.4 0.6 0.2 0.4 0.6 0.8

Figure 6.6: Fundamental frequency contours (upper panel) and spectrograms (lower
panel) of  produced in citation form with an intensive meaning ‘very low’.
The fact that both the high and low of the H+L* accent are located within the
second syllable in these reduplicated words suggests that it is the F0 fall that is important
for locating prominence in these words. This may explain why some speakers produce
slight variations in the F0 pattern associated with the first syllable.
The next set of examples show the intensive word in statement intonation in both
final and non-final position.
-206-
greenINdf2 SG greenINdnf2 SG
200
100
4000
2000
0
tones H* H+L* L% H* H* L%
phones g  i n g  i n g  i n g  i n
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2

panels) of the reduplicated word ‘very green’ produced in statement intonation
in final position in the sentence, im want wan we griingriin ‘He wants one that is very
green’ (left graph) and in non-final position, in the sentence, im want wan griingriin wan
an som yam ‘He wants a very green one and some yams’ (right graph).
In this case we see that the H+L* is maintained in both final and non-final position.
Notice though that in the non-final context, the second H+L* accent is realized on the
following accented phrase som yam ‘some yams’.
The examples below in Figure 6.8 show the intensive word in question intonation
in both final and non-final position. In this case we see that there is a high pitch
throughout the word to the phrase final high of the H% of the question intonation. The
fall in pitch onto the stressed syllable is not seen. Rather, we have a high-rise F0 pattern.
-207-
greenINynf1 SG greenINynnf1 SG
400
300
200
100
4000
2000
0
tones H* H* H% H* H* H%
phones g  i n g  i n g  i ng i n
words im want wan we griingrrin im want wan griingrrin wan an som yam
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2

panels) of the reduplicated word ‘very green’ produced in question intonation in
final position in the sentence, im want wan we griingriin ‘Does he want one that is very
green?’ (left graph) and in non-final position, in the sentence, im want wan griingriin wan
an som yam ‘Does he want a very green one and some yams?’ (right graph).
6.2.1.3. Summary
According to the analysis proposed here, the alignment of the H+L* pitch accent in the
word accounts for the difference between monosyllabic base intensive and distributive
reduplicated words that occur in nuclear position Figure 6.9 compares a monosyllabic
base reduplicated word produced with a distributive meaning (left graph) to one with an
intensive meaning (right graph) by speaker EW.
-208-
nierDISis EW nierINis EW
250
200
150
100
4000
2000
0
tones H+L* H* H+L*
phones n !  n !  n !  n ! 
words niernier niernier
0.2 0.4 0.6 0.2 0.4 0.6 0.8

panels) of  produced with a distributive meaning ‘not quite near’ (left graph) and
with an intensive meaning ‘very near’ (right graph) by EW.
As noted in the previous sections, there is a fall in pitch onto the second syllable
of monosyllabic based words in both the distributive and the intensive renditions. We
also noted that in the distributive meaning (left graph) the high portion of the F0 peak is
located in the preceding syllable whereas in intensive meaning, both the fall and the high
are located within the same syllable. That is, there is an early fall in the word with a
distributive reading and a late fall in the word with an intensive meaning. Words with the
intensive meaning have two pitch peaks (indicated by the arrows) while words with the
distributive meaning have only one. I interpret this as evidence of a prenuclear rise onto
the initial stressed syllable of words with an intensive meaning which is crucially absent
in words with a distributive meaning. In Table 6.2 I summarize the pitch accent types
associated with the stressed syllables in monosyllabic based intensive and distributive
reduplication in the different prosodic contexts.
-209-
Prosodic Condition
Final Non-final
Distributive
Citation H+L*
Statement H+L* (L%) H+L* (L%)
Yes-No question H+L* (H%) H+L* (H%)
Intensive
Citation H* H+L*
Statement H* H+L* H+H* (L%)
Yes-No question H* H*(H%) H* H* (H%)
Table 6.2: Summary of Pitch Accents in Intensive and Distributive Monosyllabic based
Words
6.2.2. Bisyllabic Based Words
These reduplicated words are formed from bisyllabic bases. In some cases the words
have a distributive meaning and in other cases they have an intensive meaning, for
example,  ‘gummy’ reduplicates as  ‘very gummy/ gummy all over’. In
the next few sections, I examine the F0 contour associated with words of both types. I
should also note that although I made reference to the syllable in describing the patterns
in monosyllabic based words in section 6.2.1, I will hereafter refer to the patterns on both
bisyllabic base words and monosyllabic base words in terms of the foot. In the examples
discussed in this section, the observed F0 patterns are identical to those seen in
monosyllabic base words, except that they are extended over two syllables. This suggests
that the foot rather than the syllable is the relevant prosodic category.
-210-
6.2.2.1. Distributive Reduplication
In the example in Figure 6.10, we can see a pattern that is identical to that seen in
monosyllabic base words but is extended over four syllables instead of two. There is a
late fall in pitch onto the second foot such that the F0 peak is on the first foot of the word.
As before, I analyse this as a H+L* pitch accent.
yeloDISis2 SG
250
200
150
100
4000
2000
0
tones H+L* L%
phones j ! l o j ! l o
words yeloyelo
0.4 0.6 0.8 1

Time (seconds)
panel) for the reduplicated word  ‘scattered yellow’ (distributive meaning)
produced citation.
This pattern is similar to that seen for quadrasyllabic compound words e.g. 
‘pregnant woman’ (section 4.7.3) and quadrasyllabic monomorphemic words, e.g.
 ‘type of wood (section 4.7.2). Figure 6.11 shows a pitch fall onto the foot
-211-
with main stress when the word occurs in final and in non-final position. Thus, in both
cases the H+L* pitch accent is still maintained.
yeloDISdf SG yeloDISdnf SG
250
200
150
100
4000
2000
0
phones j ! l o j ! l o j ! l o j ! l o
words im waant wan we yeloyelo im want wan yeloyelo mango
0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
panels) for reduplicated word  ‘scattered yellow’ (distributive meaning) produced
in statement intonation in final position in the sentence, im want wan we yeloyelo ‘He
wants one that is scattered yellow’ (left graph) and in non-final position in the sentence,
im want wan yeloyelo mango ‘He wants a scattered yellow mango’ (right graph).
-212-
Figure 6.12 below shows the word  ‘scattered yellow’ in final and non-
final position in a question. As we saw in the previous examples, the H+L* pitch accent
is still realized on the word in both contexts.
yeloDISynf SG yeloDISynnf1 SG
300
250
200
150
100
4000
2000
0
tones H+L* H% H*+L H%
words yu want wan we yeloyelo yu waant wan yeloyelo mango
1.2 1.6 2 0.4 0.8 1.2

panels) for the reduplicated word  ‘scattered yellow’ (distributive meaning)
produced in phrase final position in yes-no question intonation in the phrase yu waant
wan we yeloyelo ‘Do you want one that is scattered yellow?’ and in non final phrase
position in the question yu waant wan yeloyelo mango ‘Do you want a scattered yellow
mango?’
-213-
6.2.2.2. Intensive Reduplication
In this section, I examine reduplicated words with an intensive meaning. As shown in
Figure 6.13, there is a prenuclear F0 rise on the first foot and a late fall in pitch on the
second foot which forms the ‘hat-like’ F0 pattern we saw in monosyllabic base words. As
above, this is analysed as a prenuclear H* pitch accent on the first foot and a H+L* pitch
accent on the second foot of the word.
yeloINis SG
250
200
150
100
4000
2000
0
tones H* H+L*
phones j ! l o j ! l o
words yeloyelo
0.2 0.4 0.6 0.8

Time (seconds)
panel) for the reduplicated word  ‘very yellow’ (intensive meaning) produced in
citation.
The next set of examples in Figure 6.14 shows the reduplicated word 
‘very yellow’ in final and non-final position in a statement. The prenuclear rise onto the
first foot and the late fall on the second foot is seen only when the word occurs in final
-214-
position (left graph). When the word occurs in non-final position, a high F0 is maintained
throughout the word and the H+L* is realized on the following accented word, mango.
yeloINdf SG yeloINdnf1 SG
250
200
150
100
4000
2000
0
words di wan we im want yeloyelo im want wan yeloyelo mango
0.4 0.8 1.2 1.6 1.2 1.6 2 2.4

panels) for the reduplicated word  ‘very yellow’ (intensive meaning) produced in
statement intonation final position in the sentence Di wan we im want yeloyelo ‘The one
he wants is very yellow’ (left graph) and in non-final position in the sentence im want
wan yeloyelo mango ‘He wants a very yellow mango’ (right graph).
In Figure 6.15, the reduplicated words are in final and non-final position in a question. In
both cases, there is a high rise to the terminal contour of the question intonation. In this
case we have two H* pitch accents, on each foot.
-215-
yeloINynf1 SG yeloINynnf3 SG
300
200
100
4000
2000
0
words yu want wan we yeloyelo yu want wan yeloyelo mango
0.8 1.2 1.6 2 0.8 1.2 1.6 2

panels) for the reduplicated word  ‘very yellow’ (intensive meaning) produced in
question intonation in the sentence final position in the sentence yu waant wan we
yeloyelo ‘Do you want one that is very yellow?’ (left graph) and in non final position in
the sentence yu waant wan yeloyelo mango ‘Do you want a very yellow mango?’ (right
graph).
6.2.2.3. F0 Pattern on Narrow Focus Phrases
The data presented in this section are not an exhaustive analysis of phrase types in JC but
rather are meant to demonstrate that intensive reduplication patterns with phrases that have
two accented syllables, such as the narrow focus phrase types examined here. Figure 6.16
shows the phrase  ‘yellow mango’ in citation form. In this and other examples,
the property item ‘yellow’ has narrow focus so that both yellow and mango are accented25.
As was seen for intensive reduplicated words, there is a prenuclear rise on the first foot, a
H* pitch accent, and a late fall in pitch on the second foot, a H+L* pitch accent.
25
In the comparable phrase produced with broad focus, there is only a single H+L* pitch accent on the second foot in
the phrase.
-216-
yelomangofis SG
250
200
150
100
4000
2000
0
tones H* H+L*
phones j ! l o m a  g o
words yelo mango
0.4 0.6 0.8 1

Time (seconds)
panel) for the phrase  ‘yellow mango’ produced in citation.
Figure 6.17 shows the phrase produced in sentence intonation in final and non-final
position. Notice that when the phrase is in non-final position, a high F0 is generally
maintained throughout and the H+L* on the second foot is realized on the following
phrase som yam ‘some yams’.
-217-
yelomangofdf SG yelomangofdnf SG
250
200
150
100
4000
2000
0
phones j ! l o m a  g o j ! l o m a  go
words im want wan yelo mango im want wan yelo mango an som yam
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 2

panels) for the phrase  ‘yellow mango’ produced in statement intonation
in the sentence im want wan yelo mango ‘He wants a yellow mango’ (left graph) and in
non-final position in the sentence im want wan yelo mango an som yam ‘He wants a
yellow mango and some yams’ (right graph).
Figure 6.18 shows the phrase in final and non-final position in question intonation. In
this case, there is a high rise to the high terminal contour as we saw in intensive
reduplication and there are two H* pitch accents.
-218-
yelomangofynf SG yelomangofynnf2 SG
350
300
250
200
150
100
4000
2000
0
phones j ! l o m a  g o j ! l o m a  g o
words yu want wan yelo mango yu want wan yelo mango
0.8 1.2 1.6 0.4 0.8 1.2 1.6

panels) for the phrase  ‘yellow mango’ produced in question intonation
in the sentence yu want wan yelo mango ‘Do you want a yellow mango?’ (left graph) and
in non-final position in the sentence yu want wan yelo mango an som yam ‘Do you want
a yellow mango and some yams?’ (right graph).
6.2.2.4. Summary
As discussed above, I characterize the prosodic difference between bisyllabic base
intensive and distributive reduplicated words as a difference in the presence or absence of a
prenuclear rise on the initial foot of the word as well as a difference in the location of the F0
peak of the H+L* pitch accent in each word. Just as in monosyllabic base words, in
bisyllabic base words there is a single F0 peak (indicated by the arrow) in the words with
the distributive meaning (left graph) whereas the word with the intensive meaning (right
graph) has two peaks (Figure 6.19).
-219-
yeloDISis1 HF yeloINis2 HF
150
125
100
75
4000
2000
0
tones H+L*
H* H+L*
words yeloyelo yeloyelo
0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8 1

panels) of  produced in citation with a distributive meaning ‘yellow all over -
scattered’(left graph) and an intensive meaning ‘very yellow’ (right graph) by HF.
Thus, in both bisyllabic-based and monosyllabic-based intensive reduplications,
there is a late fall onto the foot that bears main stress (the rightmost foot) such that the F0
peak and fall are within this syllable. This is similar to the pattern seen for 
‘mother’ which also had the F0 peak and fall on the stressed initial syllable (section
4.5.2). In addition, there is a rise in pitch to a F0 peak on the initial syllable of the word.
Given these observations, I propose that intensives have two accentable syllables, the first
of which is a prenuclear rise to a H* tonal accent and the second is the target of a H+L*
pitch accent. This analysis is supported by the narrow focus data in section 6.2.2.3 which
show patterns similar to intensive reduplication. Unlike intensives, in distributive words
there is an early fall onto the foot that bears main stress (the rightmost foot) and
-220-
consequently the F0 peak is on the preceding foot. Distributives also have a H+L* but
there is only a single accentable syllable onto which the pitch falls consistently.
Table 6.3 summarizes the pitch accent associated with the stressed syllables in
bisyllabic based intensive and distributive reduplication in the different prosodic
contexts.
Prosodic Condition
Final Non-final
Distributive
Citation H+L*
Statement H+L* (L%) H+L* (L%)
Yes-No question H+L* (H%) H+L* (H%)
Intensive
Citation H* H+L*
Statement H* H+L* H* H* (L%)
Yes-No question H* H*(H%) H* H* (H%)
Table 6.3: Summary of Pitch Accents in Intensive and Distributive Bisyllabic based
Words
The strategies for producing the contrast between intensive and distributive reduplications
discussed here are generally consistent with the data produced by other speakers.
Representative examples of both types of words in the different prosodic contexts produced
by different speakers are given in Appendix B.
-221-
6.3. Discussion
As discussed in chapter 4, the alignment of the tonal melody accounts for the contrast in
words in JC. The reduplication data discussed here can also be accounted for with
reference to the alignment of the F0 contour with the word. In distributive words, the low
of the HL melody corresponds to an early fall in pitch on the foot that had main stress
(head foot) whereas in intensive words there was a late fall in pitch on the foot with main
stress. In addition, we saw evidence for a prenuclear accent on the first foot which is
absent in words with a distributive meaning.
Since intensive reduplications have two accentable syllables, when they occur in
citation form and finally in a statement (nuclear position) the nuclear H+L* accent is
realized on the foot with main stress and the prenuclear H* accent is realized on the first
foot. Distributives reduplications, on the other hand, have only one accentable syllable so
when they occur in comparable contexts they take the nuclear H+L* pitch accent on the
foot with main stress. The reason for this is directly related to differences in the prosodic
structure of the words as is illustrated in the diagrams in (100) through (102) below. That
is, distributives pattern like a single prosodic word while intensives pattern like two
prosodic words. These diagrams show metrical grid representations of distributive and
intensive reduplication. For comparison, I have also included the metrical grids for a
quadrasyllabic monomorphemic word, a quadrasyllabic compound and a narrow focus
phrase.
-222-
(100) x PPh x PPh
x Prwd accented x Prwd accented
Distributive x x Foot stressed x x Foot stressed
x x Syllable x x x x Syllable
‘green all over’ ‘yellow all over’
x PPh
x Prwd accented
Quadrasyllabic x x Foot stressed
Word x x x x Syllable
‘alligator’
x PPh
x Prwd accented
Quadrasyllabic x x Foot stressed
Compound x x x x Syllable
‘pregnant woman’
As shown in these metrical grids, distributive reduplications, like quadrasyllabic
monomorphemic words and compounds, have only one accentable syllable. This is
because there is only one syllable which is in a metrically stronger position than other
syllables in the words at the level of the prosodic word. For clarity, I have marked this in
a box. When these patterns are compared to that for the intensive reduplications in (101),
we see that both distributives and intensives have identical stress patterns at the level of
the foot. However, intensives have two metrically strong syllables at the level of the
prosodic word and are therefore eligible to have two pitch accents. This is identical to the
pattern seen in the narrow focus phrase in (102).
-223-
(101) x PPh x PPh
x x Prwd accented x x Prwd accented
Intensive x x Foot stressed x x Foot stressed
x x Syllable x x x x Syllable
‘very green’ ‘very yellow’
(102) x PPh
Adjectival Phrase x x Prwd accented
x x Foot stressed
x x xx Syllable
‘yellow mango’
This distinction in the potential to bear two versus one pitch accent is reminiscent
of the distinction in English between compound words like paperclip or blackbird which
are both accented on the first foot, and sequences of two words like paper tiger and black
bird which have two accents, one on each foot. The former has been referred to in the
literature as a compound and the latter as a phrasal compound (Bauer, 1983; Nespor and
Vogel, 1986; Vogel and Raimy, 2002).
Further, we saw that the pitch accent on distributives is not affected by the
question intonation as it was for the intensive forms. The H+L* pitch accent is
maintained in all contexts for the distributives, while it appears only in citation form and
statement intonation for intensives. More specifically, there is a distinction between a
rise from a high in the intensive words and a fall-rise pattern in the distributive words.
Since both intensive and distributive reduplicated words have a pitch fall onto the
foot with main stress (head foot of the word), in this respect, they both pattern like
compound words in that they have right-headed prominences. However, distributive
reduplications are more similar to compound words than are intensive reduplications. In
compound words, as in distributive reduplications, there is only one accented syllable. In

