University of Ghana http://ugspace.ug.edu.gh THE PHONOLOGY OF AKAN LOANWORDS IN GA AND DANGME BY KWASI ADOMAKO (10050871) THIS THESIS/DISSERTATION IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF PHD IN LINGUISTICS DEGREE JULY 2018 University of Ghana http://ugspace.ug.edu.gh i DECLARATION I do declare that this thesis is the result of my own original research, and has not been presented either in whole or in part for award of any degree elsewhere. References to other sources of information used in this work have been duly acknowledged. All errors and mistakes found are solely mine. Dr. George Akanlig-Pare PRINCIPAL SUPERVISOR DATE University of Ghana http://ugspace.ug.edu.gh ii DEDICATION I dedicate this work to my entire family, especially my wife, Grace and my sons Amos and Gabriel, for the support, patience, and prayers throughout the years I have spent on my education all up-and-coming researchers in Ghanaian languages. University of Ghana http://ugspace.ug.edu.gh iii ACKNOWLEDGEMENT My foremost gratitude goes to the Allmektige Gud (Almighty God) for the knowledge, wisdom, strength, and life He has given me. It has been by His grace that I have come this far. I also wish to express my profound gratitude to my indefatigable supervisors; Dr. George Akanlig-Pare (my principal supervisor), Dr. Clement Kwamina Insaidoo Appah (both in the Department of Linguistics, University of Ghana, Legon), and Dr. Regina Oforiwah Caesar (Department of Ga-Dangme, University of Education, Winneba). I appreciate your time, patience, and guidance throughout the process of writing this thesis. You put your other equally important appointments and tasks aside and made time for me all the time. May the Almighty God bless you! Through the past four years, I have been under the tutelage of other Faculty members of the Department of Linguistics and School of Languages, University of Ghana, Legon including Prof. Nana Aba Amfo, Prof. Helen Lauer, Dr. Seth Antwi Ofori, Dr. Fusheini Hudu, and Dr. Kofi Dorvlo. I appreciate the knowledge and guidance given me. Prof. Kofi Agyekum and Prof. Kofi K. Saah have also been of immense support and inspiration to me. I am very grateful to you, Profs. I also thank Dr. Evershed K. Amuzu (immediate past HoD, Department of Linguistics) and Dr. James A. N. Saanchi (current HoD, Department of Linguistics) for their support throughout the years I spent in the Department pursuing this doctoral programme. I also wish to express my profound gratitude to my employers, University of Education, Winneba for granting leave and sponsoring my studies, and also granting me permission for the short visit to the University of Cologne, Germany. I University of Ghana http://ugspace.ug.edu.gh iv am grateful for all the support from the staff of my department; Department of Akan-Nzema Education, College of Languages Education, Ajumako. Special thanks also goes to my language consultants; Dr. Mrs. Yvonne A. Ollennu (for Ga), and Mr. Francis Tsatsu Owulah (for Dangme). I extend the same appreciation to all the other teaching staff of the Department of Ga-Dangme Education for the assistance. And to my Ga and Dangme student respondents for this study, I express my profound gratitude for your time during the data recordings sessions. I wish to thank the Federal Republic of Germany government for granting me a research grant (Short-term grants ID # 57314022) sponsored by the German Academic Exchange Service (Deutscher Akademischer Austauschdienst (DAAD)) as a guest scholar at the University of Cologne, Germany. Through this opportunity, I was able to establish important contacts, which have positively impacted on the writing of this thesis. Special mentioning goes to Prof. Dr. Frank Kügler, University of Cologne (Currently at the Johann Wolfgang Goethe University, Frankfurt) and Jonas Heinen, his assistant, who hosted me while at the University of Cologne. I was guided by Prof. Dr. Kügler during my 6-month stay in Germany. He exposed me to a lot of literature and also connected me to as many as possible distinguished linguists. I also my thanks goes the following linguists for the immense support: Prof. Dr. Martine Grice, Head of Institute for Linguistics-Phonetics, University of Cologne, Germany and others wonderful people I came across such as Timo Roetger, Francesco Cangemi, and Albert Aviad. I thank Prof. Dr. Caroline Féry, Johann Wolfgang Goethe University, Frankfurt, for inviting me to give a talk on my chapter drafts. I express my gratitude to Dr. Christian Uffmann, University of Dusseldorf, Germany for the University of Ghana http://ugspace.ug.edu.gh v invitation to his class as a guest lecturer. I thank Prof. Sharon Rose for the brief discussions we had on draft while she was at the University of Cologne. I also thank the participants of Colloquium on African Languages and Linguistics (CALL 2017) at Leiden University, Netherlands especially Prof. Juggar, for their invaluable comments and suggestions on some chapters I presented at the conference. To my wonderful colleague and friend, Mr. Samuel Alhassan Issah, currently a PhD candidate at the Johann Wolfgang Goethe University, Frankfurt, I wish to express my profound appreciations for introducing the whole short-term grants idea to me, and greatly assisting in the process. I am very grateful, Sam. He and his family were my ‘hosts’ from afar in Frankfurt (Marburg). Also to Dr. Obimpeh and his family for hosting me prior to my departure back to Ghana, I say a big thank you. Back home, to my wonderful colleague candidates in the Department of Linguistics, namely Mrs. Abigail Kuwornu (our indefatigable course representative), Mr. Mohammed Sadat, Pastor Mrs. Mavis Gardiner, Ms. Mabel Nunoofio, Mr. Innocent Selasie Akorli, Mr. Albert A. Wornyo (presently at University of Venda, South Africa) and Mr. Novieto Setor Donné (presently at Victoria University of Wellington, New Zealand). You have been a wonderful academic family. Finally, my grand gratitude to my entire family for the wonderful support I have received from you throughout the years I have been on the programme, especially to my lovely wife Grace Adomako (nee Boafo) and my sons Amos Amoah Adomako and Gabriel Owusu Adomako. God richly bless you all. University of Ghana http://ugspace.ug.edu.gh vi ABSTRACT This thesis studies the loanword phonologies of Ga and Dangme (Kwa, Ga- Dangme group) languages. It focuses on investigating how Akan (Kwa, Tano Central group) source words