-224-
neither case do we see the prenuclear accent in these words that is observable in words
with an intensive meaning.
This chapter presented phonetic evidence for the metrical stress contrast between
intensive and distributive reduplications. We saw that while both types of reduplication
bear a similar propensity for stress they are differentiated by their potential to bear
multiple pitch accents. In the following chapter, I present a formal phonological analysis
of reduplication. The difference in the formal analysis of the words allows us to capture
the proposed differences in their prosodic structures.
-225-
CHAPTER 7
AN OPTIMALITY THEORETIC ACCOUNT OF JC REDUPLICATION
7.1 Introduction
This chapter presents a formal phonological analysis of four different patterns of
reduplication in JC: i.e. iterative, characteristic, distributive and intensive. Particular
attention is paid to the prosodic properties of distributive and intensive (simple input)
reduplications. These reduplications yield output forms that are segmentally identical,
yet, as shown in the previous chapter, have different prosodic properties. I argue that
distributive reduplication patterns like a phonological word with respect to its prosodic
properties, while intensive reduplication does not. I show that this distinction permits
different distributions of pitch accents which then accounts for the contrast between the
two types of reduplication. Further, I show that both full and partial reduplication are
prosodically constrained; when the prosodic requirements are not met, no reduplication is
observed. For example, in iterative reduplication the reduplicant copies the base
completely (the base is identical to the input) e.g. dZukop ‘to pierce’ Æ dZukopdZukop ‘to
-226-
pierce repeatedly’. However, if the base is too large, there is no reduplication, e.g.
*tSambaoptSambaop.
The chapter is outlined as follows. In the first three sections I present data and
analyses for processes of reduplication in which a prosodic contrast is not involved:
section 7.2 focuses on iterative reduplication; section 07.3 on characteristic type
reduplication and section 7.4 on intensive (complex input) reduplication. In section 7.5, I
present data and analyses of distributive and intensive (simple input) reduplication which
were shown to be prosodically contrastive.
7.2 Iterative Reduplication
As discussed in Chapter 2, the input to iterative reduplication may be a complex stem
comprised of a root and an affix, e.g. maak + op ‘to mark on (deliberately)’ or a simple
stem comprised of a bare root, e.g. tiif ‘to steal’. In reduplication, the entire stem is
copied, as shown in (103). As can be seen in (103c), e.g. njakaop ‘to cut crudely’, when
the stem is longer than two syllables, no reduplication occurs.
(103) Root +suffix = Stem Reduplicated Form

(a) bU°odop ‘to seal with boards’ bU°odopbU°odop ‘to seal with boards repeatedly’
pItop ‘to spit on’ pItoppItop ‘to spit on repeatedly’
maakop ‘to mark on’ maakopmaakop ‘to mark on repeatedly’
njamo°ut ‘to eat some of’ njamo°utnjamo°ut ‘to eat some of repeatedly
(b) kot ‘to cut’ kotkot ‘to cut repeatedly’
tiif ‘to steal’ tiiftiif ‘to steal repeatedly’
dZuk ‘to pierce’ dZukdZuk ‘to pierce repeatedly’
-227-
rob ‘to rub’ robrob ‘to rub repeatedly’
(c) tSambaop ‘to cut/chop coarsely’ *
njakaop ‘to cut/chop crudely’ *
kjatao°ut ‘to scatter *
(deliberately)’
As was noted, while some forms freely undergo reduplication as in (90a) and (90b), other
forms like those in (103c) are not reduplicated, e.g. *njakaopnjakaop *‘to cut/chop
crudely repeatedly’. While this process is a case of full reduplication in the sense that all
of the segments of the base are copied in the reduplicant, these data suggest that the size
of the base is important in iterative reduplication. In every case where reduplication is
observed, the stem is either bisyllabic as in (103a) or monosyllabic as in (103b). It is
trisyllabic in those cases where there is no reduplication. The process can therefore be
said to be prosodically restricted since no reduplication is observed when the stems are
too large. The generalization is that the reduplicant cannot be larger than two syllables. I
will first focus on the analysis of forms with reduplication and then return to the cases
where no reduplication is observed.
The constraints given in (104) formalize the observation that the reduplicant (the
copied material) copies the entire base (the material available for copy).
(104) Maximize Base-Reduplicant (MAXBR): Every element of the base has a
correspondent in the reduplicant (McCarthy and Prince, 1995b).
-228-
As shown in Tableau 7.1, candidates that copy less material incur violations of this
constraint. For clarity, in this and subsequent tableaux each part of the word is
subscripted with an R for reduplicant and a B for base. In addition, I assume that a
reduplicative morpheme, RED, is associated with specific reduplication processes and as
such I include a subscript on each reduplicative morpheme to identify the specific
process. In this case, the reduplicative morpheme is RED[ITER]. I also assume that the
reduplicant is prefixed to the base though this is not crucial to the analysis.
/RED[ITER] + pItop/ MAXBR

) a. pItopRpItopB
b. pItoR pItop B p!
c. pIR pItopB t,o,p!
Tableau 7.1: [pItoppItop] ‘to spit on repeatedly’
The candidate that wins is the one in which all of the segments in the base are also copied
to the reduplicant, candidate (a). Candidate (b) fails to copy one segment (p) and incurs
one violation of MAXBR; candidate (c) fails to copy three segments (t, o, p), incurring
three violations. Both of these candidates are ruled out as possible surface forms in lieu
of candidate (a).
Iterative reduplications with complex stems are evaluated in a similar way as
illustrated in Tableau 7.2 with the word maakopmaakop ‘to mark repeatedly’.
-229-
/RED[ITER] + maakop/ MAXBR
) a. maakopRmaakopB
b. maakR maakop B o, p!
c. maaR maakopB k, o, p!
Tableau 7.2: [maakopmaakop] ‘to mark repeatedly’
In the case of words with complex stems, as in words with simple stems, the reduplicant
copies the entire base, anything less is unacceptable. The MAXBR constraint thus accounts
for iterative reduplicated words with simple stems like kot ‘to cut’Æ kotkot ‘to cut
repeatedly’ as well as those with complex stems like dZukop ‘to pierce (deliberately)’
Æ dZukopdZukop ‘to pierce repeatedly’.
Though the reduplicants in the former case are either monosyllabic or bisyllabic
they share the property of being a prosodic foot. The constraint in (105) formalizes this
observation.
(105) Reduplicant=Foot (RED[ITER].=FT): The iterative reduplicant is exactly one foot.
The constraint RED=FT requires that the size of the iterative reduplicant is a foot.
As the constraint is formulated, it allows for feet to be bimoraic as in (kot)FT kot ‘to cut
repeatedly’ or bisyllabic as in (dZu.kop)FT dZukop ‘to pierce repeatedly’. The important
thing is that in both cases the reduplicant is exactly one foot.
Tableau 7.3 illustrates that the correct surface form is still predicted for words in
(103a) and (103b), if we assume the constraint RED[ITER]=FT. I illustrate with pItoppItop
-230-
‘to spit repeatedly’. I assume that the actual surface form of both the base and the
reduplicant will have prosodic structure. However, for simplicity I mark foot
constituency only on the reduplicant with parentheses.
/RED[ITER] + pItop/ MAXBR RED[ITER]=FT

) a. (pIt.op)R pItopB
b. (pIt)R pItopB **!
c. (pI.to)R pItopB *!
d. pI.top R .pI.top B *
The optimal candidate in (a) is selected since it incurs no violations of the constraints
MAXBR or RED=FT. The reduplicant in candidate (b) does not copy two of the segments
seen in the base and as such incurs two violations of MAXBR which rules it out. The
reduplicant in candidate (c) fares better than candidate (b) but still copies too little of the
base; thus it has a fatal violation of MAXBR and is ruled out. The final candidate in (d)
does not have foot structure and as such fails on RED=FT.
The constraint in (106) militates against reduplicants which are not binary feet.
(106) Foot Binarity (FTBIN): Feet are binary under a moraic or syllabic analysis
(Prince and Smolensky, 1993; McCarthy and Prince, 1986, 1993).
The constraint FTBIN requires that prosodic feet are well-formed. It can be satisfied with
reference to syllables or moras, so feet can either be bisyllabic or bimoraic. As illustrated
-231-
in Tableau 7.4, both candidates completely copy the base and satisfy RED=FT. However,
candidate (b) is ruled out since its reduplicant has a foot (pI), which is neither bimoraic,
nor bisyllabic.
/RED[ITER] + pItop/ MAXBR RED[ITER]=FT FTBIN

) a. (pI.top)RpItopB
b. (pI)(top) R pItop B (pI)!
The importance of FTBIN to the analysis of iterative reduplication is seen when
we examine forms like those in (103c) which have no corresponding reduplicated forms.
MAXBR and RED[ITER]=FT are insufficient to account for these forms as they do not
explain why reduplication does not occur. As shown in Tableau 7.5, candidate (a)
satisfies both MAXBR and RED[ITER]=FT since its reduplicant copies the entire base and is a
foot. However, it is the incorrect surface form since words like these are not
reduplicated. This is indicated the symbol 0. An additional constraint is therefore
needed to rule out forms like these.
/RED[ITER] + njakaop/ MAXBR RED[ITER]=FT

0 a. (nja.ka.op)RnjakaopB
b. (nja.ka)(op)R njakaop B *!
c. (nja)(ka)(op)R njakaopB **!
Tableau 7.5: [njakaop] ‘to cut/chop crudely’
-232-
Tableau 7.6 reevaluates these candidates with the constraint FTBIN. In this case,
candidate (a) has a violation of FTBIN because the foot is not binary. Notice that we are
still unable to predict the correct output form, that being, no reduplication, since all the
other candidates incur fatal violations of the constraints considered. Though candidate
(a) is selected as optimal it is not the correct surface form.
/RED[ITER] + njakaop/ MAXBR RED[ITER]=FT FTBIN

0 a. (nja.ka.op)R njakaopB *
b. (nja.ka)(op)R njakaop B *!
c. (nja)(ka)(op)R njakaopB **! **
I assume that the constraint MPARSE in (107) is operative in the language.
MPARSE militates against the potential output of forms such as njakaop ‘to cut crudely’,
which have no reduplicated form. As I show just below, MPARSE is critical in predicting
the correct surface form.
(107) Morphological Parse (MPARSE): Penalise failure to assign morphological
structure (Prince & Smolensky, 1993; Poletto, 1998)
Following Prince and Smolensky (1993), I assume that there are no reduplicated outputs
of words like njakaop ‘to cut crudely’ and others in (103c) because they have no
-233-
phonological or morphological structure associated with segments, i.e. they are unparsed.
Thus, an unparsed item cannot be interpreted semantically or in terms of higher
morphological structure and does not violate any constraint on morphological or
phonological well-formedness. Such a candidate is referred to as the null parse.
Tableau 7.7 shows the correct selection of njakaop ‘to cut/chop crudely’.
/RED[ITER] + njakaop/ MAXBR FTBIN RED[ITER]=FT MPARSE

a. (nja.ka.op)RnjakaopB *!
b. (nja.ka)R njakaopB **!
c. (nja)(ka)(op)R njakaopB *!* *
) d. Ø *
The interaction of MPARSE and FTBIN shown in Tableau 7.7 ensures that the illformed
candidate (a) is not predicted as the surface form. Its reduplicant is trisyllabic and
therefore exceeds the binary maximum on footing imposed by FTBIN. Thus MPARSE is
crucially ranked below FTBIN. Candidate (b) does not copy enough of the base and so
incurs a fatal violation of MAXBR and candidate (c) violates FTBIN and RED=FT. The null
parse candidate, (d), incurs a violation of MPARSE but is selected as optimal since it
satisfies the higher ranked constraints. With MPARSE ranked lowest, a potential
reduplicant which would be larger than a binary foot never surfaces. This shows that for
words like njakaop ‘to cut/chop crudely’, which have no means of satisfying the prosodic
requirements for iterative reduplication, the only alternative for a grammatical output is
-234-
the null parse. Thus, while the iterative reduplicant must copy all of the base, it must do
so only if it satisfies the prosodic requirement on size. If the reduplicant would be too
large, no reduplication is observed.
Summarizing the arguments presented above, in iterative reduplication, copying the whole
stem is obligatory. However, no reduplication is possible if the requirement for the size of
the prosodic feet is not met. This is reflected in the constraint hierarchy for iterative
reduplication shown in (108).
(108) Constraint hierarchy for Iterative Reduplication
MAXBR, RED[ITER]=FT, FTBIN
MPARSE
7.3 Characteristic Reduplication
This section examines reduplicated forms which express the characteristic nature of the
base form, i.e. having the characteristic of X, where X refers to the semantics of the base
(cf. Kouwenberg and LaCharité, 2001). This type of reduplication may also be seen as full
reduplication in the sense that the base and the reduplicant are identical, e.g. naasI ‘nasty,
filthy’ Æ naasInaasI ‘having nasty, filthy characteristics’. However, as shown in (96b), a
-235-
segment [I] not seen in the input form is present both in the reduplicant and in the base,
e.g. laaf ‘to laugh’ Æ laafIlaafI ‘inclined to laughter’
Stem Reduplicated Form
(109) (a) grIEnI ‘grainy grIEnIgrIEnI ‘having grainy, coarse characteristics’

naasI ‘nasty, filthy naasInaasI ‘having nasty, filthy characteristics’
(b) laaf ‘to laugh laafIlaafI ‘inclined to laughter’
dZuk ‘to pierce dZukIdZukI ‘prickly or needle-like’
blak ‘black blakIblakI ‘having black spots or areas’
bwaI ‘boy bwaIIbwaII ‘characteristically boyish’
(c) lIzad ‘lizard’ *lIzadIlIzadI
tSupId ‘stupid’ *tSupIdItSupIdI
arindZ ‘orange’ *arIndZIarIndZI
(d) pIkInI ‘child’ *pIkInIpIkInI
Notice that all reduplicants are bisyllabic in the words in (109a), in addition, they all have
[I] as the final vowel which is also present in the stem. When [I] is not present in the stem
as in the words in (109b), the vowel is inserted. When adding the vowel would create a
reduplicant which is trisyllabic or longer no reduplication occurs as in (109c) and (109d).
I propose that the vowel is inserted in words in (109b) to satisfy a prosodic
requirement that the reduplicant be a bisyllabic foot. In addition, the vowel serves a
morphological function of marking adjectives of this type, making them distinct from
other adjectives with an intensive meaning (see examples in section 7.5.2) or those with a
distributive meaning (section 7.5.1). It is important to note that though bases in which
the vowel occurs, like *blakIADJ, *dZukIADJ, are phonologically identical to bases like
-236-
grIEnI ‘grainy’, they are not free form lexical items in the language26. While it is possible
to argue for an intrinsic characteristic interpretation for words like grIEnI ‘grainy’, this
argument cannot be extended to other words in this class.
Tableau 7.8 illustrates the selection of words with [I] in the input form with the
constraint MAXBR.
/RED[CHARAC] + naasI/ MAXBR

) a. naasIRnaasIB
b. naasR naasI B *!
c. naaR naasIB *!*
Tableau 7.8: [naasInaasI] ‘having nasty, filthy characteristics’
The best candidate is candidate (a) which completely copies the base. The other
candidates (b) and (c) do not copy all of the segments of the base and as such they incur
fatal violations of MAXBR and are ruled out.
Not only is MAXBR completely satisfied in these cases, but the reduplicant must
also be bisyllabic. The constraint in (110) formalizes the observation that well-formed
reduplicants in (109) are bisyllabic. Input words which are not bisyllabic would therefore
have to be augmented to satisfy this constraint.
(110) Reduplicant = Foot (RED[CHARAC]=FT[σσ]): The characteristic reduplicant is
exactly one bisyllabic foot.
26
In other CECs like Guyanese Creole (Devonish, 2003) and Trinadadian Creole (Donald Winford p.c.), these words
exist as free forms and can therefore have the ‘characteristic’ interpretation.
-237-
Tableau 7.9 shows the selection of laafIlaafI ‘inclined to laughter’. The reduplicants for
all of the candidates shown are exactly one foot. However, the candidates in (b) and (c)
have monosyllabic reduplicants and do not copy all the segments in their respective
bases. They are ruled out in favor of candidate (a) since it incurs no violations of the
constraints considered. It copies the entire base and is a bisyllabic foot.
/RED[CHARAC] + laaf/ MAXBR RED=FT[σσ]

) a. (laa.fI)RlaafIB
b. (laaf)R laafI B *! *
c. (laa)R laafIB *!* *
Tableau 7.9: [laafIlaafI] ‘inclined to laughter’
As was noted above, the vowel present in the base and the reduplicant in the
words in (109b) is not present in the input. These reduplicants are not merely bisyllabic,
they are crucially v-final with [I] as the final vowel, a vowel which is copied from the
base. The constraint in (111) formalizes this observation.
(111) Align Base Right, [I] (ALIGNBASE[CHARAC][I]): align the right edge of the base (in
characteristic reduplication) with the vowel [I].
The ALIGNBASE[CHARAC][I] requires the final vowel of the base to be [I]. As noted above,
considering MAXBR and ALIGNBASE[CHARAC][I] together implies that the reduplicant
-238-
copies the base. McCarthy and Prince (1993) argue that both base-reduplicant and
reduplicant-base copy are predicted by an OT analysis. In this analysis however, there is
no crucial evidence that would motivate one approach over the other.
The constraint as formulated forces the appearance of [I] in forms which do not
already have it as Tableau 7.10 illustrates.
/RED[CHARAC] + laaf/ MAXBR ALIGNBASE[CHARAC] [I] RED=FT[σσ]