have been phonologically adapted into Ga and Dangme. It has been observed that both Ga and Dangme have borrowed comparatively more Akan words into their vocabularies than Akan has borrowed from Ga and Dangme. However, not much studies have been carried out to establish the nature and extent of borrowings between Akan, on the one hand, and Ga and Dangme, on the other hand. The thesis, therefore, seeks to investigate how the two target languages have adapted some Akan segments, especially non-native segments (both vowels and consonants) into their vocabularies. It also explores how Akan source prosodic features such as syllabic nasals and tone have been realised in the two target languages. The data for this study were collected from the entries in both Ga-English and Dangme-English dictionaries. The study concludes that the phonological adaptations in general are motivated by a combination of phonological and perceptual factors. Phonologically, the adaptations of non-native vocalic and consonantal segments can be accounted for within OT due to the crucial ranking of some markedness constraints. In terms of tone, it has been concluded that the source tone melody is generally preserved in both languages. However, additionally, the same source melodies can be adapted into other melodies present in the targel languages. University of Ghana http://ugspace.ug.edu.gh vii LIST OF ABBREVIATIONS & SYMBOLS Acute accent High ( )́ tone C Consonant DF Distinctive Feature theory F1 First Formant F2 Second Formant Grave accent Low ( ̀) tone Macron Mid( )̄tone N Syllabic nasal OT Optimality Theory S Syllabic consonant UNSD United Nations Statistical Division V Vowel ꜝH Downstepped High * Constraint violation mark *! Fatal constraint violation mark σ Syllable University of Ghana http://ugspace.ug.edu.gh viii LIST OF TABLES Table 1: Akan consonantal inventory ....................................................................... 9 Table 2: The Ga sociolects (Adopted from Dakubu 2002:2) ................................. 11 Table 3: Akan vs. Ga vowels mapping ................................................................... 12 Table 4: Ga consonantal phonemes (Adopted from Dakubu 2002: 61) ................. 13 Table 5: Akan vs. Ga consonant mapping .............................................................. 14 Table 6: Akan vs. Dangme vowel mapping ........................................................... 17 Table 7: Dangme consonantal phonemes (Adapted from Dakubu 1987:13) ......... 18 Table 8: Akan vs. Dangme consonant mapping ..................................................... 18 Table 9: Duration of recordings of Ga/Dangme words & simple sentences .......... 28 Table 10: Summary of information about the respondents for the triangulation exercise ................................................................................................................... 32 Table 11: Adaptations of Akan source non-native vowels in Ga and Dangme .... 117 Table 12: Adaptations of some Akan source consonants in Ga and Dangme ...... 158 Table 13: Adaptations of source prosodic features in Ga and Dangme ............... 207 Table 14: Adaptations of Akan source tonal melodies in Ga and Dangme .......... 251 University of Ghana http://ugspace.ug.edu.gh ix LIST OF TABLEAUX Tableau 1: The adaptation of the Akan source [ʊ] as [u] in Ga ............................. 71 Tableau 2: The adaptation of an Akan source [ʊ] as [ɔ] in Ga ............................... 74 Tableau 3: The adaptation of an Akan source [ɪ] as [i] in Ga ................................ 79 Tableau 4: The adaptation of source [ɪ] as [ɛ] ........................................................ 83 Tableau 5: The realisation of [æ] in Akan .............................................................. 89 Tableau 6: The adaptation of the source [æ] as [a] in Ga ....................................... 89 Tableau 7: The adaptation of the Akan source [ʊ] as [u] in Dangme .................... 99 Tableau 8: The adaptation of the Akan source [ʊ] as [ɔ] in Dangme ................... 102 Tableau 9: The adaptation of Akan source [ɪ] as [i] in Dangme ......................... 106 Tableau 10: The adaptation of source [ɪ] as [ɛ] .................................................... 109 Tableau 11: The adaptation of the Akan source [æ] as [a] in Dangme ................ 111 Tableau 11: The adaptation of source [ʥɥ] as [ʤʷ] in Ga ................................... 124 Tableau 12: The adaptation of source [ɕɥ] as [ʃʷ] in Ga ...................................... 126 Tableau 13: The adaptation of source [ɥ] as [w] in Ga ........................................ 130 Tableau 14: The adaptation of source [Cɥ] as [Cw] in Ga ..................................... 132 Tableau 15: The adaptation of source word-final [w] as [o] in Ga ...................... 137 Tableau 16: The adaptation of source word-final [w] as [ø] in Ga ...................... 140 Tableau 17: The adaptation of source word-final [w] through compensatory lengthening in Ga .................................................................................................. 142 Tableau 18: The adaptation of the source non-native [ʥɥ] as [ʤʷ] in Dangme .. 145 Tableau 19: The adaptation of the source non-native [ɥ] as [w] in Dangme ....... 150 Tableau 20: The adaptation of the source word-final [w] as [o] in Dangme ........ 154 Tableau 21: The adaptation of the source nasality in [ɪ]̃ in Ga ............................. 165 Tableau 22: The adaptation of the source nasality in [ʊ̃] in Ga ........................... 167 Tableau 23: The adaptation of the source [ɪ]̃ in Dangme ..................................... 173 Tableau 24: The adaptation of the source nasality in [ɪ]̃ in Dangme .................... 175 Tableau 25: The deletion of the Akan source word-initial syllabic nasal in Dangme .............................................................................................................................. 192 Tableau 26: The deletion of the Akan source syllabic nasal in NN context in Dangme ................................................................................................................. 198 University of Ghana http://ugspace.ug.edu.gh x LIST OF FIGURES Figure 1: The Standard OT Architecture (Adapted from McCarthy 2003a: 211) .. 47 Figure 2: Results of Ga speakers listening experiment .......................................... 62 Figure 3: Results of Dangme speakers listening experiment.................................. 