) a. (laa.fI)RlaafIB
b. (laaf)R laafB *! *
Candidate (a) is selected as optimal since it inserts a vowel [I] and copies the entire base.
The competing candidate (b) does not insert a vowel so it incurs violations of
ALIGNBASE[CHARAC] [I] and RED=FT[σσ] and is ruled out.
Since the bases as well as the reduplicants of these words have the inserted
segment, the constraint seen in (112) is important to their evaluation. DEPIO militates
against insertion of segments in the base in cases where the input does not have an [I].
(112) DEPIO: every segment in the output has a correspondent in the input
(McCarthy & Prince, 1995a).
-239-
A consequence of segment insertion in the base is that the reduplicant must also insert a
segment in order to be completely similar to the base. The constraint in (113) penalizes
the insertion of a segment in the reduplicant.
(113) DEPBR: every segment in the reduplicant has a correspondent in the base
The difference in the relationship between the input and the output on the one hand and
the base and the reduplicant on the other, is schematized in (114) with the word
laafIRlaafIB ‘inclined to laughter’.
(114) Input /AfRED + laaf/
I-O Faithfulness
[laafIRlaafIB]
Output R B
R-B Identity
With the constraints RED=FT[σσ], and the DEP constraints, we can predict that the
base of characteristic reduplication is no larger than a bisyllabic foot but will be
augmented to the appropriate size if it would be too small. As shown, since all
reduplicated forms in (96b) have an epenthetic vowel, the DEP constraints must
necessarily be ranked low to ensure that these forms are not ruled out (Tableau 7.11).
-240-
/RED[CHARAC] + laaf/ MAXBR ALIGNBASE RED=FT[σσ] DEPBR DEPIO
[CHARAC][I]
) a. (laa.fI)R laafIB *
b. (laa)R laaf B *! * *
c. (laaf)R laaf B *! *
d. (laaf)R laafIB *! *
Candidate (a) is selected as optimal since its reduplicant is a bisyllabic foot and is
identical to the base. Candidate (a) also has the required [I] vowel in both the base and
the reduplicant. Since this vowel in not in the input candidate (a) violates DEPIO. The
reduplicant in candidate (b) does not copy the entire base and is a monosyllabic foot. It
incurs violations of MAXBR, RED=FT and ALIGNBASE[CHARAC][I] and is ruled out. The
reduplicant in candidate (c) is identical to its base but is also ruled out since it is a
monosyllabic foot. This tableau also illustrates what happens if the reduplicant does not
copy the base. Here, candidate (d) does just that. It is eliminated due to a fatal violation
of DEPBR. This shows that DEPIO must be ranked below DEPBR or else we would
incorrectly predict candidate (d) as the surface form.
Forms in which the base has the vowel [I] are accounted for trivially as shown in
Tableau 7.12.
-241-
/RED[CHARAC] + grIEnI/ MAXBR ALIGNBASE RED=FT[σσ] DEPBR DEPIO
[CHARAC] [I]
) a. (grIE.nI)R grIEnI B
b. (grIEn)R grIEnIB *! *
c. (grIE.nI.I)R grIEnIIB * *!
d. (grIE.nI.I)R grIEnIB * *!
Tableau 7.12:[grIEnIgrIEnI] ‘having grainy, coarse characteristics’
The candidate in (a) wins since it completely copies its base; it has a bisyllabic
reduplicant with the required vowel [I] in final position in the base and the reduplicant.
Since the vowel [I] is already in this candidate, it satisfies both DEPBR and DEPIO. This
is also the case for candidate (b), it also satisfies both DEP constraints. However, it fails
to copy the entire base and has a monosyllabic reduplicant. Consequently, it incurs
violations of MAXBR, ALIGNBASE[CHARAC] [I] and RED=FT and is ruled out. Candidates (c)
and (d) both have epenthetic vowels and thus satisfy ALIGNBASE[CHARAC][I]. However,
candidate (c) violates DEPIO since its base has one more segment than is seen in the input
form. Candidate (d) on the other hand, copies its base completely but also has one
additional segment not found in the base. It incurs a fatal violation of DEPBR and is ruled
out.
Neither the requirement for a bisyllabic foot nor for the occurrence of the vowel
[I] is sufficient to account for words like those in (109c), which also have bisyllabic
bases. For example, if having a bisyllabic foot were the only requirement we might
-242-
expect to find characteristic reduplications of the form *tSupIdtSupId but in fact there are
no such forms. Tableau 7.13 illustrates this using the same set of constraints as in
Tableau 7.9.
/RED[CHARAC] + tSupId/ MAXBR ALIGNBASE RED=FT[σσ] DEPBR DEPIO

[CHARAC] [I]
0 a. (tSu.pId)R tSupIdB *
b. (tSu.pI)R tSupIdIB *! *
c. (tSup)R tSupIdB *!* * *
Tableau 7.13: *tSupIdtSupId
The illformed candidate (a) is the predicted winner though it has a violation of
ALIGNBASE[CHARAC][I]. None of the other candidates are desirable since they make wrong
predictions about the surface form. For example, although candidate (b) satisfies
ALIGNBASE[CHARAC][I], it is suboptimal as it does not copy the base completely. As
illustrated in Tableau 7.14, the null parse plays a critical role in the evaluation of these
forms, just as seen in the previous section for iterative reduplication.
-243-
/RED[CHARAC] + tSupId/
ALIGNBASE
RED=FT[σσ]
[CHARAC] [I]
MPARSE
MAXBR
DEPBR
DEPIO
a. (tSupIdI)R tSupIdIB *! *
b. (tSupI)R tSupId B *!
c. (tSu.pId)R tSupId B *!
) d. Ø *
Tableau 7.14: *tSupIdItSupIdI
With the constraint MPARSE ranked lowest, reduplicants in which the vowel [I] appears
but which do not have a bisyllabic foot, are never selected as optimal. The unparsed
candidate (c) is selected as optimal since it does not violate any of the constraints. It is
the only option for satisfying all the conditions necessary for an optimal output of
characteristic reduplication. I propose that no reduplication is observed for input forms
such as this because the input stems are too large (see Kouwenberg and LaCharité, 2001)
and Gooden, 2003 for similar observations). The occurrence of the vowel [I] would yield
undesirable reduplicants which have trisyllabic rather than bisyllabic feet, e.g.
*tSupIdItSupIdI or *pIkInII-pIkInII.
The characteristic reduplicant is a bisyllabic foot and must end in the vowel [I] which
appears in base. However, if the stem is already bisyllabic and has [I] as a final vowel, the
-244-
vowel does not appear. In bisyllabic or trisyllabic stems where the appearance of the vowel
would yield a reduplicant that is larger than two syllables, no reduplication is observed.
This highlights the crucial relationship of identity which holds between a base and its
reduplicant. In these examples, base-reduplicant identity is enforced even when
requirements are vital only for the base. These preferences are formally expressed in the
constraint hierarchy for characteristic reduplication, shown in (115).
(115) Constraint hierarchy for Characteristic Reduplication
MAXBR, RED=FT[σσ], ALIGNBASE[CHARAC] [I],DEPBR
DEPIO
MPARSE
7.4 Intensive Reduplication (Complex Input)
The third type of reduplication to be discussed in this chapter is intensive reduplication.
As mentioned in Chapter 2, the stems that serve as the input for this type of reduplication
can be morphologically simple, comprised of a bare stem, or morphologically complex,
comprised of a root plus an affix (stem = root + affix). In this section, I discuss intensive
reduplications that are formed from morphologically complex stems. I reserve the
discussion of words with morphologically simple stems for section 7.5.2 where I discuss
prosodically disambiguated reduplications.
-245-
Complex input intensive reduplications contain suffixes like op, aut or aaf, which
give the interpretation of an action done intentionally or of an accomplished action
(Gooden, 2003). Examples are shown in (116) below.
(116) Stem Reduplicated Form

(a) grIn-op ‘to grin…’ grIn-grInop ‘to smile with someone
a lot’
blak-op ‘to make black’ blak-blakop ‘to make very black’
baks-op ‘to hit in the face..’ baks-baksop ‘to hit hard in the face’
a°Il-op ‘to soil/cover with oil ...’ a°Il-a°Ilop ‘to soil/cover

completely with oil ...’
kri°Ep-a°ut ‘to scrape out’ kri°Ep-kri°Epa°ut ‘to scrape out

completely’
brok-a°ut ‘to break out’ brok-broka°ut ‘to break out
completely’
kri°Ep-aaf ‘to scrape off’ kri°Ep-kri°Epaaf ‘to scrape off
completely’
njam-aaf ‘to eat completely’ njam-njamaaf ‘to eat completely…’
(b) naasI-op ‘to make filthy ...’ naasI-naasIop ‘to make completely
filthy’
plaasta-op ‘to daub/plaster ...’ plaasta-plaastaop ‘to daub/plaster ...’
dotI-op ‘to soil ...’ dotI-dotIop ‘to soil...’
(c) ari°EndZ-op ‘to arrange things’ *
sIdoN-op ‘to sit upright/sit relaxed *
for an extended period’
(d) maalIhak-op ‘to make a mess / to *
ruin’
batabruz-op ‘to bruise extensively...’ *
-246-
Where reduplication is observed, the reduplicant is a foot whether monosyllabic as in
(116a) or bisyllabic as in (116b). Of interest is the observation that there are no
reduplicated forms for the words in (116c) or (116d). This cannot simply be attributed to
a general prohibition on copying the affix, since as we saw in section 7.2, stems with
these same suffixes undergo iterative reduplication. I will argue here that the illegitimacy
of surface reduplicated forms in these words is directly related to the prosodic shape and
size of their potential bases. In all of these cases I assume that the base is identical to the
root. I motivate this assumption further below. This follows from the observation that the
intensive reduplicant maintains identity with the root (stem without suffix) of the input
form. For clarity, the proposed root constituency is indicated by brackets [ ].
For the words seen in (116d), the bases from which the potential reduplicants
would copy material are trisyllabic, however intensive reduplicants are maximally
bisyllabic. Further, while the surface reduplicants in (116a) and (116b) are exactly one
foot, the forms in (116d) would create more than one foot. No reduplication is seen in
these forms since the bases are too large. The constraint in (117) captures the
generalization that the intensive reduplicant is maximally one foot.
(117) Reduplicant=Foot (RED[INT]=FT): the intensive reduplicant is exactly one foot.
Tableau 7.15 shows that the only licit means for an output in these cases is the
null parse candidate, in (c), since all the other candidates violate the higher ranked
constraints. The reduplicant in candidate (a) has more than one foot and the reduplicant
-247-
in candidate (b) does not copy enough of the base. As seen above for iterative
reduplication, here there is no reduplication if the base is too large.
/RED[INT] + maalIhakop / MAXBR RED[INT]=FT MPARSE

a. (maa.lI)(hak)R[maalIhak]Bop *!
b. (maa.lI)R [maalIhak]Bop *!**
) c. Ø *
Tableau 7.15: maalIhakop ‘to make a tremendous mess of…’
The behaviour of the forms in (116c) suggest that the RED[INT]=FT constraint needs
to be refined. The bases from which the potential reduplicants would copy material in
forms like sIdoNop ‘to sit upright/relaxed for a prolonged period’ are exactly one foot and
thus satisfy the RED[INT]=FT constraint. Still, no reduplication occurs in these forms.
Interestingly, the forms in (116c) have stems made up of a sequence of LH syllables, that
is, they form iambic feet. In all the cases where reduplication is observed, the reduplicant
is comprised of a H foot, a LL foot or a HL foot. What is common among all of these
forms is that the reduplicant is a trochaic foot. We have independent evidence from stress
assignment, which motivated the analysis of JC feet in terms of trochees. The constraint
in (118) reflects this observation. Reduplication is not observed in the words in (116c)
since this would yield reduplicants with iambic feet.
(118) Reduplicant=Trochaic Foot (RED[INT]=FT[TROCH]): the intensive reduplicant is
exactly one trochaic foot (LL, H, HL, L ) (cf. Prince, 1990)
-248-
Following Prince (1990), I assume that trochaic feet comprised of two light (LL)
syllables or a single heavy (H) syllable, are preferable to those comprised of a sequence
of a heavy and a light (HL) syllable. These are in turn preferable to one comprised of a
single light (L) syllable. In cases where the output form cannot have the correct foot
type, the only legitimate output is the null parse as is illustrated in Tableau 7.16.
/RED[INT] + sIdoNop/ MAXBR RED[INT]=FT[TROCH MPARSE

a. (sI.doN)R [sIdoN]B op *!
b. (sI.do)R [sIdoN]B op *!
) c. Ø *
Tableau 7.16: sIdoNop ‘to sit upright/relaxed for a prolonged period’
Candidate (a) is completely faithful to the base but it fails on RED[INT]=FT[TROCH] since it is
an iambic foot (LH). Similarly, candidate (b) is ruled out because it violates the highly
ranked MAXBR as it fails to copy all the segments in the base. This leaves only the null
parse candidate in (c) which has only a violation of lowly ranked MPARSE since it has no
phonological or morphological content.
Tableau 7.17 demonstrates that words with heavy reduplicants as in (116a) are
easily accounted for under this analysis.
-249-
/RED[INT] + baksop/ MAX BR RED[INT]=FT[TROCH] MPARSE
) a. (baks)R [baks]B op
b. (bak)(sop)R [baksop]B *!
c. (baks)R [baksop]B *!*
d. Ø *!
Tableau 7.17: [baksbaksop] ‘to hit hard in the face’
The candidate in (a) with the root as base is selected as optimal since it has no violations
of the constraints considered. Candidates (b) and (c) have the base analysed as the stem.
Candidate (b) satisfies MAXBR but fails on RED[INT]=FT[TROCH] since it has two feet each a
heavy syllable. Candidate (c) is phonetically identical to the winning candidate (a) but
has the stem as base. It incurs two violations of MAXBR and is ruled out. The MPARSE
candidate in (d) is also ruled out.
Reduplicated forms with heavy-light (HL) reduplicants are also correctly selected
under this analysis, as illustrated in Tableau 7.18.
/RED[INT] + naasIop/ MAXBR RED[INT]=FT[TROCH] MPARSE

) a. (naa.sI)R [naasI] B op
b. (naa)R [naasI] Bop *!*
c. (naas)R [naasI] B op *!
d. (naa.sI.op)R [naasIop] B *!
e. (naa.sI)R [naasIop] B *!*
f. Ø *!
Tableau 7.18:[naasInaasIop] ‘to make completely filthy’
-250-
The optimal form, candidate (a), incurs no violations of the constraints considered.
Candidates (b) and (c) both have fatal violations of MAXBR as they do not copy all of
their bases. The base in candidates (d) and (e) is the entire stem. As in the example above,
this is problematic. Candidate (d) copies the entire base, however, this forms a trisyllabic
HLH foot causing a fatal violation of RED[INT]=FT[TROCH]. Candidate (e) copies less of its
base at the expense of fatal violations of MAXBR, which eliminates it. The null parse
candidate is eliminated due to its violation of the MPARSE. The evaluation of these
candidates illustrates that it is just as important for the reduplicant to copy the base
completely, as it is for the base to be a morphological root. In each case, the root is
synonymous with the base, all of which must be copied. However, as demonstrated these
roots/bases must also be trochaic feet.
To summarize the discussion and arguments presented above, it has been shown that an
intensive reduplicant is a trochaic foot and must copy the entire base. In addition, where the
base is too large or is the wrong foot type no reduplication is observed. In these cases we
saw that the MPARSE constraint was instrumental in accounting for these forms. I propose
therefore that the surface realization of the intensive reduplicant involves more than failure
to copy the input suffix. The generalization is that the reduplicant copies as much of the
base as possible without exceeding the prosodic requirement on foot size or foot form. In
(119) I summarize the constraint hierarchy for this subtype of reduplication.
-251-
(119) Constraint hierarchy for Intensive (Complex input) Reduplication27
MAXBR, RED[INT]=FT[TROCH]
MPARSE
7.5 Prosodically Disambiguated Reduplications
In this section, I discuss the two types of reduplication which yield segmentally identical
forms i.e. intensives and distributives. The phonetic analysis in the previous chapter
showed that these reduplicated words are different phonetically. We saw that when the
distributive reduplications occur in citation form or in final position in a statement
(nuclear position) they have a H+L* pitch accent on the foot with main stress (head foot).
Intensive reduplications, on the other hand, have a H+L* pitch accent on the analogous
foot in addition to a H* pitch accent on the preceding foot. Recall that we accounted for
the difference between the words as a difference in the alignment of the F0 peak within
the word. The goal is to interpret the phonetic differences within a formal phonological
analysis. I focus on the accent pattern on the reduplicated words in citation form and
statement final position for two reasons. The alignment difference between intensive and
distributive reduplications is most robust in these contexts. In addition, the types of pitch
accent that are realized on the words in other prosodic contexts correspond to differences
in the shape of the F0 contours associated with the contexts themselves.