94 Figure 4: A spectrogram showing the syllable structure of the adapted Ga word mɔ̃́ dɛ̃́ ŋ̃́ ‘laudable effort’ ......................................................................................... 182 Figure 5: A spectrogram showing the syllable structure of the adapted Dangme word ‘pity, sorrow’ mɔ̃́ bɔ̀. ..................................................................................... 193 Figure 6: A spectrogram showing the pitch levels of m̀mɔ̃́ bɔ̃́ ‘pity, sorrow’ in Akan .............................................................................................................................. 219 Figure 7: A spectrogram showing the pitch levels of mɔ̃́ bɔ̀........ ‘pity, sorrow’ in Ga 219 Figure 8: A spectrogram showing the pitch levels of mɔ̃́ bɔ̀ ‘pity, sorrow’ in Dangme ................................................................................................................. 220 Figure 9: A spectrogram showing the pitch levels of sùbã́ ŋ̃́ ‘character’ in Akan 231 Figure 10: A spectrogram showing the pitch levels of súbã́ ŋ̃́ ‘character’ in Ga .. 232 Figure 11: A spectrogram showing the pitches in àmà nɪ̀b ɔ̃́ ‘mission, report, news’ in Akan ................................................................................................................. 234 Figure 12: A spectrogram showing the pitches in āmã̄ níɛ̃́bɔ̀ ‘mission, report, news’ in Dangme ............................................................................................................. 235 Figure 13: A spectrogram showing downstep in ŋ̀káꜝtɪ̃́ɛ̃́ ‘groundnut, peanut’ in Akan ...................................................................................................................... 237 Figure 14: A spectrogram showing downstep in ŋ̀káꜝtíɛ̃́ ‘groundnut, peanut’ in Ga .............................................................................................................................. 238 Figure 15: A spectrogram showing the pitch level of ɔ̀dáꜝsɪ̃́ ‘testimony’ in Akan .............................................................................................................................. 241 Figure 16: A spectrogram showing the pitch level of òdásè ‘testimony’ in Ga ... 242 Figure 17: A spectrogram showing the pitch level of àdɪ̃́ꜝbɔ̃́ ‘creation, nature’ in Akan ...................................................................................................................... 243 Figure 18: A spectrogram showing the pitch level of ādébɔ̀ ‘creation, nature’ in Dangme ................................................................................................................. 244 Figure 19: A spectrogram showing the pitch level of ōdásē ‘testimony, witness’ in Dangme ................................................................................................................. 246 Figure 20: A spectrogram showing the pitch levels of ākáté ‘groundnut, peanut’ in Dangme ................................................................................................................. 248 University of Ghana http://ugspace.ug.edu.gh xi LIST OF APPENDICES Appendix 1: Elicitation of words (for Ga) ............................................................ 267 Appendix 2: Elicitation of words (for Dangme) ................................................... 269 Appendix 3: Recordings of social events ............................................................. 271 Appendix 4: List of Akan loanwords in Ga .......................................................... 272 Appendix 5: List of Akan loanwords in Dangme ................................................. 301 Appendix 6: Ga war song ..................................................................................... 311 Appendix 7: Dangme war song ............................................................................ 313 University of Ghana http://ugspace.ug.edu.gh xii TABLE OF CONTENTS DECLARATION ....................................................................................................... i DEDICATION .......................................................................................................... ii ACKNOWLEDGEMENT .......................................................................................iii ABSTRACT ............................................................................................................ vi LIST OF ABBREVIATIONS & SYMBOLS ........................................................ vii LIST OF TABLES .................................................................................................viii LIST OF TABLEAUX ............................................................................................ ix LIST OF FIGURES .................................................................................................. x TABLE OF CONTENTS ....................................................................................... xii CHAPTER ONE ....................................................................................................... 1 GENERAL INTRODUCTION ................................................................................ 1 1.1. Introduction ................................................................................................ 1 1.1.1. Language Information ........................................................................ 3 1.1.1.1. Overview of Akan language ........................................................... 3 1.1.1.2. Akan vowel system ......................................................................... 3 1.1.1.3. Akan consonant system ................................................................... 6 1.1.1.4. The syllable in Akan ....................................................................... 9 1.1.1.5. Tone in Akan ................................................................................. 10 1.1.1.6. Overview of Ga language ............................................................. 11 1.1.1.7. Ga vowel system ........................................................................... 12 1.1.1.8. Ga consonant system ..................................................................... 13 1.1.1.9. The syllable in Ga ......................................................................... 14 1.1.1.10. The tone in Ga ........................................................................... 14 1.1.1.11. Overview of Dangme language ................................................. 16 1.1.1.12. The Dangme vowel system ....................................................... 16 1.1.1.13. The Dangme consonant system ................................................. 17 1.1.1.14. The syllable in Dangme ............................................................. 19 1.1.1.15. The tone in Dangme .................................................................. 19 1.2. Problem statement ................................................................................... 20 University of Ghana http://ugspace.ug.edu.gh xiii 1.3. Objectives ................................................................................................. 21 1.4. Research Questions/Hypotheses .............................................................. 22 1.5. Methodology ............................................................................................ 22 1.5.1. Description of research methods ...................................................... 22 1.5.2. The Data source ................................................................................ 22 1.5.3. Data ................................................................................................... 23 1.5.4. Data collection instruments and techniques ..................................... 26 1.5.5. Sample/metadata ............................................................................... 30 1.5.6. Selection of respondents ................................................................... 33 1.6. Ethical considerations ............................................................................. 33 1.6.1. Consent procedures ........................................................................... 33 1.6.2. Confidentiality .................................................................................. 34 1.6.3. Privacy .............................................................................................. 34 1.6.4. Risks and benefits ............................................................................. 35 1.6.5. Compensation ................................................................................... 35 1.6.6. Withdrawal from Study .................................................................... 35 1.7. Thesis Overview ....................................................................................... 35 1.8. Summary of the chapter ........................................................................... 37 CHAPTER TWO .................................................................................................... 38 LITERATURE REVIEW ....................................................................................... 38 2.1. Introduction .............................................................................................. 38 2.2. What is loanword adaptation? ................................................................. 41 2.3. Overview of literature on loanword among some Ghanaian languages . 42 2.4. Theoretical Framework ........................................................................... 44 2.4.1. Distinctive Feature (DF) theory ........................................................ 44 2.4.2. Optimality Theory (OT) ........................................................................ 45 2.5. Models/Approaches to loanword adaptation ........................................... 49 2.5.1. The production-based model ............................................................ 49 2.5.2. The perception-based model ............................................................. 52 2.5.3. The ‘hybrid’ model ........................................................................... 57 2.6. Summary of the chapter ........................................................................... 59 CHAPTER THREE ................................................................................................ 60 AKAN VOCALIC ADAPTATIONS IN GA AND DANGME ............................. 60 University of Ghana http://ugspace.ug.edu.gh xiv 3.1. Introduction .................................................................................................. 60 3.2. Adaptations of Akan source non-native vowels in Ga ................................. 61 3.2.1. The adaptations of source [ʊ] ................................................................ 64 3.2.1.1. Adaptation of source [ʊ] as [o] ....................................................... 65 3.2.1.2. Adaptation of source [ʊ] as [u] ....................................................... 67 3.2.1.3. Adaptation of source [ʊ] as [ɔ] ....................................................... 72 3.2.2. The adaptations of source [ɪ] ................................................................. 75 3.2.2.1. Adaptation of source [ɪ] as [e] ........................................................ 76 3.2.2.2. Adaptation of source [ɪ] as [i] ......................................................... 77 3.2.2.3. Adaptation of source [ɪ] as [ɛ] ........................................................ 80 3.2.3. The adaptations of source [æ] ............................................................... 84 3.2.3.1. Adaptation of source [æ] as [a] ....................................................... 84 3.2.3.2. Summary of discussions ................................................................. 90 3.3. Adaptations of Akan source non-native vowels in Dangme ......................... 93 3.3.1. The adaptations of source [ʊ] ................................................................ 96 3.3.1.1. Adaptation of source [ʊ] as [o] ....................................................... 96 3.3.1.2. Adaptation of source [ʊ] as [u] ....................................................... 98 3.3.1.3. Adaptation of source [ʊ] as [ɔ] ..................................................... 100 3.3.2. The adaptations of source [ɪ] in Dangme ............................................ 103 3.3.2.1. Adaptation of source [ɪ] as [e] ...................................................... 103 3.3.2.2. Adaptation of source [ɪ] as [i] ....................................................... 105 3.3.2.3. Adaptation of source [ɪ] as [ɛ] ...................................................... 107 3.3.3. The adaptations of source [æ] in Dangme ........................................... 110 3.3.3.1. Adaptation of source [æ]as [a] ...................................................... 