27
An alternate ranking of RED[INT]=FT>>MPARSE>>MAXBR correctly predicts the surface forms for words like
naasInaasaIop ‘to make filthy’ but is unable to account for the absence of surface forms for words like *sIdoNop. This
reinforces the idea that MPARSE must be ranked lowest to account for those cases in which all the conditions for the
intensive reduplication cannot be satisfied.
-252-
Based on the results of the phonetic analysis, I suggest that the bitonal H+L* in
both types of reduplication is associated to the prosodic head of a word, i.e. the main
stress in the word. Distributive reduplication patterns like other words in the language
such as trisyllabic, quadrasyllabic and compound words and as such can be viewed as a
prosodic word. The prosodic word (a phonological unit) in this case is equivalent to a
morphological word and has a single pitch accent (H+L*) in citation form. I also assume
that the requirement is, all else being equal, only one pitch accent per prosodic word. This
is based on the observation that in the general case all non-reduplicated words have at
most a single H+L* pitch accent associated with the stressed syllable of the words in the
different prosodic contexts in which they where tested. Thus the pattern on distributives
reflects the regular pattern observed for non-reduplicated words in the language.
Conversely, since intensive reduplications have more than one pitch accent, I propose
that they form two prosodic words; they pattern with doubly accented phrases such as the
narrow focus constructions seen in section 6.2.2.3.
The data discussed in Chapter 4 showed how the stress patterns on reduplicated
words might be accounted for. We saw that main stress is predicted to fall on the
leftmost heavy syllable and in the absence of heavy syllables stress falls on the
penultimate syllable. Recall that quadrasyllabic monomorphemic words and
quadrasyllabic compounds are predicted to have primary stress on the penultimate
syllable and secondary stress on the initial syllable. Tableau 4.13 (repeated as Tableau
7.19) illustrates how the stress pattern on quadrasyllabic words is predicted.
-253-
) a. [(ÆgjalI)("mEnta)] **
b. [(ÆgjalI) ("mEn)ta] *! * **
c.[(Ægja)(lI"mEn)ta ] *! * * * ***, *
d.[(ÆgjalI)mEn("ta)] *! * * **
e. [(gjaÆlI) ("mEnta)] *! * **
f. [("gjalI)(ÆmEnta)] **! **
Tableau 7.19: [ÆgjalI"mEnta] ‘type of wood’
Candidate (a) wins since it completely satisfies the constraints considered. It has
secondary stress on the initial syllable and primary stress on the penultimate syllable.
These constraints also accounted for the location of primary stress in bisyllabic words as
shown in Tableau 4.14 (repeated as Tableau 7.20).

a. [("gaa)lIn] * *! * *
) b. [("gaalIn)] *
c.[("gaa)(ÆlIn)] *! * *
d.[gaa("lIn)] * *!
Tableau 7.20: ["gaalIn] ‘egret’
Note that it is the violation of PARSYL which rules out candidate (d) in favor of candidate
(b), which has the predicted stress pattern.
In bisyllabic (monosyllabic-based) distributive and intensive reduplication, main
stress falls on the heavy syllable or on the penultimate syllable if they are both light. In
-254-
quadrasyllabic (bisyllabic-based) distributive and intensive reduplication, main stress
falls on the penultimate syllable and secondary stress falls on the initial syllable. It is
with these syllables that the pitch accents are associated. Crucial to this analysis,
however, is the observation that different pitch accents are associated with the stressed
syllables of the reduplicated words. I account for these differences just below.
7.5.1 Distributive Reduplication
As noted above, the output of distributive reduplication is segmentally identical to that of
the intensive reduplications discussed in the previous section. I show here however that
distributive reduplication has a different prosodic structure such that it has only one
accentable syllable, since there is only one main stress in the word. Representative
examples are shown in (120). In these examples, primary stress is assigned to the
rightmost foot in the word and secondary stress to the leftmost foot.
(120) (a) grIEnI ‘grainy’ ÆgrIEnI"grIEnI ‘ grainy all over’

gomI ‘gummy’ ÆgomI"gomI ‘ gummy all over’
oglI ‘ugly’ ÆoglI"oglI ‘ ugly all over/in parts’
jala~jElo ‘yellow’ ÆjElo"jElo ‘yellowish all over’
haafa ‘halved’ Æhaafa"haafa ‘in halves’
pisa ‘spliced’ Æpisa"pisa ‘in pieces’
(b) blak ‘black’ Æblak"blak ‘black spots all over’

wayt ‘white’ Æwayt"wayt ‘whitish all over’
daag ‘dog’ Ædaag"daag ‘scattered dogs/many
different dogs’
-255-
grin ‘green’ Ægrin"grin ‘green spots all
over/greenish all over’
(c) blak ‘black’ ÆblakI"blakI ‘black spots all over’

grin ‘green’ ÆgrinI"grinI ‘green spots all over’
swit ‘sweet’ ÆswitI"switI ‘sweet ingredients all over’
wayt ‘white’ ÆwaytI"waytI ‘white spots all over’
As shown in these examples, there are two possible output forms for distributive
reduplication, either a quadrasyllabic form as seen in (120a) and (120c) or a bisyllabic
form as seen in (120b). Notice that the forms in (120a) have bisyllabic inputs while those
in (120b) and (120c) have monosyllabic inputs. As was seen for characteristic
reduplication, the input forms to distributive reduplication seen in (120c) are augmented
to create a bisyllabic base. Unlike characteristic reduplication however, there is an
alternative strategy for creating distributive reduplication as is illustrated by the forms in
(120b). I argue here that despite the difference in the segmental shape of these words,
they share a common prosodic property of bearing a single nuclear pitch accent.
As I show below in Tableau 7.21 and Tableau 7.22, the constraints MAXBR and
RED=FT[TROCH] introduced in the account of intensive reduplications, can also account for
shape and size of the distributive reduplicant. This is because the distributive reduplicant
is maximally a bisyllabic trochaic foot (LL, HL or H). In essence, any restriction on the
size and shape of the intensive reduplicant will also hold for the distributive reduplicant
given their segmental similarity.
To illustrate this first consider the selection of the forms in (107a) which have
bisyllabic inputs.
-256-
/RED[DIST] + grIEnI/ MAXBR RED[DIST]=FT[TROCH]
) a. (grIE.nI)R grIEnI B
b. (grIE)R grIEnI B *!*
Tableau 7.21:[ grIEnIgrIEnI] ‘grainy all over’
As shown in Tableau 7.21, the optimal candidate is (a) since it copies the entire base and
is a trochaic foot. Candidate (a) is thus better than candidate (b) which copies less
material from the base. Tableau 7.22 demonstrates that we can account for the size and
shape of the words in (120b) in a similar way. In this case as well, the optimal form is the
one which copies the base completely. Although candidate (b) has a heavy reduplicant
and satisfies RED[DIST]=FT[TROCH] it does not copy all of the base. Thus, it incurs a fatal
violation of MAXBR and is ruled out.
/RED[DIST] + grin/ MAXBR RED[DIST]=FT[TROCH]

) a. (grin)R grin B
b. (gri)R grin B *!
Tableau 7.22: [gringrin ] ‘green all over’
I now move the discussion to the account of alternate shape for words with
monosyllabic inputs. Tableau 7.23 illustrates the selection of forms with an epenthetic
vowel. As shown in (120c), a segment [I] is present both in the reduplicant and in the base
even though it is not seen in the input form. It is apparent that a constraint analogous to the
-257-
ALIGNBASE[I] constraint seen in (111) is needed here in addition to the DEP constraints
introduced in (112) and (113). For reference, I repeat these constraints in (121) below.
(121) a. Align Base Right, [I] (ALIGNBASE[DIST][I]): align the right edge the base of a
distributive reduplicant with the vowel [I].
b. DEPBR: every segment in the reduplicant has a correspondent in the base
c. DEPIO: every segment in the output has a correspondent in the input
In this case, candidate (a) is chosen as optimal since it satisfies all of the higher ranked
constraints. Candidate (b) is eliminated since it does not copy all of the base and
candidate (c) fails on ALIGNBASE[DIST][I] since the vowel [I] is not in the base.
/RED[DIST] + grin/ MAXBR RED[DIST]= DEPBR DEPIO ALIGNBASE

FT[TROCH] [DIST][I]
) a. (gri.nI)R grinI B *
b. (grin)R grinI B *!
c. (gri.nI)R grin B *! *
Tableau 7.23: [grinIgrinI]‘green all over’
Tableaux 7.22 and 7.23 thus show that distributive reduplicants must remain faithful to
their bases, whether or not these bases are monosyllabic or bisyllabic.
-258-
Tableau 7.24 shows the selection of surface forms with both monosyllabic and
bisyllabic bases. As noted above, the distributive reduplicant may be a bisyllabic foot but
need not be. This results in a low ranking of ALIGNBASE[DIST][I]. As shown here, since
there is no crucial ranking between ALIGNBASE[DIST][I] and DEPIO, both of the possible
output candidates (a) and (b) exemplifying the shape and size of the distributive
reduplicant are correctly predicted.
/RED[DIST] + blak/ MAXBR RED[DIST]= DEPBR DEPIO ALIGNBASE

FT[TROCH] [DIST][I]
) a. (blakI)R blakIB *
) b. (blak)R blak B *
c. (bla)R blak B *! *
d. (blak)R blakI B *!
e. (blakI)R blak B *! *
Tableau 7.24: blakIblakI ‘black all over’
Candidate (a) is selected as optimal since its reduplicant is identical to the base. Notice
however, that to achieve this it inserts a segment in the base and the reduplicant at the
expense of violating DEPIO. Candidate (b) is also a legitimate surface form for
distributive reduplication. It is identical to its base but violates ALIGNBASE[I] since it
does not have [I] as the final vowel in the reduplicant. The violations incurred by
candidates (a) and (b) are not sufficient to eliminate them as these constraints are lowly
ranked. Candidates (c) and (d) both fail on MAXBR and candidate (e) fails on DEPBR.
-259-
Although we have accounted for the segmental properties of the surface form of
distributive reduplication, we must bear in mind that segmental identity between the base
and the reduplicant is only part of the requirement, as the accentable syllable also gets a
H+L* pitch accent. The constraint given (122) formalizes the observation that the
rightmost foot in reduplicated words have a H+L* pitch accent in nuclear position.
(122) ANCHOR: A H+L* pitch accent is anchored to the prosodic head at the right
edge of a prosodic word. (ANCHORHL( PRWD))
ANCHORHL(PRWD) defines the location of the nuclear pitch accent on the
rightmost accentable syllable in the reduplicated word. I propose that this Anchor
constraint is a general constraint in the language. It reflects the observation that all words
have a HL melody in citation form and in statement intonation (nuclear position). As
discussed earlier, in chapter 6, distributive reduplications pattern with quadrasyllabic
non-reduplicated words and compounds since they all have main stress on the rightmost
foot in the word. In addition, in all these words there is a single H+L* pitch accent on the
stressed syllable of the head-foot of the word, i.e. the foot with main stress. In this way,
distributive reduplication behaves like a single prosodic word bearing a single nuclear
pitch accent. As I discuss further below (section 7.5.2), in contrast to compound words
and distributive reduplications, intensive reduplications have two syllables each capable
of bearing a pitch accent.
-260-
Tableau 7.25 shows the selection of the monosyllabic base reduplicated word
gringrin ‘green all over’. The constraints seen in the analysis of stress assignment
predict where in the word these syllables with main stress will occur. I have included the
constraints ALIGNHD(R) and ALIGNFT(R) here to demonstrate how the location of main
stress in the reduplicated word is predicted.
/RED[DIST] + grin/ ANCHORHL( PRWD) ALIGNHD(R) ALIGNFT(R)

H+L* *
|
) a. [(grin)R (grinB )]
H+L* *! * *
|
b. [(grin)R (grinB )]
Tableau 7.25: [Ægrin"grin] ‘green all over’
In this tableau, the winning candidate (a) has a H+L* pitch accent on the prosodic head of
the word and thus satisfies ANCHORHL(PRWD). Candidate (b) has main stress on the
initial syllable and incurs a violation of ALIGNHD. The H+L* pitch accent is also
incorrectly assigned to this foot causing a violation of ANCHORHL(PRWD).
In Tableau 7.26 we see the evaluation of a bisyllabic based reduplicated word,
gomIgomI ‘gummy all over’. As with the monosyllabic based word, the optimal
candidate, (a), has the H+L* pitch accent on the prosodic head-foot of the word, which in
this case is the penultimate syllable. Candidate (b) incurs a fatal violation of AnchorHL
-261-
since it makes the wrong prediction about which syllable is accented. It places the
nuclear pitch accent the initial foot of the word instead of on the final foot.
/RED[DIST] + gomI/ ANCHORHL( PRWD) ALIGN ALIGN

HD(R) FT(R)
H+L* **
|
) a. [(gomI)R (gomI B )]
H+L* *! ** **
|
b. [(gomI)R (gomI B )]
Tableau 7.26: [ÆgomI"gomI] ‘gummy all over’
Tableau 7.27 evaluates a bisyllabic based distributive word considering both the
constraints on segmental and prosodic well-formedness. For simplicity, I have excluded
the align constraints which refer to stress assignment.
-262-
/RED[DIST] + grinI/
RED=FT[TROCH]
ALIGNBASE
HL(PRWD)
ALIGNFT(R)
ALIGNHD(R)
ANCHOR
MAXBR
DEPBR
[DIST][I]
DEPIO
H+L* **
|
) a.[(grinI)R (grinI B )]
H+L* *! ** **
|
b. [(grinI)R (grinI B )]
H+L* ** *!
|
c.[(grin)R (grinI B )]
H+L* ** *! *
|
d. [(grinI)R (grin B )]
Tableau 7.27: [ÆgrinI"grinI] ‘green all over’
As shown here, candidate (a) is chosen as optimal since it best satisfies both the
segmental and the prosodic constraints. Candidate (b) has a violation of DEPIO as does
the winning candidate. However, candidate (b) is eliminated due to a fatal violation of
ANCHORHL. Candidates (c) and (d) satisfy ANCHORHL but incur fatal violations of the
constraints MAXBR and DEPBR respectively. This shows that constraints on segmental
and prosodic well-formedness are equally important for predicting the correct surface
form of distributive reduplication; though some constraints like DEPIO and
ALIGNBASE[DIST][I]are of less importance than others.
-263-
The data discussed in this section, shows that the distributive reduplicant is exactly one
foot in size. It copies the base completely but must also be correctly accented.
Specifically, distributive reduplications have a single H+L* pitch accent on the head-foot
of the word. The hierarchy of constraints for distributives is summarized in (123).
(123) Constraint hierarchy for Distributive Reduplication
MAXBR, RED[DIST]=FT[TROCH], ANCHORHL(PRWD), ALIGNHD(R)
DEPBR ALIGNFT(R)
DEPIO, ALIGNBASE[DIST][I]
7.5.2 Intensive Reduplication (Simple Input)
In this section I discuss the data for the simple input intensive reduplications. In
comparison to the examples seen in section 7.4, which had complex input forms, in this
case the entire stem, which is a single morpheme, serves as the base and in each case
copy is complete, e.g. laN ‘long’Æ laNlaN ‘very long’. In the examples shown in (124),
primary stress occurs on the second syllable in the word as well as on the initial syllable.
Thus, this type of reduplication has two stressed syllables which can potentially bear
pitch accents. This property appears to be restricted to this subtype of intensive
reduplication and does not appear to be relevant for intensive reduplications derived from
-264-
complex inputs. In all the examples seen below, the reduplicants are trochaic feet, thus
the constraint RED=FT[TROCH] introduced in (118) for the evaluation of intensives with
complex stems is relevant for these forms as well.
(124) (a)blak ‘black’ "blak"blak ‘very black’

laN ‘long’ "laN"laN ‘very long’
fUl ‘full’ "fUl"fUl ‘very full’
lU°o ‘low’ "lU°o"lU°o ‘very low’
wayt ‘white’ "wayt"wayt ‘very white’
grin ‘green’ "grin"grin ‘very green’
swit ‘sweet’ "swit"swit ‘very sweet’
niEr ‘near "niEr"niEr ‘very near’
(b) maaga ‘skinny’ "maaga"maaga ‘very skinny’
grIEnI ‘grainy’ "grIEnI"grIEnI ‘very grainy’
gomI ‘gummy’ "gomI"gomI ‘very gummy’
jala~jElo ‘yellow’ "jElo"jElo ‘very yellow’
Tableau 7.28 and Tableau 7.29 illustrate that MAXBR and RED=FT[TROCH] are able
to account for the segmental shape of both monosyllabic-based words and bisyllabic-
based words. In Tableau 7.28 the optimal form of the monosyllabic-based word is
candidate (a) since it copies all the segments of the base. As such it is better than the
other candidates in (b) and (c) which copy less material from the base.
-265-
/RED[INT] + laN/ MAXBR
) a. laNR laN B
b. laR laN B *!
c. lR laN B *!*
Tableau 7.28: [ "laN"laN] ‘very long’
For the bisyllabic-based word shown in Tableau 7.29, candidate (a) is chosen as optimal
since it satisfies all of the constraints considered. Candidate (b) has a bimoraic
reduplicant and satisfies RED=FT. However, not all segments in the base have
correspondents in the reduplicant so the candidate incurs a fatal violation of MAXBR and
is ruled out. Candidate (c) is also ruled out since its reduplicant does not have any feet.
MAXBR RED[INT]=FT[TROCH]
/RED[INT] + gomI/
) a. (go.mI)R gomI B
b. (gom)R gomI B *!
c. go.mIR gomI B *!
Tableau 7.29: ["gomI"gomI] ‘very gummy’
However, as noted for distributives, these constraints are insufficient to account
for the prosodic facts in this type of intensive reduplication. Specifically, in nuclear
position, the rightmost foot of intensive reduplications has a H+L* nuclear pitch accent
and the initial foot has a H* prenuclear pitch accent. The data presented in chapter 6
showed that this pattern is identical to that seen in narrow focus phrases. As proposed,
-266-
intensive reduplications can thus be analysed as having two prosodic words each with its
own pitch accent. As discussed in section 6.3, the presence of the additional grid mark at
the level of the prosodic word in intensive words licenses the presence of the prenuclear
accent.
Tableau 7.30 illustrates the selection of a monosyllabic-based intensive word,
gringrin ‘very green’. As before, prosodic word constituency is indicated by square
brackets.
/RED[INT] + gomI/ ANCHORHL(PRWD)