110 3.3.3.2. Adaptation of source [æ] as [ɛ] ..................................................... 112 3.3.4. Summary of discussions ...................................................................... 114 3.4. Summary of the chapter ............................................................................. 115 CHAPTER FOUR ................................................................................................ 118 AKAN CONSONANTAL ADAPTATIONS IN GA AND DANGME .............. 118 4.1. Introduction ................................................................................................ 118 4.2. Adaptations of Akan source non-native consonants in Ga ........................ 119 4.2.1. Adaptations of source non-native consonants in Ga ........................... 121 4.2.1.1. Adaptation of source [ʨɥ] as [ʧʷ] ................................................. 121 4.2.1.2. Adaptation of source [ɕɥ] as [ʃʷ] .................................................. 125 4.2.1.3. Adaptation of source [ɥ] as [w] .................................................... 127 4.2.1.4. Adaptation of source derived [Cᶣ] as [Cʷ] .................................... 130 4.2.1.5. Adaptation of source [ɕ] as [h] ..................................................... 133 University of Ghana http://ugspace.ug.edu.gh xv 4.2.1.6. Adaptations of source word-final [w] ........................................... 135 4.3. Adaptations of Akan source non-native consonants in Dangme ............... 143 4.3.1. Adaptations of Akan source labial-palatals in Dangme ...................... 144 4.3.1.1. Adaptation of source [ʥɥ] as [ʤʷ]............................................... 144 4.3.1.2. Adaptation of source [ʨɥ] ............................................................ 146 4.3.1.3. Adaptation of source [ɕɥ] ............................................................. 147 4.3.1.4. Adaptation of source non-native [ɥ] as [w] .................................. 149 4.3.1.5. Adaptations of source non-native [Cᶣ] as [Cʷ] ............................. 151 4.3.2. Adaptations of source word-final consonants in Dangme .................. 152 4.3.2.1. Adaptations of source word-final [w] as [o] ................................. 152 4.4. Summary of the chapter ............................................................................. 156 CHAPTER FIVE .................................................................................................. 159 AKAN PROSODIC ADAPTATIONS IN GA AND DANGME ......................... 159 5.1. Introduction ................................................................................................ 159 5.2. Adaptations of Akan source vocalic nasality in Ga ................................... 160 5.2.1. Adaptations of Akan source nasalised [ɪ]̃ ............................................ 161 5.2.2. Adaptations of Akan source nasalised [ʊ̃] ........................................... 166 5.2.3. Adaptation of Akan source non-nasalised vowels as nasalised vowels ....................................................................................................................... 168 5.2.4. Adaptation of Akan source nasalised vowels as non-nasalised vowels ....................................................................................................................... 169 5.3. Adaptations of Akan source nasality in Dangme ....................................... 170 5.3.1. Adaptations of Akan source nasalised [ɪ]̃ ............................................ 171 5.3.2. Adaptations of Akan source nasalised [ʊ̃] ........................................... 174 5.3.3. Adaptation of Akan source non-nasalised vowels as nasalised vowels ....................................................................................................................... 176 5.3.4. Adaptation of Akan source nasalised vowels as non-nasalised vowels ....................................................................................................................... 177 5.4. Adaptations of Akan source syllabic nasals in Ga ..................................... 179 5.4.1. Adaptations of Akan source word-initial syllabic nasals .................... 179 5.4.1.1. Where the source word-initial syllabic N is deleted ..................... 180 5.4.1.2. Where source word-initial syllabic N is adapted as V .................. 182 5.4.2. Adaptations of source word-medial syllabic nasals ............................ 184 5.4.3. Adaptations of Akan source word-final syllabic nasal ........................ 185 5.4.4. Truncation of Akan source syllable structure ..................................... 188 5.5. Adaptations of Akan source syllabic nasals in Dangme ............................ 190 University of Ghana http://ugspace.ug.edu.gh xvi 5.5.1. Adaptations of source word-initial syllabic nasals .............................. 190 5.5.1.1. Where word-initial N is realised as V .......................................... 194 5.5.2. Adaptations of Akan source word-medial syllabic nasals .................. 195 5.5.3. Adaptations of source word-final syllabic nasals ................................ 199 5.5.3.1. Where word-final syllabic N is adapted as V ............................... 200 5.5.3.2. Where V is epenthesised to the word-final syllabic N ................. 202 5.5.4. Truncation of Akan source syllable structure ..................................... 203 5.6. Summary of the chapter ............................................................................. 204 CHAPTER SIX ..................................................................................................... 208 AKAN TONES ADAPTATIONS IN GA AND DANGME................................ 208 6.1. Introduction ................................................................................................ 208 6.2. Adaptations of Akan source lexical tone system in Ga and Dangme ........ 212 6.2.1. Adaptations of tone of Akan source nominal stems ............................ 214 6.2.1.1. Adaptations of Akan source H ...................................................... 214 6.2.1.2. Adaptations of Akan source HH(H) ............................................. 216 6.2.1.3. Adaptations of Akan source LH ................................................... 222 6.2.1.4. Adaptations of Akan source (L/H)HLH ....................................... 227 6.2.1.5. Adaptations of Akan source HL ................................................... 