H+L*
|
0 a. [(grin)]R[(grin)] B
H+L*
|
b. [(grin)]R [(grin)]B *!
H* H+L*
| |
) c. [(grin)]R [(grin)]B
Tableau 7.30: ["grin"grin] ‘very green’
Candidate (b) incurs a violation of ANCHORHL because the H+L* pitch accent is on the
leftmost foot instead of on the rightmost one, it loses in favor of candidates (a) and (c)
which have the accent appropriately anchored to the rightmost foot. However, candidate
(c) rather than candidate (a) is the correct output form. As was noted above, there is
another property of intensive reduplications that would rule out candidate (a). Namely, a
H* pitch accent appears on the initial foot of intensive reduplications as is seen in
candidate (c). The constraint given in (125) specifies this.
-267-
(125) ANCHOR (H* LF): A H* pitch accent is anchored to the left edge of a
phonological phrase in intensive reduplications. (ANCHORH(LF, PPH) )
This constraint is specific to intensive reduplications in that it defines the location of the
prenuclear pitch accent on the initial foot. As before, the location of the pitch accents
themselves is predictable from the stress pattern on the word. According to the Prosodic
Hierarchy (section 1.6.2), every instance of the category Prosodic Word (PRWD) is
dominated by at least one phonological phrase. This means that the phonological phrase
can be seen as the host for both the prenuclear and the nuclear pitch accents. The
analysis in terms of the phonological phrase accounts for why the leftmost PRWD in
intensive reduplications does not get a nuclear accent. I suggest, that this position is
reserved for the first accent in the phonological phrase, a prenuclear accent. In addition,
as discussed in section 6.3, since the presence of the additional grid mark at the level of
the prosodic word licenses the presence of the prenuclear accent in intensive words, the
accent cannot appear in distributives given that they have a different metrical structure.
-268-
Tableau 7.31 shows the full evaluation of a monosyllabic-based intensive word.
/RED[INT] + grin/ ANCHORHL(PRWD) ANCHORH (LF, PPH)

H* H+L*
| |
) a. [(grin)]R[(grin)] B
H+L*
|
b. [(grin)]R [(grin)]B *! *
H+L*
|
c. [(grin)]R [(grin)]B *!
H*
|
d. [(grin)]R [(grin B)] *!
H+L* H*
| |
e. [(grin)]R [(grin)]B *! *
Tableau 7.31: ["grin"grin ] ‘very green’
Candidate (a) wins since it satisfies both of the constraints considered. It has two
accented feet, the first of which is assigned a prenuclear H* accent and the second a
H+L* accent. The other candidates incur violations of both ANCHORHL and ANCHORH.
Candidate (b) has only one accent, a nuclear pitch accent, which is incorrectly anchored
in the initial foot, thus it violates ANCHORHL. It also violates ANCHORH since it does not
have a prenuclear accent. Candidate (c) likewise has only one accented foot, which is
appropriately anchored to the foot with main stress. This is the pattern seen in
distributives, but in this case the candidate violates ANCHORH since it does not have a
prenuclear accent. Candidate (d) shows that having only a prenuclear accent is
prohibited; it fatally violates ANCHORHL. Candidate (e) has both pitch accents but they
-269-
are misplaced causing violations of both ANCHORHL and ANCHORH. This shows that a
well-formed surface form for intensive reduplication must have both a prenuclear H*
pitch accent and a nuclear H+L* pitch accent appropriately associated with the initial and
rightmost feet respectively. That is, the left and right edge of the phonological phrase.
Tableau 7.32 shows the selection of a reduplicated word with a bisyllabic-base.
The candidate in (a) is the best choice since it satisfies both of the anchor constraints. It
has a prenuclear accent on the leftmost foot and a nuclear accent on the rightmost foot.
These mark the edges of the phonological phrase. Both candidates (b) and (c) incur fatal
violations of the anchor constraints and are ruled out.
/RED[INT] + gomI/ ANCHORHL(PRWD) ANCHORH (PPH)

H* H+L*
| |
) a. [(gomI)]R [gomI] B
H+L*
|
b.[(gomI)]R [gomI] B *!
H*
|
c.[(gomI)]R [gomI] B *!
Tableau 7.32: [ "gomI"gomI] ‘very gummy’
Tableau 7.33 shows the selection of a bisyllabic based word with the inclusion of
the constraints on the segmental shape of the word and its prosodic well-formedness.
-270-
/RED[INT] + gomI/ ANCHOR ANCHORH MAXBR RED[INT]=FT[TROCH]
HL( PRWD) (LF, PPH)
H* H+L*
| |
) a.[(gomI)R (gomI B )]
H+L* H* *! *
| |
b. [(gomI)R (gomI B )]
H* H+L* *!
| |
c.[(gom)R (gomI B )]
Tableau 7.33: [ "gomI"gomI] ‘very gummy’
As before, the optimal candidate is the one which satisfies both the segmental constraints
and the prosodic constraints, in this case candidate (a). Candidate (b) satisfies both
constraints on segmental shape of the output form but fails on the Anchor constraints
since the pitch accents are associated incorrectly with the word. Candidate (c) has the
appropriate prosodic form, both a prenuclear and nuclear accent but the entire base is not
copied. This reinforces the idea that intensive reduplications must equally satisfy
conditions on segmental and prosodic well-formedness to produce grammatical surface
forms.
In summary, the intensive reduplicant is a prosodic foot. In this subtype of intensives, the
reduplicant copies the base completely while adhering to the prosodic requirements on
foot structure. Further, it must adhere to other prosodic requirements which permit its
-271-
stressed syllables to be accented. That is, it must have two accented syllables in order to
be a well-formed output. The hierarchy of constraints in summarized in (126).
(126) Constraint hierarchy for Intensive (Simple Input) Reduplication
MAXBR, RED[INT]=FT[TROCH], ANCHORHL(PRWD), ANCHORH (LF, PPH)
7.5.3 Section Summary
The main points discussed in this section are as follows. First, both intensive and
distributive reduplications are right prominent, having their nuclear stressed syllables
dominated by the rightmost foot in the word. Both are similar to compound words which
are also right prominent, having their main stress on the rightmost foot in the word.
Second, while both intensive and distributive reduplications involve full reduplications,
they crucially satisfy different prosodic requirements to achieve a well-formed output.
With respect to their stress patterns, for example, intensive reduplications have two
metrically strong (accentable) syllables which each carry a pitch accent in citation form
and in statement intonation. Distributive reduplications, on the other hand, have only one
metrically strong (accentable) syllable which carries a pitch accent in the same contexts.
Third, as noted above, unlike intensives, distributive reduplications pattern with non-
reduplicated words in the language such as compounds, quadrasyllabic words and non-
-272-
reduplicated words. All of these have a single pitch accent in citation form and in
statement intonation. Thus, distributives reflect the regular pattern.
7.6 Previous Accounts of JC Reduplication
In this section, I review two previous phonological analyses of JC reduplication.
McCarthy (1983) proposed a segmental analysis of JC reduplication in which the
template defining the JC base is as in (127).
(127) Onset V (C(V)) where ‘onset’ is a possible syllable onset in JC
This template is based on data from DeCamp (1974) and so does not cover the range of
data examined in this dissertation. Consequently, it does not make the right predictions
about the possibilities for reduplicants in JC reduplication. For example, the template
incorrectly predicts that coda consonants are prohibited in the syllables of reduplicants.
Also, the template predicts that there are no long vowels or diphthongs in reduplicants.
To the contrary we have seen several examples with input forms with long vowels e.g.
maaga ‘skinny’ and diphthongs e.g. grIEnI ‘grainy’. Apart from the empirical issues, a
segment based account misses the generalization that in all cases the JC reduplicant is a
prosodic foot. Within the OT framework assumed here, the fact that some reduplicants
are a bisyllabic foot is predicted by the same constraints as those that predict that other
reduplicants can be either a bimoraic or a bisyllabic foot. A segment-based account
would require different templates to account for these differences.

-273-
Alderete (1993) proposed a syllable-based account of JC reduplication within a
prosodic morphology approach. The analysis is also based on data from DeCamp (1974)
and so is restricted in the range of data covered. Alderete proposed that the base for JC
reduplication is a minimal prosodic word that contains at least one foot, a moraic trochee.
That is, the base is a foot comprised of LL syllables or a H syllable. The approach
assumes that in HL cases, the final L syllable is unfooted. Alderete’s analysis is similar to
the one presented here in that phonological conditions on the process are delimited in
terms of authentic units of prosody following the assumptions of prosodic morphology
(McCarthy and Prince, 1993; McCarthy and Prince, 1995a; McCarthy and Prince,
1995b). However, the analysis presented here does not crucially assume a moraic
trochee analysis. Evidence that the moraic trochee analysis is not satisfactory can be
taken from the fact that the portion of the word corresponding to the reduplicant is not
necessarily a moraic trochee. For example, there are LL reduplicants as in yEloRyEloB
‘very yellow’; HL reduplicants as in grIEnIRgrIEnIB ‘very grainy’; HH reduplicants as in
njamopRnjamopB ‘to eat all of repeatedly’; LH reduplicants as in blakopRblakopB ‘to
blacken repeatedly’ and H reduplicants as in blakRblak Bop ‘to make very black’.
Alderete’s analysis cannot explain why the LH reduplicant is permissible in words like
blakopRblakopB ‘to blacken repeatedly’ but not in words like blakRblak Bop ‘to make
very black’ or why HH reduplicants are permissible in words like njamopRnjamopB ‘to
eat all of repeatedly’. As I discuss below, this is directly related to the theoretical
treatment of the base in the reduplication process.
-274-
The shortcomings of both of these proposals is partly an empirical issue and
partly theoretical. First, as I have noted above, the analyses are based on restricted
datasets which do not reflect the full range of possibilities for reduplication in JC.
Second, from a theoretical perspective, both of these analyses specify conditions on the
size and shape of the input in JC reduplication whereas this analysis specifies conditions
on the reduplicant and the base. This is a reflection of the theoretical approach taken here
that both the base and the reduplicant must satisfy phonological as well as morphological
requirements. This was shown to be instrumental in accounting for Characteristic type
reduplications (section 7.3) in which a vowel not seen in the input was present both in the
base and the reduplicant in the output form. Second, in this approach, the base is not
necessarily identical to the input form since the base is the portion of the input from
which segmental material is copied. This conceptualization of the base is vital in
accounting for differences between iterative reduplication whose base is a stem
equivalent to the input and intensive reduplication (complex input) whose base is a root
(i.e. the stem minus the affix of the input). Third, neither of the previous approaches
considered the role of JC prosody in disambiguating segmentally identical reduplicated
words like the intensive and distributive reduplications discussed here.
7.7 Chapter Summary
In this chapter JC reduplication was shown to be systematic and predictable. Four types
of reduplication processes were identified based on the semantic ideas expressed by each
output form. The discussion of reduplication in terms of prosody was vital to the analysis
-275-
presented. In particular, differences between the reduplication processes were shown to
be directly related to the type of prosodic restrictions imposed on them. Further, each
process has a specific reduplicative morpheme which has separate phonological
requirements. I review these differences below.
The first issue has to do with the shape and size of the reduplicant. In all of the
cases seen, the JC reduplicant is completely faithful to the base, i.e. MAXBR is always
satisfied (full reduplication). However, the input form was not always identical to the
reduplicant. In these cases, the size and form of the reduplicant was shown to be
prosodically determined. The intensive reduplicant was shown to be a trochaic foot; the
characteristic type reduplicant was obligatorily a bisyllabic foot; the iterative and
distributive reduplicants could be either a bisyllabic or a bimoraic foot. The
generalization is that reduplicants across all the different types of reduplication processes
are consistently a prosodic foot. However, the phonological shape and form of these feet
vary according to the type of reduplication process involved.
The second issue is the location of the reduplicant with respect to the base. In
processes of full reduplication, it is not always apparent which portion of the output
string is the base and which is the reduplicant. I have assumed that the reduplicant is
prefixed to the base since it is not crucial to the analysis and there is no crucial evidence
to the contrary.
The third issue concerns the fixed vowel segment /I/ in both the reduplicant and
the base of characteristic and some distributive reduplications. While this is obligatory
for characteristic type reduplication it is optional for distributive reduplication. The fact
-276-
that the quality of the vowel is invariant in these forms is suggestive of a morphological
function denoting reduplicated adjectives that can express semantic notions of the
characteristic nature of the base form.
Fourth, there were several cases in which the reduplicant could not satisfy the
prosodic requirements of its subtype and, as a result, no reduplication was observed. In
these cases, the constraint MParse proved crucial to the analysis. In particular, iterative
and characteristic type reduplications were blocked when the reduplicant would be larger
than two syllables. Intensive Reduplication (complex input) was blocked when the foot
type and foot size requirements could not be satisfied.
Lastly, the prosodic structure of intensive and distributive reduplication with
respect to their stress patterns determined the type and number of pitch accents associated
with them. While intensives were doubly accented, distributives were accented only
once. Since doubly accenting the reduplicated word was only observed in the intensive
reduplication process, this suggests that it may serve a pragmatic function of marking
‘emphasis’ in intensive reduplications. As noted earlier, a similar property was seen in
narrow focus constructions. In addition, while distributives pattern with phonological
words in the language, while intensives do not.
To summarize, some constraints proposed above are prosodically motivated.
These are aimed at forming a preferred foot, whether by failing to copy an affix,
restricting the length of the input, augmenting the reduplicant or prohibiting reduplication
(see 128). In each case the reduplicants copied as much of the input material as possible
without exceeding the prosodic limit. In addition, the constraints specific to the
-277-
contrastive intensive and distributive reduplications aimed at forming well formed
prosodic units in terms of their rhythmic organization.
A summary of the constraints posited for JC reduplication is given in (128).
These constraints are able to account for the range of reduplicated forms in the language.
For simplicity, the constraint RED=FT is stated in its general form and constraints related
to stress assignment are omitted.
(128) Summary of Constraints for JC Reduplication.
DEPBR, MAXBR, RED=FT, FTBIN, ALIGNBASE[CHARAC][I], ANCHORHL(PRWD), ANCHORH (LF, PPH)
DEPIO, ALIGNBASE[DIST][I]
MPARSE
-278-
CHAPTER 8
CONCLUSION
The research discussed in this dissertation serves to elucidate the phonological and
phonetic properties of reduplication processes in Jamaican Creole. I have shown that
surface differences in the shape of the reduplicant across different reduplication processes
are straightforwardly accounted for by referring to the prosodic properties of the
reduplicated words. Further, I have shown that segmental ambiguities between
semantically different reduplicated words are resolved at a higher level in the prosodic
hierarchy, i.e. the level of the prosodic word. It is hoped that this will lead to comparable
research into reduplication processes in other Caribbean English Creoles (CEC) and will
thus help us to better understand how the prosodic systems of CECs interact with
morphological processes in the languages.
A secondary goal of this dissertation was to situate the word-level prosodic
system of JC in a wider taxonomy of prosodic systems, i.e. tonal, stress-accent or non
stress-accent. I have shown that the JC has a stress-accent system in which lexical
contrasts result from differences in the alignment of the F0 contour with the stressed
syllables of words. The treatment of the JC prosodic system provided the
-279-
backdrop against which the proposed prosodic differences between the segmentally
identical intensive and distributive reduplications were interpreted.
In Chapter 2 I showed that the processes of reduplication found in Jamaican
Creole are also represented in other CECs. However, the semantic types in JC
reduplication are a subset of those found in other CECs. In particular, while some CECs
have a distinct class of attributive reduplication, I suggest that this meaning might be
expressed through distributive or characteristic reduplication in JC. In terms of the size
of the copied material, the majority of processes in CECs are processes of full
reduplication though some cases of partial reduplication are reported. I discussed data
showing that in many instances reduplication created segmentally identical output forms.
In these cases CECs manipulated either the stress or the tonal properties of the
reduplicated words to signal differences in meaning.
In Chapter 4 I discussed phonetic evidence for a characterization of the JC
prosodic system as a stress system. I showed that lexical contrasts in JC are signalled by
differences in the alignment of the F0 contour within the word. This is in contrast to
‘pure’ tonal systems in which we can attribute lexical contrasts to specific F0 shapes. In
particular, I showed that main stress in JC is characterized by a fall in pitch onto that
syllable. In addition, the preceding F0 peak may or may not be aligned with the same
syllable as the F0 fall. The first alignment pattern was discussed in this chapter with
regard to unreduplicated words. We saw that lexical contrasts are signaled by differences
in the alignment of the F0 fall with the prosodic head of the word.
-280-
In Chapter 5, I reviewed phonological evidence showing that the prosodic system
of JC is best described as a weight-sensitive stress-accent system. I showed that in the
unmarked case, stress in JC is assigned preferentially to heavy syllables and in the
absence of heavy syllables to penultimate syllables. I argued that segmentally identical
reduplicated words are differentiated by their stress pattern. This chapter also presented
an analysis of word-level stress.
Chapter 6 reviewed the phonetic properties of in intensive and distributive
reduplicated words and a second alignment pattern was seen. Specifically, contrasts are
signaled by differences in the alignment of the F0 peak with the word. In addition, there
was a difference in whether or not there was a prenuclear rise on the initial foot of the
word. Both types of reduplicated words are similar in that they have utterance-final falls.
However, in intensive words this is preceded by a prominence lending rise on the initial
foot of the word.
With regard to the formal Optimality Theoretic account of reduplicated words, in
Chapter 7, I showed that reduplicants in JC copy as much of their base as possible
without exceeding prosodic limits. In fact, the process can be described informally as an
‘all or nothing’ strategy since if full copy is not attainable within the specified conditions,
reduplication does not occur. Further, whereas intensive reduplications are doubly
accented, distributive reduplications have a single nuclear pitch accent. I argued that
these differences are related to differences in the prosodic structure of the words, by
which the pitch accents are licensed.
-281-
This study has import for the study of reduplication processes crosslinguistically.
Processes of full reduplication are often overlooked as being phonologically uninteresting
since no phonological alternations are evident. The JC data presented here suggest that
there are in fact phonological alternations albeit at higher levels of the prosodic hierarchy.
In JC the morphological properties of the words interact with stress at the level of the
prosodic word, to resolve segmental ambiguity in two reduplication processes.
Empirically, this dissertation adds to the stock of cross-linguistic research on the study of
prosodic disambiguation in languages. It provides evidence for an important link between
the prosodic properties of reduplication processes and their semantic properties.
The research discussed in this dissertation is also important in that it contributes
to the study of the prosodic system of Creoles, an area of research largely unexplored in
Creole linguistics. It fortifies the type of groundwork necessary for using facts about
the prosodic system of CECs to elucidate the discussions about the origins of these
Creole varieties, as well as their relationship to each other (cf. Devonish, 1989, 2002).
This study has provoked interest in several aspects of research which merit future
investigation; these are outlined below. First is the question of whether there are other
acoustic cues to JC stress. As we saw in Chapter 4, the stressed syllable in JC
consistently has a pitch fall onto that syllable. However, we need to discover what other
acoustic cues are important in signaling stress. For example, cross-linguistic studies on
the acoustic cues to stress contrast show that the most consistent and significant
correlates are fundamental frequency and duration though there are other cues such as
intensity and vowel quality differences (Fry, 1955, 1965; Lehiste, 1961; Bruce, 1977
-282-
among others). This is related to second issue, the listeners’ perception of stress. It would
appear that the percept of stress in JC may be elicited by reference to the alignment of
pitch falls with the word. Another aspect for further investigation is to find out whether
there are differences in the perception of stress related to the timing of the pitch fall.
Verhoeven (1994), for example, showed that in Dutch, listener judgements were more
sensitive to variations in the timing of F0 falls than to variations in the timing of F0 rises
in utterances. The data discussed in Chapter 6 showed that there is a difference in the
timing of the fall on the word-final foot in intensive and distributive reduplicated words.
However, we do not know what effect the late versus early fall has on listeners’
perception of semantic differences between these words. We also saw that some speakers
produced intensive reduplications with a higher pitch on the initial foot in lieu of a
prenuclear rise. The question is how are listener judgments affected by the presence or
absence of the prenuclear rise in intensive words?
Following research by Kawasaki (1993) and Ohala (1993), I suggest that the
perceptibility of stress might be enhanced where there is greater modulation in the F0
signal, for example, where there is a fall in the pitch onto stressed syllables. According
to Kawasaki and Ohala’s research, greater modulation in a sequence contributes to
greater perceptibility. However, in some cases we saw that the fall into the stressed
syllable was absent in question intonation and further that this had the potential for
contrast neutralization in the speech of some speakers. In these contexts, I suggest that
listeners may have to rely on other acoustic cues to stress such as duration or vowel
-283-
quality differences as the timing cue will be less useful. I leave these issues open for
future consideration.
-284-
APPENDIX A
Informant Data Sheet

Tape #: _____________
Name: _____________________________
Age: ___________
Occupation (or parent for minors):_____________________________
Education (highest level):_____________________________________
Places of residence:
Present:______________________ from 19_______ to ________
Past: _________________________ from 19_______ to ________
Languages spoken:____________________________________________
Language used most often:______________________________________
- in general:____________________________
- at home:______________________________
- talking about important things (school, business, etc):__________
- talking about family matters (quarrels, etc):___________________
- talking to children:________________________________________
- for gossiping, jokes, secrets etc:____________________________
- among close friends:____________________________
- in the community/neighborhood:__________________________
- Any other situations in which these languages are typically used
- 285 -
APPENDIX B
The following graphs show representative examples of the monosyllabic based
reduplicated word gringrin, produced with an intensive meaning and a distributive meaning
by RP.
greenDISdf4 RP greenINdf1 RP
150
125
100
75
4000
2000
0
tones L% H+L*L%
phones g ® i n g ® i g ® i n g ® i n
words mi want di wan dem we griingriin mi want di mango dem we greengreen
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6

Figure B.1. Fundamental frequency contour of the reduplicated word gringrin produced
with a distributive meaning ‘scattered green (left graph) and an intensive meaning ‘very
green’ (right graph) in statement intonation final position.
- 286 -
greenDISdnf3 RP greenINdnf1 RP
150
125
100
75
4000
2000
0
tones L% H+L* L%
phones g ® i n g ® i n g ® i n g ® i n
words mi want di griingriin mango dem want di griingriin mango dem
0.4 0.8 1.2 0.4 0.8 1.2

Figure B.2. Fundamental frequency contour of the reduplicated word gringrin

produced with a distributive meaning ‘scattered green (left graph) and an intensive
meaning ‘very green’ (right graph) in statement intonation non-final position.
greenDISynf2 RP greenINynf2 RP
150
125
100
75
4000
2000
0
tones H% H* H%
words yu want di wan dem we griingriin yu want di mango dem we griingriin
0.4 0.8 1.2 0.4 0.8 1.2

green’ (right graph) in question intonation final position.
- 287 -
greenDISynnf4 RP greenINynnf1 RP
150
125
100
75
4000
2000
0
tones H% H+L* H%
words yu want di griingriin mango dem yu want di griingriin mango dem
0.4 0.8 1.2 0.4 0.8 1.2

green’ (right graph) in question intonation non-final position.
- 288 -
BIBLIOGRAPHY
ADAMSON, LILLIAN and SMITH, NORVAL. 1999. Derivational Reduplication in Sranan.

Paper presented at 4th Westminster Creolistics Workshop on Reduplication in
Contact Languages, April 8th -10th, London.
—. 2003. ed. by Silvia Kouwenberg: Battlebridge Publications.
AKERS, GLEN. 1981. Phonological Variation in the Jamaican Continuum. Ann Arbor:
Karoma.
ALDERETE, JOHN. 1993. The Prosodic Morphology of Jamaican Creole Iteratives.

University of Massachusetts Occasional Papers 20, the UMOP in Indigenous
Languages, ed. by Elena Benedicto, 29-50. Amherst: Graduate Linguistic Student
Association.
ALLEYNE, MERVYN. 1980. Comparative Afro-American: A Historical Comparative Study

of English-based Afro-American Dialects of the New World. Ann Arbor: Karoma.
ALLEYNE, MERVYN. 1987. Predicate Structures in Saramaccan. Studies in Saramaccan

Language Structure, ed. by Mervyn Alleyne, 71 -87. Amsterdam/Jamaica:
University of Amsterdam/University of the West Indies.
ALLSOPP, RICHARD. 1996. Dictionary of Caribbean English Usage. Oxford: Oxford

University Press.
ARENDS, JACQUES. 1995. The early stages of creolization. Amsterdam ; Philadelphia: John
Benjamins Publishing Company.
ARCHANGELI, DIANA and LANGENDOEN, D.TERENCE (eds.) 1997. Optimality theory: an

overview. Explaining Linguistics. Oxford: Blackwell.
ARVANITI, AMALIA. 1992. Secondary Stress: evidence from Modern Greek. Papers in
Laboratory Phonology II: Gesture, Segment, Prosody, ed. by G. Docherty and D.
Ladd. Cambridge: CUP.
BAKER, PHILIP. 1999. Reduplication in Mauritian Creole. Paper presented at 4th

Westminster Creolistics Workshop on Reduplication in Contact Languages,
London.
- 290 -
BAKKER, PETER. 1987. Reduplications in Saramaccan. Studies in Saramaccan Language
Structure: Caribbean Culture Studies 2, ed. by Mervyn Alleyne, 17 - 40.
BECKFORD-WASSINK, ALICIA. 1999. A Sociophonetic Study of Jamaican Vowels,

Linguistics, University of Michigan: PhD Dissertation.
—. 2001. Theme and variation in Jamaican Vowels. Language Variation and Change,
13.135-59.
BECKMAN, MARY E. 1984. Toward Phonetic Criteria for a Typology of Lexical Accent,
Department of Linguistics, Cornell University: PhD Dissertation.
—. 1986. Stress and Non-Stress Accent: Netherlands Phonetic Archives. Dordrecht -

Holland/ Riverton -USA: Foris Publications.
—. 1996. The Parsing of Prosody. Language and Cognitive Processes, 11.17 -67.
BECKMAN, MARY E. and PIERREHUMBERT, JANET. 1986. Intonational Structure in Japanese

and English. Phonology Yearbook 3, 255 -310.
BROUSSEAU, ANNE-MARIE. 2003. The Accentual System of Haitian Creole: The Role of
Transfer and Markedness Values. To appear in Linguistische Arbeiten, ed. by Ingo
Plag: Niemeyer Publishers.
BROSELOW, E and MCCARTHY, JOHN. 1983. A Theory of Internal Reduplication. The

Linguistic Review, 3.25 - 88.
BRUCE, GOSTA. 1977. Swedish Word Accents in Sentence Perspective: Travaux de

L'institute de Linguistique de Lund 12: CWK Gleerup.
BUCKLEY, EUGENE. 1997. Alignment in Manam Stress. Linguistic Inquiry.
CARTER, HAZEL. 1980. How to be a Tone Language: Theoretical Considerations involved

in the Classification of Creole Languages as Tonal (or otherwise). Paper presented
at Society for Caribbean Linguistics, Aruba.
—. 1983. How to be a Tone Language: Theoretical Considerations involved in the

Classification of Creole Languages as Tonal (or otherwise). Studies in Caribbean
Language, ed. by Lawrence Carrington, 90-111. Society for Caribbean Linguistics.
—. 1987. Suprasegmentals in Guyanese: some African comparisons. Pidgin and Creole

Languages, ed. by Glen Gilbert, 213 - 63. Honolulu: University of Hawaii Press.
CASSIDY, FREDERICK G. 1957. Iteration as a Word Formation Device in Jamaican Folk

Speech. American Speech, 32. 49 -53.
- 290 -
CASSIDY, FREDERICK G. 1961. Jamaica Talk: Three Hundred Years of the English
Language in Jamaica. London: Macmillan.
CASSIDY, FREDERICK G and LEPAGE, ROBERT. 1967. Dictionary of Jamaican English.

Cambridge: University of Cambridge Press. [Reprinted in 1980].
CLEMENTS, G.N. 1985. The Problem of Transfer in Non-Linear Morphology. Cornell

Working Papers in Linguistics, 7.38 - 73.
CRYSTAL, DAVID. 1997. A Dictionary of Linguistics and Phonetics: Basil Blackwell.
DECAMP, DAVID. 1971. Towards a Generative Analysis of a Post-Creole Continuum.

Piginization and Creolization of Languages, ed. by Dell Hymes, 349 - 70.
Cambridge: Cambridge University Press.
DECAMP, DAVID.1974. Neutralization, Iteratives and Ideophones: the locus of language in

Jamaica. Pidgins and Creoles: current trends and prospects, ed. by David DeCamp
and Ian Hancock, 46 -50. Georgetown University Press.
DE JONG, KENNETH and ZAWAYDEH, B. A. 1999. Stress, Duration and Intonation in Arabic
Word-level Prosody. Journal of Phonetics, 27.3 -22.
DE LACY, PAUL. 2002. The interaction of tone and stress in Optimality Theory. Phonology,
19.1-32.
DEVONISH, HUBERT. 1989. Talking in Tones: A Study of Tone in Afro-European Creole

Languages. London and Christ Church: Karia Press and Caribbean Academic
Publications.
—. 2002. Talking Rhythm Stressing Tones: The role of prominence in Anglo-West African
Creole Languages. Kingston: Arawak Publications.
—. 2003. Reduplication as Lexical and Syntactic Aspect Marking: The Case of Guyanese
Creole. Twice as Meaningful: Morphological Reduplication in Pidgins, Creoles and
other Contact Languages, ed. by Silvia Kouwenberg, 47-60. London: Battlebridge.
DEVONISH, HUBERT and SEILER, W. 1991. A Reanalysis of the Phonological System of

Jamaican Creole. Society for Caribbean Linguistics Occasional Paper, 24.
D'IMPERIO, MARIAPAOLA. 2000. The Role of Perception in Defining Tonal Targets and
Their Alignment, The Ohio State University: PhD Dissertation.
FERGUSON, CHARLES. 1959. Diglossia. Language in Social Context, ed. by P Giglioli, 232 -
51. Harmondsworth: Penguin.
- 291 -
FRY, DENNIS. 1955. Duration and Intensity as Physical Correlates of Linguistic Stress.
Journal of the Acoustic Society of America, 45.765 -68.
—. 1958. Experiments in Speech Perception. Language and Speech, 1.126-52.
—. 1965. The Dependence of Stress Judgements on Vowel Formant Structure. Proceedings

of the Fifth International Congress of Phonetic Science.306 -11.
GOLDSMITH, JOHN. 1990. Autosegmental and Metrical Phonology. Cambridge: Blackwell.
GOLDSMITH, JOHN. 1995. The Handbook of Phonology Theory. Cambridge, MA:

Blackwell.
GOOD, JEFF. 2003. Tone and accent in Saramaccan: Charting a deep split in the phonology
of a language. Unpublished Manuscript
GOODEN, SHELOME. 2002. Past Time Reference in Belizean Creole. UPenn Working
Papers in Linguistics: Selected Papers from NWAV 30, 8.
—. 2003a. Reduplication in Jamaican Creole: Semantic Functions and Prosodic

Constraints. Twice as Meaningful: Morphological Reduplication in Pidgins,
Creoles and other Contact Languages, ed. by Silvia Kouwenberg, 93-103. London:
Battlebridge Publications.
—.2003b. Prosodic Contrast in Jamaican Creole Reduplication. in Linguistische Arbeiten,

ed. by Ingo Plag, 193-208. Tübingen: Niemeyer.
GOODEN, SHELOME, KOUWENBERG, SILIVIA and LACHARITE, DARLENE. 2003. An Overview

of Jamaican Creole Reduplication. Twice as Meaningful: Morphological
Reduplication in Pidgins, Creoles and other Contact Languages, ed. by Silvia
Kouwenberg, 105-110. London: Battlebridge Publications.
HALLE, MORRIS and VERGNAUD, JEAN-ROGER.1987. An Essay on Stress. Cambridge: MIT

Press.
HAMMOND, MICHAEL. 1999. The Phonology of English: A Prosodic Optimality-Theoretic

Approach. Oxford: Oxford University Press.
HARRY, OTELEMATE G. and DEVONISH, HUBERT. 1998. The Suprasegmental System of

Berbice Dutch Creole. Paper presented at Society for Caribbean Linguistics, St.
Lucia.
HAYES, BRUCE. 1989. Compensatory Lengthening in Moraic Phonology. Linguistic

Inquiry, 20. 253 -306.
- 292 -
—. 1995. Metrical Stress Theory: Principles and Case Studies. Chicago and London: The
University of Chicago Press.
HOLM, JOHN. 1988. Pidgins and Creoles: Theory and Structure.Vol. 1. Cambridge:
HOPKINS, TOMETRO. 2000. Morphology in a Developing Creole. Paper presented at Society

for Caribbean Linguistics 13th Biennial Conference, UWI, Jamaica.
HUSS, V. 1978. English Word Stress in the Post-Nuclear Position. Phonetics.86 - 105.
HUTTAR, GEORGE and HUTTAR, MARY. 1997. Reduplication in Ndjuka. The Structure and
Status of Pidgins and Creoles (CLL 19), ed. by Arthur Spears and Donald Winford,
395 - 414. Amsterdam and Philadelphia: John Benjamins.
HYMAN, LARRY M. 1978. Tone and/or Accent. Elements of Tone, Stress an Intonation:
Papers presented at an interest group of the Georgetown University Round Table on
Languages and Linguistics, 1977, ed. by Donna J. Napoli, 1 - 20: Georgetown
University Press.
IRVINE, ALLISON. 1994. Dialect Variation in Jamaican English: A study of the phonology
of social group marking. English World Wide, 15.55 -78.
KAGER, RENÉ. 1995. The Metrical Theory of Word Stress. The Handbook of Phonological
Theory, ed. by John Goldsmith, 367 - 402. Oxford and Cambridge MA: Basil
Blackwell.
—. 1999. Optimality Theory: Cambridge Textbooks in Linguistics. Cambridge: Cambridge
University Press.
KATAMBA, FRANCIS. 1993. Morphology: Modern Linguistic Series. London: Macmillan

Press.
KAWASAKI, HARUKO. 1982. An Acoustic Basis for Universal Constraints on Sound

Sequences, University of California at Berkeley: PhD Dissertation.
KENSTOWICZ, MICHAEL. 1994. Phonology in Generative Grammar. Oxford and Cambridge

MA: Basil Blackwell.
KOUWENBERG, SILVIA and LACHARITÉ, DARLENE. 1998. Inflectional versus Non-iconic

Derivational Reduplication in Creole Languages: The Creolist Archives. Papers
Online.
—. 2001a. The Mysterious Case of yalayala. Due Respect. Papers on English and English-
Related Creoles in Honour of Robert B. LePage, ed. by Pauline Christie,124 -34.
Jamaica: University of the West Indies Press.
- 293 -
—. 2001b. The Iconic Interpretations of Reduplication: Issues in the Study of
Reduplication in Caribbean Creole Languages. European Journal of English
Studies, 5.59 - 80.
—. 2003. The Meanings of 'more of the same': Iconicity in reduplication and the Evidence
for Substrate Transfer in the Genesis of Caribbean Creole Languages. Twice as
Meaningful: Morphological Reduplication in Pidgins, Creoles and other Contact
Languages, ed. by Silvia Kouwenberg, 7 -18. London: Battlebridge Publications.
LADD, D.R. 1996. Intonational Phonology. Cambridge: Cambridge University Press.
LAWTON, DAVID. 1963. Suprasegmental Phenomena in Jamaican Creole, Department of

English, University of Michigan: Ph.D Thesis.
—. 1968. The Implications of Tone for Jamaican Creole. Anthropological Linguistics,

10.22-26.
LEFEBVRE, CLAIRE.1998. Creole Genesis and the Acquisition of Grammar: the case of
Haitian Creole. New York: Cambridge University Press.
LEHISTE, ISLE. 1961. Some Acoustic Correlates of Accent in Serbo-Croatian. Phonetica,

7.114-47.
LEPAGE, ROBERT B. 1960. An Historical Introduction to Jamaican Creole. Jamaican

Creole: Creole Language Studies I, ed. by Robert B. LePage and David DeCamp, 1
- 124. London:Macmillan.
LIBERMAN, MARK and PRINCE, ALAN. 1977. On Stress and Linguistic Rhythm. Linguistic
Inquiry, 8. 249 - 336.
LIEBERMAN, P. 1960. Some Acoustic Correlates of Word Stress in American English.