228 6.2.1.6. Adaptations of tones in Akan source compound stems ................ 229 6.3. Adaptations of Akan source downstepped H in Ga and Dangme .............. 235 6.3.1. Adaptation of Akan source ꜝH as ꜝH ................................................... 236 6.3.2. Adaptation of Akan source ꜝH as L ..................................................... 239 6.3.3. Akan source ꜝH as M ........................................................................... 245 6.3.4. Adaptation of Akan source ꜝH as H .................................................... 247 6.3.5. Adaptation of Akan source H as ꜝH .................................................... 248 6.4. Summary of the chapter ............................................................................. 249 CHAPTER SEVEN .............................................................................................. 252 SUMMARY, RECOMMENDATION, AND CONCLUSION ............................ 252 7.1. Introduction ................................................................................................ 252 7.2. Summary of preceding chapters ................................................................. 252 7.3. Concluding remarks ................................................................................... 257 7.4. Significance of the study/contribution to knowledge ................................. 262 7.5. Recommendations for future studies .......................................................... 263 University of Ghana http://ugspace.ug.edu.gh xvii APPENDICES ...................................................................................................... 267 Appendix 1: Elicitation of words (for Ga) ........................................................ 267 INSTRUMENT #1. ....................................................................................... 267 Appendix 2: Elicitation of words (for Dangme) ............................................... 269 INSTRUMENT #2. ....................................................................................... 269 Appendix 3: Recordings of social events .......................................................... 271 INSTRUMENT #3. ....................................................................................... 271 Appendix 4: List of Akan loanwords in Ga ....................................................... 272 Appendix 5: List of Akan loanwords in Dangme .............................................. 301 REFERENCES ..................................................................................................... 316 University of Ghana http://ugspace.ug.edu.gh 1 CHAPTER ONE GENERAL INTRODUCTION 1.1. Introduction This study focuses on investigating how Akan source words have been phonologically adapted into Ga and Dangme. Speakers of Akan, Ga and Dangme have had contacts through trade, intermarriages, and education (Dakubu 1987, Dakubu 2009a). These three languages have had influence on each other. This influence has mainly resulted in borrowing some aspects of their cultural practices as well as some words into the vocabulary of the respective languages. According to Dakubu (1999, 2009a, 2013) and Owusu-Adjei (2007), Ga and Dangme have borrowed more Akan words into their vocabularies than Akan has borrowed from Ga and Dangme. Though one can identify Akan source words in Ga and Dangme, not much study has been carried out on the segmental and prosodic structure of Akan words that have been loaned into Ga and Dangme. This study, therefore, seeks to investigate how Akan source words have been segmentally and prosodically adapted to the phonological structure of both Ga and Dangme. The study also investigates the tonal structure of Akan loanwords in Ga and Dangme. From the data gathered for the present study, which is confirmed by some data from Dakubu’s (1999, 2009b, 2013) dictionaries, the majority of the borrowed words from Akan in both Ga and Dangme, as has been indicated in the dictionaries, come from the Akuapem dialect of Akan. Therefore, unless otherwise stated, the Akan data discussed in this study would come from the Akuapem dialect. University of Ghana http://ugspace.ug.edu.gh 2 There are three schools of thought that have emerged concerning the approach or model to explain the loanword phenomenon. These are production- based model (cf. Paradis & LaCharité 1997, Silverman, 1992, Boersma 2007, Boersma & Hamann 2009, among others), perception-based model (cf. Peperkamp & Dupoux 2003, Vendelin & Peperkamp 2006), and a third view that posits a combination of factors based on production and perception, and sometimes on non-grammatical factors, which I have termed it ‘hybrid’ in this thesis (cf. Kenstowicz 2001, Uffmann 2013). These approaches are discussed in chapter two of this thesis. This study, therefore, focuses on providing explanations for the adaptations of Akan source segments, that is, vocalic adaptation which focuses on the phonological patterns for adapting source non-native vowels by both Ga and Dangme speakers in chapter three. The segmental adaptations also look at how speakers of the two target languages adapt some source consonants, including non- native consonants in chapter four. Another important discussion that will engage our attention in the present thesis will be that of the adaptations of prosodic features of the Akan source words by both Ga and Dangme speakers. These prosodic features include the syllable structure in chapter five, and tone in chapter six. In the rest of this chapter, I provide the relevant background information on the phonology of Akan, Ga and Dangme. University of Ghana http://ugspace.ug.edu.gh 3 1.1.1. Language Information 1.1.1.1. Overview of Akan language Akan is a member of the Kwa (Tano Central) sub-family of the Niger-Congo language family. Akan is spoken mainly in Ghana and some parts of the eastern part of Côte d`Ivoire both in West Africa. The language is the most widely used language presently in Ghana. It is spoken as first language (L1) in six out of the present ten administrative regions in Ghana. These regions are Ashanti, Brong Ahafo, Central, Eastern, Western and small part of Volta. In terms of its speaker population, it is estimated that currently, Akan has over 9 million speakers, which is divided into over 8 million L1 speakers and approximately 1 million L2 (second language) speakers in Ghana alone (cf. Simons & Fennig 2017). The number of its dialects usually varies between ten and twelve in existing literature. While some authors, including Agyekum (2006) list as many as twelve dialects, namely Asante, Akuapem, Akwamu, Fante, Akyem, Agona, Assin, Denkyira, Twifo, Wassaw, Kwawu, and Bron, a very recent report by Simons & Fennig (2017) puts the figure at ten distinctive dialects. They include Agona, Akuapem, Akyem, Asante, Asen, Dankyira, Fante (Anomabo, Abura), Kwawu (Kwahu), Gomua, and Ahafo. Out of these dialects, three have received literary status in Ghana, namely Akuapem, Asante, and Fante. The first two major dialects together form the Twi group. 1.1.1.2. Akan vowel system In terms of its vocalic phonology, Akan phonologists are still debating the actual numbers that constitute the vocalic inventory of Akan (cf. Dolphyne 1967, 1988; University of Ghana http://ugspace.ug.edu.gh 4 Stewart 1967; Abakah 2002, 2005a, 2013; Boadi 2009). Generally, it is accepted that Akan has nine (9) phonemic vowels, namely /i, ɪ, e, ɛ, u, ʊ, o, ɔ, a/. However, the status of the tenth vowel [æ] is still being debated by Akan scholars as to whether it is phonetic or phonemic. While majority of authors including Dolphyne (1988/2006), Adu Manyah (2003), and Boadi (2009) posit that Akan has nine (9) oral phonemic vowels, others, including Dolphyne (1967), Stewart (1967)1, Abakah (2002, 2004,2005a, 2013) argue that the number of vocalic phonemes is ten (10), which includes /æ/. The explanation has been that while the Twi dialects operate the ten-vowel system, many varieties of the Fante dialect, on the other hand, have nine phonemic vowels in their system. However, Abakah (2002) provides evidence to support his claim that the vowel is contrastive in some sub- dialects (especially Boka) of the Fante dialect. As to whether this vowel is still contrastive in Akan, Boadi (2009: 6-11), asserts that the advanced low vowel seems to be losing its contrast with its unadvanced counterpart, /a/ synchronically in Akan. It can be deduced from above that the number of phonemic vowels in Akan is dialectal; while it could be ten in Fante, as posited by Abakah (op cit), it is nine in both Akuapem and Asante. The tenth vowel, which is [æ], is realised at the phonetic level of representation in both Akuapem and Asante. In the present study, my reference to the ‘Akan vowels’ includes [æ], which is phonetic in the two Twi dialects. Since the discussions on vocalic adaptations will focus on height, I group all the ten vowels according to their height in example (1) below. 1Stewart (1967: 185) represents his advanced low vowel /æ/ as /ɜ/ in his paper. University of Ghana http://ugspace.ug.edu.gh 5 (1). Classification of Akan vowels by tongue height High: i, ɪ, u, ʊ Mid: e, ɛ, o, ɔ Low: æ, a According to Dolphyne (1988), in addition to the ten oral vowels, there are six nasal vowels in Akan, which are /ũ, ʊ̃, ĩ, ɪ,̃ ã, æ̃/. Out of these six nasal vowels, five are present in the Twi dialects and six one, that is, /æ̃/ can be phonemic in Fante. The two main types of harmonies that apply in Akan are Advanced Tongue Root (ATR) and rounding harmonies. The former has received much attention in extant literature (cf. Stewart 1967, 1983; Clements 1981; Dolphyne 1967, 1988/2006; Ballard 2010; Abakah 2012; Kügler 2015; among others). The ATR harmony is a phonological process that requires that vowels that occur within a stem or morpheme or a phonological word, phrase or even a sentence belong to one set of vowels, either advanced or unadvanced. Based on this rule, all the ten vowels in Akan are grouped into two sets in terms of their ATR values as shown in (2a). (2). The Akan ATR sets a. ATR pairing [+ATR]: i, e, u, o, æ [-ATR]: ɪ, ɛ, ʊ, ɔ, a University of Ghana http://ugspace.ug.edu.gh 6 The other type of harmony associated with Akan vowels is the rounding harmony. In Akan rounding harmony, the Akan vowels are divided into two sets based on their values for the feature [round]. This rounding pairing is exemplified in (2b). (2). The Akan round sets b. Rounding set [+round]: u, ʊ, o, ɔ [-round]: i, ɪ, e, ɛ, æ, a For details discussions and examples of Akan words that exhibit both ATR and rounding harmonies within stems and across morpheme boundaries, see Dolphyne (1988). 1.1.1.3. Akan consonant system Just like the vocalic inventory of Akan, its consonantal inventory is also not without a debate. This assertion finds expression in a statement made by Abakah (2006) that: “…modern linguistic study of the various dialects of Akan started by the close of the nineteenth century and to date there seems to be controversy over the number of systematic phonemes contained in the consonantal inventory of the language” (Abakah 2006: 22). University of Ghana http://ugspace.ug.edu.gh 7 This was in direct response to a claim made in the preliminary report on Akan phonology by Schachter & Fromkin (1968), in which they reported that Akan has only eight ‘true’ consonants (1968:35). Abakah (2006) has challenged this claim and rather opined instead that there are fourteen (14) phonemic consonants in Akan which he listed as follows: p, b, t, d, k, g, m, n, r, f, s, h, j, w. Aside from these phonemic consonants, Abakah presented a chart of sixty-four (64) phonetic variants of the fourteen consonantal phonemes and posited that the additional fifty (50) phonetic consonants (i.e. the 14 phonemes aside) could be derived through the application of some phonological rules. This same statement had been made previously by Christaller (1933), Dolphyne (1988), among others. The statuses of some consonants, including the following: /ʨ, ʥ, ʨɥ, ʥɥ, ɕɥ, ɲ/ still remain unclear. With the exception of /ʥ/, Abakah (2006) listed all these consonants before front vowels, implying that they have undergone palatalisation. However, there is enough evidence from Akan to prove that these consonants can occur before non-front vowels, namely central and back vowels as we can see in (3) below. (3). Some examples of alveo-palatals in Akan i. aʨaʨa a hunch-back ii. nʨo/nʨɥo defeat iii. ʥata lion iv. ʥʊnsɔ/ʥɥʊnsɔ urine University of Ghana http://ugspace.ug.edu.gh 8 v. ɛʨɥʊ hernia vi. eʨɥã scar vii. eʥɥa market viii. ʥɥoŋku hip ix. ɲɲɥʊm songs x. ɲɥunu cold It is worth pointing out that the long-standing debate over whether these consonants are phonemic (palatal) or phonetic (palatalised) does not seem to have been resolved yet (cf. Mensah 1977, Boadi 1988). Those who posit a diachronic change from plain back consonants; k, g, n into ʨ, ʥ, ɲ respectively