Journal of the Acoustic Society of America, 32.451-54.
MAIDMENT, J.A. 1989. On the Accentuation of Compounds in English. Speech, Hearing

and Language:Work in Progress (University College of London), 3.181 -88.
MARANTZ, ALEC. 1982. Re Reduplication. Linguistic Inquiry, 13.435-82.
MCCAIN, LOIS. 1996. Caribbean Studies Thesis. University of the West Indies Mona.
Unpublished Undergraduate Thesis.
MCCARTHY, JOHN.1982. Prosodic Templates, Morphemic Templates and Morphemic

Tiers. The Structure of Phonological Representation Part I, ed. by Van der Hulst
and Norval Smith. Dordrecht: Foris.
- 294 -
MCCARTHY, JOHN. 2002. A Thematic Guide to Optimality Theory: Cambridge University
Press.
MCCARTHY, JOHN and PRINCE, ALAN. 1993. Prosodic Morphology I: Constraint Interaction
and Satisfaction: University of Massachusetts at Amherst and Rutgers University.
—. 1995a. Prosodic Morphology. Handbook of Phonological Theory, ed. by John

Goldsmith. Cambridge MA and Oxford UK: Blackwell Publishers.
—. 1995b. Faithfulness and Reduplicative Identity. Proceedings of the Urecht Workshop

on Prosodic Morphology, ed. by René Kager, H van der Hulst and W Zonneveld.
The Hague: Mouton.
MEADE, ROCKY. 1996. On the Phonology and Orthography of Jamaican Creole. Journal of
Pidgin and Creole Languages, 11.335-41.
—. 2001. Acquisition of Jamaican phonology: LOT dissertations 41. Delft: De Systeem

Drukkers.
MIGGE, BETTINA. 1998. Substrate influence in the formation of the Surinamese Plantation
Creole : a consideration of sociohistorical data and linguistic data from Ndyuka and
Gbe, The Ohio State University: PhD Dissertation.
MIGGE, BETTINA. 2003. The Origin of Predicate Reduplication in Suriname Eastern

Maroon Creole. Twice as Meaningful: Morphological Reduplication in Pidgins,
Creoles and other Contact Languages, ed. by Silvia Kouwenberg, 61 -71. London:
MILROY, LESLEY. 1980. Language and Social Networks. Oxford: Blackwell.
MORAVSCIK, EDITH. 1978. Reduplicative Constructions. Universals of Human Language,

ed. by Joseph Greenberg. California: Stanford University Press.
NESPOR, MARINA and VOGEL, IRENE. 1986. Prosodic Phonology. Dordrecht: Foris.
NYLANDER, DUDLEY. 1999. Reduplication in Krio. Paper presented at 4th Westminster

Creolistics Workshop on Reduplication in Contact Languages, London.
OHALA, JOHN. 1993. The Perceptual Basis of Some Sound Patterns. Papers in Laboratory
Phonology, Vol. 1: Between the grammar and the physics of speech. ed. by D.
A.Connell and Amalia Arvaniti, 87 -94.Cambridge: Cambridge University Press.
PATRICK, PETER L. 1999. Urban Jamaican creole : variation in the mesolect. Amsterdam ;
Philadelphia: J. Benjamins Publishing Company.
- 295 -
PLAG, INGO and UFFMAN, CHRISTIAN. 2000. Phonological Restructuring in Creole: The
development of paragoge in Sranan. Degrees of Restructuring in Creole Languages
(CLL 22), ed. by Neumann-Holzchuh and Edgar Schneider, 309 -36. Amsterdam:
John Benjamins.
POLETTO, ROBERT. 1998. Topics in Runyankore Phonology, The Ohio State University:
PhD Dissertation.
PRINCE, ALAN. 1990. Quantitative Consequences of Rhythmic Organization. CLS 26:

Papers from the Parasession on the Syllable in Phonetics and Phonology, ed. by
Michael Ziolkowski, Manuela Noske and Karen Deaton, 355-98. Chicago: Chicago
Linguistic Society.
PRINCE, ALAN and SMOLENKSY, PAUL. 1993 (ROA Version 8/2002). Optimality Theory:
Constraints in Generative Grammar.
REINECKE, JOHN and TOKIMASA, AIKO. 1934. The English Dialect of Hawaii. American
Speech, 9. 48 -58.
REMIJSEN, ALBERT C.L. 2001. Word-prosodic Systems of Raja Ampat Languages: LOT.
RIVERA-CASTILLO, YOLANDA. 1998. Tone and Stress in Papiamentu: the contribution of a

constraint-based analysis to the problem of creole genesis. Journal of Pidgin and
Creole Languages, 13.297-335.
ROSENTHALL, SAM.1994. Vowel/Glide Alternation in a Theory of Constraint Interaction,

University of Massachusetts, Amherst: PhD Dissertation.
SCOTT, N.C. 1939. An Experiment on Stress Perception. Le Maître Phonétique, 67.44-45.
SEBBA, MARK. 1981. Derivational Regularities in a Creole Lexicon: the case of Sranan.
Linguistics, 19.101 -17.
SELKIRK, E. 1984. Phonology and Syntax: the relationship between sound and structure.
Cambridge, MA: MIT Press.
SHATTUCK-HUFNAGEL, OSTENDORF, M and ROSS, K. 1994. Stress Shift and Early Pitch
Accent Placement in Lexical Items in American English. Journal of Phonetics,
22.357- 88.
SMITH, NORVAL.1987. The Genesis of the Creole Languages of Surinam, University of

Amsterdam: PhD Dissertation.
SPENCER, ANDREW. 1991. Chapter 5: Non-Linear Approaches to Morphology.

Morphological Theory. Cambridge MA and Oxford UK: Blackwell Publishers.
- 296 -
STATIN. 2002. Population Census 2001: Preliminary report. Kingston, Jamaica: Statistical
Institute of Jamaica.
STERIADE, DONCA. 1988. Reduplication and syllable transfer in Sanskrit and elsewhere.
Phonology.73-155.
SUTCLIFFE, DAVID. 1986. Jamaican Creole Tonality and the Implications for Syntactic
Analysis. Society for Caribbean Linguistics Occasional Paper, 2.
SYLVAIN, S. 1936. Le Creole Haitien: Morphologie et Syntax. Port-au-Prince (Haiti)

L'auteur: Wetern (Belgique) Imprimerie De Meester.
URBANCZYK, S. 2001. Patterns of Reduplication in Lushootseed. New York: Garland.
VAN COETSEM, FRANS. 1996. Towards a Typology of Lexical Accent: 'Stress Accent and
'Pitch Accent' in a Renewed Perspective. Heidelberg: Universitätsverlag C. Winter.
VERHOEVEN. 1994. The Discrimination of Pitch Movement Alignment in Dutch. Journal

of Phonetics, 22.65 - 85.
VOGEL, IRENE and RAIMY, ERIC. 2002. The acquisition of compound vs. phrasal stress: the
role of prosodic constituents. Journal of Child Language, 29.225-50.
WELLS, J.C. 1973. Jamaican Pronunciation in London: Publications of the Philological

Society XXV. Oxford: Basil Blackwell.
WHITTLE, D. 1989. The Syllable structure of Jamaican Creole, Linguistics, University of

the West Indies: Undergraduate Thesis.
WILBUR, RONNIE. 1973. The Phonology of Reduplication: University of Rochester

Dissertation: Reproduced by University of Indiana Linguistics Club.
WINFORD, DONALD. 1997. Re-examining Caribbean English Creole Continua. World

Englishes, 16. 233 - 279.
WINFORD, DONALD. 1997. Property Items and Predication in Sranan. Journal of Pidgin and
Creole Languages, 12.237 - 301.
- 297 -
APPENDIX A
Informant Data Sheet

Tape #: _____________
Name: _____________________________
Age: ___________
Occupation (or parent for minors):_____________________________
Education (highest level):_____________________________________
Places of residence:
Present:______________________ from 19_______ to ________
Past: _________________________ from 19_______ to ________
Languages spoken:____________________________________________
Language used most often:______________________________________
- in general:____________________________
- at home:______________________________
- talking about important things (school, business, etc):__________
- talking about family matters (quarrels, etc):___________________
- talking to children:________________________________________
- for gossiping, jokes, secrets etc:____________________________
- among close friends:____________________________
- in the community/neighborhood:__________________________
- Any other situations in which these languages are typically used
- 285 -
APPENDIX B
The following graphs show representative examples of the monosyllabic based
reduplicated word gringrin, produced with an intensive meaning and a distributive meaning
by RP.
greenDISdf4 RP greenINdf1 RP
150
125
100
75
4000
2000
0
tones L% H+L*L%
phones g ® i n g ® i g ® i n g ® i n
words mi want di wan dem we griingriin mi want di mango dem we greengreen
0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6

green’ (right graph) in statement intonation final position.
- 286 -
greenDISdnf3 RP greenINdnf1 RP
150
125
100
75
4000
2000
0
tones L% H+L* L%
words mi want di griingriin mango dem want di griingriin mango dem
0.4 0.8 1.2 0.4 0.8 1.2

Figure B.2. Fundamental frequency contour of the reduplicated word gringrin

produced with a distributive meaning ‘scattered green (left graph) and an intensive
meaning ‘very green’ (right graph) in statement intonation non-final position.
greenDISynf2 RP greenINynf2 RP
150
125
100
75
4000
2000
0
tones H% H* H%
words yu want di wan dem we griingriin yu want di mango dem we griingriin
0.4 0.8 1.2 0.4 0.8 1.2

green’ (right graph) in question intonation final position.
- 287 -
greenDISynnf4 RP greenINynnf1 RP
150
125
100
75
4000
2000
0
tones H% H+L* H%
words yu want di griingriin mango dem yu want di griingriin mango dem
0.4 0.8 1.2 0.4 0.8 1.2

green’ (right graph) in question intonation non-final position.
- 288 -
BIBLIOGRAPHY
ADAMSON, LILLIAN and SMITH, NORVAL. 1999. Derivational Reduplication in Sranan.

Paper presented at 4th Westminster Creolistics Workshop on Reduplication in
Contact Languages, April 8th -10th, London.
—. 2003. ed. by Silvia Kouwenberg: Battlebridge Publications.
AKERS, GLEN. 1981. Phonological Variation in the Jamaican Continuum. Ann Arbor:
Karoma.
ALDERETE, JOHN. 1993. The Prosodic Morphology of Jamaican Creole Iteratives.

University of Massachusetts Occasional Papers 20, the UMOP in Indigenous
Languages, ed. by Elena Benedicto, 29-50. Amherst: Graduate Linguistic Student
Association.
ALLEYNE, MERVYN. 1980. Comparative Afro-American: A Historical Comparative Study

of English-based Afro-American Dialects of the New World. Ann Arbor: Karoma.
ALLEYNE, MERVYN. 1987. Predicate Structures in Saramaccan. Studies in Saramaccan

Language Structure, ed. by Mervyn Alleyne, 71 -87. Amsterdam/Jamaica:
University of Amsterdam/University of the West Indies.
ALLSOPP, RICHARD. 1996. Dictionary of Caribbean English Usage. Oxford: Oxford

University Press.
ARENDS, JACQUES. 1995. The early stages of creolization. Amsterdam ; Philadelphia: John
Benjamins Publishing Company.
ARCHANGELI, DIANA and LANGENDOEN, D.TERENCE (eds.) 1997. Optimality theory: an

overview. Explaining Linguistics. Oxford: Blackwell.
ARVANITI, AMALIA. 1992. Secondary Stress: evidence from Modern Greek. Papers in
Laboratory Phonology II: Gesture, Segment, Prosody, ed. by G. Docherty and D.
Ladd. Cambridge: CUP.
- 289 -
BAKER, PHILIP. 1999. Reduplication in Mauritian Creole. Paper presented at 4th
Westminster Creolistics Workshop on Reduplication in Contact Languages,
London.
BAKKER, PETER. 1987. Reduplications in Saramaccan. Studies in Saramaccan Language
Structure: Caribbean Culture Studies 2, ed. by Mervyn Alleyne, 17 - 40.
BECKFORD-WASSINK, ALICIA. 1999. A Sociophonetic Study of Jamaican Vowels,

Linguistics, University of Michigan: PhD Dissertation.
—. 2001. Theme and variation in Jamaican Vowels. Language Variation and Change,
13.135-59.
BECKMAN, MARY E. 1984. Toward Phonetic Criteria for a Typology of Lexical Accent,
Department of Linguistics, Cornell University: PhD Dissertation.
—. 1986. Stress and Non-Stress Accent: Netherlands Phonetic Archives. Dordrecht -

Holland/ Riverton -USA: Foris Publications.
—. 1996. The Parsing of Prosody. Language and Cognitive Processes, 11.17 -67.
BECKMAN, MARY E. and PIERREHUMBERT, JANET. 1986. Intonational Structure in Japanese

and English. Phonology Yearbook 3, 255 -310.
BROUSSEAU, ANNE-MARIE. 2003. The Accentual System of Haitian Creole: The Role of
Transfer and Markedness Values. To appear in Linguistische Arbeiten, ed. by Ingo
Plag: Niemeyer Publishers.
BROSELOW, E and MCCARTHY, JOHN. 1983. A Theory of Internal Reduplication. The

Linguistic Review, 3.25 - 88.
BRUCE, GOSTA. 1977. Swedish Word Accents in Sentence Perspective: Travaux de

L'institute de Linguistique de Lund 12: CWK Gleerup.
BUCKLEY, EUGENE. 1997. Alignment in Manam Stress. Linguistic Inquiry.
CARTER, HAZEL. 1980. How to be a Tone Language: Theoretical Considerations involved

in the Classification of Creole Languages as Tonal (or otherwise). Paper presented
at Society for Caribbean Linguistics, Aruba.
—. 1983. How to be a Tone Language: Theoretical Considerations involved in the

Classification of Creole Languages as Tonal (or otherwise). Studies in Caribbean
Language, ed. by Lawrence Carrington, 90-111. Society for Caribbean Linguistics.
—. 1987. Suprasegmentals in Guyanese: some African comparisons. Pidgin and Creole

Languages, ed. by Glen Gilbert, 213 - 63. Honolulu: University of Hawaii Press.
- 290 -
CASSIDY, FREDERICK G. 1957. Iteration as a Word Formation Device in Jamaican Folk
Speech. American Speech, 32. 49 -53.
CASSIDY, FREDERICK G. 1961. Jamaica Talk: Three Hundred Years of the English
Language in Jamaica. London: Macmillan.
CASSIDY, FREDERICK G and LEPAGE, ROBERT. 1967. Dictionary of Jamaican English.

Cambridge: University of Cambridge Press. [Reprinted in 1980].
CLEMENTS, G.N. 1985. The Problem of Transfer in Non-Linear Morphology. Cornell

Working Papers in Linguistics, 7.38 - 73.
CRYSTAL, DAVID. 1997. A Dictionary of Linguistics and Phonetics: Basil Blackwell.
DECAMP, DAVID. 1971. Towards a Generative Analysis of a Post-Creole Continuum.

Piginization and Creolization of Languages, ed. by Dell Hymes, 349 - 70.
Cambridge: Cambridge University Press.
DECAMP, DAVID.1974. Neutralization, Iteratives and Ideophones: the locus of language in

Jamaica. Pidgins and Creoles: current trends and prospects, ed. by David DeCamp
and Ian Hancock, 46 -50. Georgetown University Press.
DE JONG, KENNETH and ZAWAYDEH, B. A. 1999. Stress, Duration and Intonation in Arabic
Word-level Prosody. Journal of Phonetics, 27.3 -22.
DE LACY, PAUL. 2002. The interaction of tone and stress in Optimality Theory. Phonology,
19.1-32.
DEVONISH, HUBERT. 1989. Talking in Tones: A Study of Tone in Afro-European Creole

Languages. London and Christ Church: Karia Press and Caribbean Academic
Publications.
—. 2002. Talking Rhythm Stressing Tones: The role of prominence in Anglo-West African
Creole Languages. Kingston: Arawak Publications.
—. 2003. Reduplication as Lexical and Syntactic Aspect Marking: The Case of Guyanese
Creole. Twice as Meaningful: Morphological Reduplication in Pidgins, Creoles and
other Contact Languages, ed. by Silvia Kouwenberg, 47-60. London: Battlebridge.
DEVONISH, HUBERT and SEILER, W. 1991. A Reanalysis of the Phonological System of

Jamaican Creole. Society for Caribbean Linguistics Occasional Paper, 24.
D'IMPERIO, MARIAPAOLA. 2000. The Role of Perception in Defining Tonal Targets and
Their Alignment, The Ohio State University: PhD Dissertation.
- 291 -
FERGUSON, CHARLES. 1959. Diglossia. Language in Social Context, ed. by P Giglioli, 232 -
51. Harmondsworth: Penguin.
FRY, DENNIS. 1955. Duration and Intensity as Physical Correlates of Linguistic Stress.
Journal of the Acoustic Society of America, 45.765 -68.
—. 1958. Experiments in Speech Perception. Language and Speech, 1.126-52.
—. 1965. The Dependence of Stress Judgements on Vowel Formant Structure. Proceedings

of the Fifth International Congress of Phonetic Science.306 -11.
GOLDSMITH, JOHN. 1990. Autosegmental and Metrical Phonology. Cambridge: Blackwell.
GOLDSMITH, JOHN. 1995. The Handbook of Phonology Theory. Cambridge, MA:

Blackwell.
GOOD, JEFF. 2003. Tone and accent in Saramaccan: Charting a deep split in the phonology
of a language. Unpublished Manuscript
GOODEN, SHELOME. 2002. Past Time Reference in Belizean Creole. UPenn Working
Papers in Linguistics: Selected Papers from NWAV 30, 8.
—. 2003a. Reduplication in Jamaican Creole: Semantic Functions and Prosodic

Constraints. Twice as Meaningful: Morphological Reduplication in Pidgins,
Creoles and other Contact Languages, ed. by Silvia Kouwenberg, 93-103. London:
—.2003b. Prosodic Contrast in Jamaican Creole Reduplication. in Linguistische Arbeiten,

ed. by Ingo Plag, 193-208. Tübingen: Niemeyer.
GOODEN, SHELOME, KOUWENBERG, SILIVIA and LACHARITE, DARLENE. 2003. An Overview

of Jamaican Creole Reduplication. Twice as Meaningful: Morphological
Reduplication in Pidgins, Creoles and other Contact Languages, ed. by Silvia
Kouwenberg, 105-110. London: Battlebridge Publications.
HALLE, MORRIS and VERGNAUD, JEAN-ROGER.1987. An Essay on Stress. Cambridge: MIT

Press.
HAMMOND, MICHAEL. 1999. The Phonology of English: A Prosodic Optimality-Theoretic

Approach. Oxford: Oxford University Press.
HARRY, OTELEMATE G. and DEVONISH, HUBERT. 1998. The Suprasegmental System of

Berbice Dutch Creole. Paper presented at Society for Caribbean Linguistics, St.
Lucia.
- 292 -
HAYES, BRUCE. 1989. Compensatory Lengthening in Moraic Phonology. Linguistic
Inquiry, 20. 253 -306.
—. 1995. Metrical Stress Theory: Principles and Case Studies. Chicago and London: The
University of Chicago Press.
HOLM, JOHN. 1988. Pidgins and Creoles: Theory and Structure.Vol. 1. Cambridge:
HOPKINS, TOMETRO. 2000. Morphology in a Developing Creole. Paper presented at Society

for Caribbean Linguistics 13th Biennial Conference, UWI, Jamaica.
HUSS, V. 1978. English Word Stress in the Post-Nuclear Position. Phonetics.86 - 105.
HUTTAR, GEORGE and HUTTAR, MARY. 1997. Reduplication in Ndjuka. The Structure and
Status of Pidgins and Creoles (CLL 19), ed. by Arthur Spears and Donald Winford,
395 - 414. Amsterdam and Philadelphia: John Benjamins.
HYMAN, LARRY M. 1978. Tone and/or Accent. Elements of Tone, Stress an Intonation:
Papers presented at an interest group of the Georgetown University Round Table on
Languages and Linguistics, 1977, ed. by Donna J. Napoli, 1 - 20: Georgetown
University Press.
IRVINE, ALLISON. 1994. Dialect Variation in Jamaican English: A study of the phonology
of social group marking. English World Wide, 15.55 -78.
KAGER, RENÉ. 1995. The Metrical Theory of Word Stress. The Handbook of Phonological
Theory, ed. by John Goldsmith, 367 - 402. Oxford and Cambridge MA: Basil
Blackwell.
—. 1999. Optimality Theory: Cambridge Textbooks in Linguistics. Cambridge: Cambridge
University Press.
KATAMBA, FRANCIS. 1993. Morphology: Modern Linguistic Series. London: Macmillan

Press.
KAWASAKI, HARUKO. 1982. An Acoustic Basis for Universal Constraints on Sound

Sequences, University of California at Berkeley: PhD Dissertation.
KENSTOWICZ, MICHAEL. 1994. Phonology in Generative Grammar. Oxford and Cambridge

MA: Basil Blackwell.
KOUWENBERG, SILVIA and LACHARITÉ, DARLENE. 1998. Inflectional versus Non-iconic

Derivational Reduplication in Creole Languages: The Creolist Archives. Papers
Online.
- 293 -
—. 2001a. The Mysterious Case of yalayala. Due Respect. Papers on English and English-
Related Creoles in Honour of Robert B. LePage, ed. by Pauline Christie,124 -34.
Jamaica: University of the West Indies Press.
—. 2001b. The Iconic Interpretations of Reduplication: Issues in the Study of
Reduplication in Caribbean Creole Languages. European Journal of English
Studies, 5.59 - 80.
—. 2003. The Meanings of 'more of the same': Iconicity in reduplication and the Evidence
for Substrate Transfer in the Genesis of Caribbean Creole Languages. Twice as
Meaningful: Morphological Reduplication in Pidgins, Creoles and other Contact
Languages, ed. by Silvia Kouwenberg, 7 -18. London: Battlebridge Publications.
LADD, D.R. 1996. Intonational Phonology. Cambridge: Cambridge University Press.
LAWTON, DAVID. 1963. Suprasegmental Phenomena in Jamaican Creole, Department of

English, University of Michigan: Ph.D Thesis.
—. 1968. The Implications of Tone for Jamaican Creole. Anthropological Linguistics,

10.22-26.
LEFEBVRE, CLAIRE.1998. Creole Genesis and the Acquisition of Grammar: the case of
Haitian Creole. New York: Cambridge University Press.
LEHISTE, ISLE. 1961. Some Acoustic Correlates of Accent in Serbo-Croatian. Phonetica,

7.114-47.
LEPAGE, ROBERT B. 1960. An Historical Introduction to Jamaican Creole. Jamaican

Creole: Creole Language Studies I, ed. by Robert B. LePage and David DeCamp, 1
- 124. London:Macmillan.
LIBERMAN, MARK and PRINCE, ALAN. 1977. On Stress and Linguistic Rhythm. Linguistic
Inquiry, 8. 249 - 336.
LIEBERMAN, P. 1960. Some Acoustic Correlates of Word Stress in American English.

Journal of the Acoustic Society of America, 32.451-54.
MAIDMENT, J.A. 1989. On the Accentuation of Compounds in English. Speech, Hearing

and Language:Work in Progress (University College of London), 3.181 -88.
MARANTZ, ALEC. 1982. Re Reduplication. Linguistic Inquiry, 13.435-82.
MCCAIN, LOIS. 1996. Caribbean Studies Thesis. University of the West Indies Mona.
Unpublished Undergraduate Thesis.
- 294 -
MCCARTHY, JOHN.1983. Prosodic Templates, Morphemic Templates and Morphemic
Tiers. The Structure of Phonological Representation Part I, ed. by Van der Hulst
and Norval Smith. Dordrecht: Foris.
MCCARTHY, JOHN. 2002. A Thematic Guide to Optimality Theory: Cambridge University

Press.
MCCARTHY, JOHN and PRINCE, ALAN. 1993. Prosodic Morphology I: Constraint Interaction
and Satisfaction: University of Massachusetts at Amherst and Rutgers University.
—. 1995a. Prosodic Morphology. Handbook of Phonological Theory, ed. by John

Goldsmith. Cambridge MA and Oxford UK: Blackwell Publishers.
—. 1995b. Faithfulness and Reduplicative Identity. Proceedings of the Urecht Workshop

on Prosodic Morphology, ed. by René Kager, H van der Hulst and W Zonneveld.
The Hague: Mouton.
MEADE, ROCKY. 1996. On the Phonology and Orthography of Jamaican Creole. Journal of
Pidgin and Creole Languages, 11.335-41.
—. 2001. Acquisition of Jamaican phonology: LOT dissertations 41. Delft: De Systeem

Drukkers.
MIGGE, BETTINA. 1998. Substrate influence in the formation of the Surinamese Plantation
Creole : a consideration of sociohistorical data and linguistic data from Ndyuka and
Gbe, The Ohio State University: PhD Dissertation.
MIGGE, BETTINA. 2003. The Origin of Predicate Reduplication in Suriname Eastern

Maroon Creole. Twice as Meaningful: Morphological Reduplication in Pidgins,
Creoles and other Contact Languages, ed. by Silvia Kouwenberg, 61 -71. London:
MILROY, LESLEY. 1980. Language and Social Networks. Oxford: Blackwell.
MORAVSCIK, EDITH. 1978. Reduplicative Constructions. Universals of Human Language,

ed. by Joseph Greenberg. California: Stanford University Press.
NESPOR, MARINA and VOGEL, IRENE. 1986. Prosodic Phonology. Dordrecht: Foris.
NYLANDER, DUDLEY. 1999. Reduplication in Krio. Paper presented at 4th Westminster

Creolistics Workshop on Reduplication in Contact Languages, London.
OHALA, JOHN. 1993. The Perceptual Basis of Some Sound Patterns. Papers in Laboratory
Phonology, Vol. 1: Between the grammar and the physics of speech. ed. by D.
A.Connell and Amalia Arvaniti, 87 -94.Cambridge: Cambridge University Press.
- 295 -
PATRICK, PETER L. 1999. Urban Jamaican creole : variation in the mesolect. Amsterdam ;
Philadelphia: J. Benjamins Publishing Company.
PLAG, INGO and UFFMAN, CHRISTIAN. 2000. Phonological Restructuring in Creole: The
development of paragoge in Sranan. Degrees of Restructuring in Creole Languages
(CLL 22), ed. by Neumann-Holzchuh and Edgar Schneider, 309 -36. Amsterdam:
John Benjamins.
POLETTO, ROBERT. 1998. Topics in Runyankore Phonology, The Ohio State University:
PhD Dissertation.
PRINCE, ALAN. 1990. Quantitative Consequences of Rhythmic Organization. CLS 26:

Papers from the Parasession on the Syllable in Phonetics and Phonology, ed. by
Michael Ziolkowski, Manuela Noske and Karen Deaton, 355-98. Chicago: Chicago
Linguistic Society.
PRINCE, ALAN and SMOLENKSY, PAUL. 1993 (ROA Version 8/2002). Optimality Theory:
Constraints in Generative Grammar.
REINECKE, JOHN and TOKIMASA, AIKO. 1934. The English Dialect of Hawaii. American
Speech, 9. 48 -58.
REMIJSEN, ALBERT C.L. 2001. Word-prosodic Systems of Raja Ampat Languages: LOT.
RIVERA-CASTILLO, YOLANDA. 1998. Tone and Stress in Papiamentu: the contribution of a

constraint-based analysis to the problem of creole genesis. Journal of Pidgin and
Creole Languages, 13.297-335.
ROSENTHALL, SAM.1994. Vowel/Glide Alternation in a Theory of Constraint Interaction,

University of Massachusetts, Amherst: PhD Dissertation.
SCOTT, N.C. 1939. An Experiment on Stress Perception. Le Maître Phonétique, 67.44-45.
SEBBA, MARK. 1981. Derivational Regularities in a Creole Lexicon: the case of Sranan.
Linguistics, 19.101 -17.
SELKIRK, E. 1984. Phonology and Syntax: the relationship between sound and structure.
Cambridge, MA: MIT Press.
SHATTUCK-HUFNAGEL, OSTENDORF, M and ROSS, K. 1994. Stress Shift and Early Pitch
Accent Placement in Lexical Items in American English. Journal of Phonetics,
22.357- 88.
SMITH, NORVAL.1987. The Genesis of the Creole Languages of Surinam, University of

Amsterdam: PhD Dissertation.
- 296 -
SPENCER, ANDREW. 1991. Chapter 5: Non-Linear Approaches to Morphology.
Morphological Theory. Cambridge MA and Oxford UK: Blackwell Publishers.
STATIN. 2002. Population Census 2001: Preliminary report. Kingston, Jamaica: Statistical
Institute of Jamaica.
STERIADE, DONCA. 1988. Reduplication and syllable transfer in Sanskrit and elsewhere.
Phonology.73-155.
SUTCLIFFE, DAVID. 1986. Jamaican Creole Tonality and the Implications for Syntactic
Analysis. Society for Caribbean Linguistics Occasional Paper, 2.
SYLVAIN, S. 1936. Le Creole Haitien: Morphologie et Syntax. Port-au-Prince (Haiti)

L'auteur: Wetern (Belgique) Imprimerie De Meester.
URBANCZYK, S. 2001. Patterns of Reduplication in Lushootseed. New York: Garland.
VAN COETSEM, FRANS. 1996. Towards a Typology of Lexical Accent: 'Stress Accent and
'Pitch Accent' in a Renewed Perspective. Heidelberg: Universitätsverlag C. Winter.
VERHOEVEN. 1994. The Discrimination of Pitch Movement Alignment in Dutch. Journal

of Phonetics, 22.65 - 85.
VOGEL, IRENE and RAIMY, ERIC. 2002. The acquisition of compound vs. phrasal stress: the
role of prosodic constituents. Journal of Child Language, 29.225-50.
WELLS, J.C. 1973. Jamaican Pronunciation in London: Publications of the Philological

Society XXV. Oxford: Basil Blackwell.
WHITTLE, D. 1989. The Syllable structure of Jamaican Creole, Linguistics, University of

the West Indies: Undergraduate Thesis.
WILBUR, RONNIE. 1973. The Phonology of Reduplication: University of Rochester

Dissertation: Reproduced by University of Indiana Linguistics Club.
WINFORD, DONALD. 1997. Re-examining Caribbean English Creole Continua. World

Englishes, 16. 233 - 279.
WINFORD, DONALD. 1997. Property Items and Predication in Sranan. Journal of Pidgin and
Creole Languages, 12.237 - 301.
- 297 -

i the Phonology and Phonetics of Jamaican Creole ( Pdfdrive )

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

i the Phonology and Phonetics of Jamaican Creole ( Pdfdrive )

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

i the Phonology and Phonetics of Jamaican Creole ( Pdfdrive )

Uploaded by

Copyright:

Available Formats

THE PHONOLOGY AND PHONETICS OF

JAMAICAN CREOLE REDUPLICATION

Presented in Partial Fulfillment of the Requirements for

Shelome A. Gooden, B.A., M.A.

The Ohio State University

Dissertation Committee: Approved by

Professor Donald C. Winford, Co-adviser ________________________

Professor Mary M. Beckman

This dissertation is an extensive treatment of the phonological and phonetic

properties of Jamaican Creole (JC) reduplication. While reduplication is thought to be a

has been done on the phonetic or morpho-phonological properties of the process.

Complementing the analysis of reduplication developed in this work, is an analysis of the

a stress-based system in which lexical contrasts are signaled by differences in the

alignment of the F0 contour with the word.

Reduplication processes in JC are similar in form and semantics to those found

Optimality Theoretic perspective. The phonological aspect focused on delimiting the

constraints on the segmental properties of reduplication processes. The observation is

The phonetic aspect investigated how the phonological constraints on

reduplication interact with the phonetic properties of reduplicated words. Particular

segmentally identical words. Specifically, it is shown that distributive reduplications

nuclear pitch accent.

of reduplication in Jamaican Creole. It also contributes to our understanding of the

relationship between word-level prosody and the semantic functions of reduplicated

research on this project.

home and helping me find ‘hard to find’ books and manuscripts.

My fieldtrips to Jamaica were supported by the Department of Linguistics, the Office

share of server space to store my many sound files.

forums for discussing my work.

The incessant prayers of my mother, Hazel, and my grandmother, Agatha, have

and encouraged me to pursue my goals.

September 6th, 1973 ......................................Born - Clarendon, Jamaica W.I.

1993.................................................................B.A. (Honors) Linguistics,

University of the West Indies,

2002.................................................................M.A. Linguistics, Ohio State University

1997- 2001.......................................................Graduate Teaching and Research Associate

The Ohio State University

2001 - 2003......................................................Editorial Associate

Journal of Pidgin and Creole Languages

2. (2003a). Reduplication in Jamaican Creole: Semantic Functions and Prosodic

4. (2003c). with D. Winford, H. Dawson and R. Dodsworth. OSU Working Papers in

Major Field: Linguistics

Minor Fields: Phonology, Creole Linguistics, Sociolinguistics

List of Tableau ..................................................................................................................xv

3. Jamaican Creole in its Sociocultural Context................................................................62

4. Word Prosody: Evidence for JC as a Stress-Based System………………………….....78

5 Word Level Stress in Jamaican Creole………….……………………………………151

7. An Optimality Theoretic Account of JC Reduplication............................................226

7.1. Introduction.......................................................................................................... 226

8. Conclusion .............................................................................................................. 279

A. Sample Informant Data Sheet................................................................................285

3.1. Sources of Jamaican population (Lalla and D’Costa 1990:4) ................................... 64

3.2. Map of Jamaica Showing the Fourteen Electoral Parishes........................................ 72

3.1: Informant Demographic Data .........................................................................................75

4.1: JC Consonant Inventory ..................................................................................................82

4.2: JC Vowel Inventory.........................................................................................................82

4.3: Summary of F0 Patterns……………………………………………………………..115

4.4: Elicited Words used in Analysis...................................................................................128

4.5: Prosodic Conditions for Target Words.........................................................................128