before front vowels might not consider such consonants phonemic. The question now is, once they can occur before non-front vowels, have they synchronically assumed a phonemic status in Akan segmental phonology? Though there seems to be a lack of a clear-cut answer to this question, I include these alveo-palatals and ‘labialised’ alveo-palatals in the list of consonantal phonemes in Akan. I posit that these consonants might be going through a phonemicisation process in Akan. The Table 1 below displays the consonantal inventory of Akan. University of Ghana http://ugspace.ug.edu.gh 9 Table 1: Akan consonantal inventory Bilabial Labio Alveolar Alveo- Palatal Labialised Vela Labial- Glottal dental palatal alveo-palatal r velar Stop p, b t, d k, ɡ Affricates ʨ, ʥ ʨɥ, ʥɥ Fricatives f s ɕɥ h Nasals m n ɲ ɲɥ ŋ Approximants (w) ɹ j w 1.1.1.4. The syllable in Akan Akan operates simple CV and V as its basic syllable structure (cf. Dolphyne 1988/2006). There are also syllabic consonants, which is represented as ‘S’ due to their sonority status, such as word-final and preconsonantal nasals, liquids, and the labio-velar glide (cf. Abakah 2005a). These word-final syllabic consonants, according to Abakah (2004), are derived through the elision of the underlying [+high] final vowels. So the three basic syllable types in Akan are CV, V, and S. It is worth noting at this point the crucial role tone plays in the syllabification process in Akan. In Akan, every syllable is a tone-bearing unit. According to Dolphyne (1988), all the vowels and all the syllabic consonants usually occupy the nucleus of the syllable, and on their own can constitute syllables. University of Ghana http://ugspace.ug.edu.gh 10 1.1.1.5. Tone in Akan As it is characteristic of tone languages, tones are phonemically contrastive, i.e. a morpheme with the same syllables can have two or more different meanings depending on the kind of tones the syllables may bear. Akan has two basic contrastive tones; High (H) and Low (L). Out of these two levels, there can be compound tones, i.e. falling (HL), as well as rising (LH) tone. For example, from the native Akan words in (7) below, it is the tone that contrasts the meanings. This is the lexical function of tone in Akan (cf. Dolphyne 1988; Abakah 2004, 2005a, 2005b; Kügler & Genzel 2011a, b; Genzel 2013; Kügler 2016; among others). (4). Akan Gloss i. pápá good ii. pàpá father iii. pàpà fan The interaction of the level tones may lead to the application of tone rules such as downstep where the pitch level of a H is perceived relatively lower than preceding H when it is immediately preceded by a L. This tone rule is schematised in (5). (5). H- L– H→ H - L -ꜝH University of Ghana http://ugspace.ug.edu.gh 11 In chapter six of this thesis, I discuss how the level tonal melodies of Akan source nouns are realised in both Ga and Dangme. In the same chapter, I also consider the interpretations of the downstep rule in the adaptations of Akan source nouns in the two target languages. 1.1.1.6. Overview of Ga language Ga is a member of the Niger-Congo (Kwa) family of languages. It is one of the only two languages that belong to the Ga-Dangme group. The language is spoken mainly in the Greater Accra Region; the national capital of Ghana. Though the language does not have dialects, according to Dakubu (2002) citing Kotei’s (1969) thesis and (Mante 1971), there are variations in pronunciations of some Ga words among speakers in some areas as follows: Table 2: The Ga sociolects (Adopted from Dakubu 2002:2) Accra La Teshie Nungua Kpone Gloss i. o ̀nṵ́fṵ́ o ̀nṵ̀u ̰f́u ̰́ oǹ ṵ́ꜝfṵ́ o ̀nṵ̀u ̰́fṵ́ o ̀nṵ̀u ̰́fṵ́ snake ii. ɡ͡be ̀saŋ̀ e ̀ ɡ͡bòsa ̀ŋe ̀ ɡ͡bòsa ̀ŋe ̀ ɡ͡bòsa ̀ŋe ̀ ɡ͡bòsaŋ̀ e ̀ leathery turtle iii. o ̀dáa ̀kl̀ɔ ̀ od̀ áa ̀kl̀ɔ ̀ o ̀dáa ́ꜝkĺɔ ́ od̀ áa ̀kĺɔ ̀ o ̀dáak̀ ĺɔ ́ lizard iv. a ̀kl̀ònt̀ ḭ́a ̰̀ a ̀kɔǹ̀tiá̰ ̰́ ak̀ l̀òti ̰́a ̰̀ a ̀kl̀òtḭ́a ̰̀ ak̀ l̀òtḭ́a ̰̀ pin In terms of speaker population, Ga has a population of 745,000 L1 speakers according to Simons & Fennig’s (2017) estimates based on the 2013 UNSD report. University of Ghana http://ugspace.ug.edu.gh 12 1.1.1.7. Ga vowel system The language has seven oral vowels /i, u, o, ɔ, e, ɛ, a/ and five nasal vowels /ũ, ĩ, ɔ͂, ɛ͂, ã/ in their vocalic inventory (cf. Dakubu 1987, 2002; Wentum 1997; Campbell 2017; among others). According to Dakubu (2002), though Ga vowels can be distinguished based on ATR, thus dividing the seven oral vowels into [+ATR] i, u, o, e; and [-ATR] ɛ, ɔ, a, “…there is no vowel harmony based on this [ATR] feature, as there is in Akan.”(2002: 52). Similarly, these vowels can be grouped according to the feature [round] as seen below in (6). (6). The Ga round sets Rounding set [+round]: u, o, ɔ [-round]: i, e, ɛ, a Table 3: Akan vs. Ga vowels mapping Akan i ɪ u ʊ e ɛ o ɔ æ a Ga i u e ɛ o ɔ a As a brief comparison, it can be seen from the Ga vocalic inventory (also in Table 3) that three Akan vowels, namely /ɪ, ʊ, æ/ are not present in the Ga vowel system. The question that I will seek to address in chapter three of this thesis is, how do the Ga speakers adapt these three non-native vowels in Akan loaned words? University of Ghana http://ugspace.ug.edu.gh 13 1.1.1.8. Ga consonant system In terms of consonantal inventory, Ga has 32 consonantal phonemes (Dakubu 2002, Campbell 2017, among others). Dakubu (2002) lists these consonants as follows in Table 4. Table 4: Ga consonantal phonemes (Adopted from Dakubu 2002: 61) Bilabial Labio- Alveolar Pre- Labialised Velar Labialised Labial- dental palatal pre- velar velar palatal Stop p, b t, d k, ɡ kʷ, ɡʷ k ͡p, ɡ͡b Affricates ʧ, ʤ ʧʷ, ʤʷ Fricatives f, v s, z ʃ ʃʷ h hʷ Nasals m n ɲ ŋ ŋʷ ŋ͡m Approximants j (w) w Laterals l Retroflex r From the Table 4 above, it can be observed that though Ga has more consonantal phonemes than Akan has, there are few consonants present in Akan which are not seen in the Ga system. These consonants include: /ʨɥ, ʥɥ, ɲɥ ɕɥ/. This observation is tabulated as follows in Table 5. The question I will investigate is, are the Akan source labialised alveo-palatals adapted with labialised pre-palatals in Ga? University of Ghana http://ugspace.ug.edu.gh 14 Table 5: Akan vs. Ga consonant mapping Akan ʨɥ ʥɥ ɲɥ ɕɥ ɥ Ga ʧʷ ʤʷ ŋʷ ʃʷ hʷ/w In chapter four of the present thesis, I investigate, among other things, how these Akan source consonants are adapted in Ga. 1.1.1.9. The syllable in Ga Ga, just like Akan, also operates the open syllable system. The most basic syllable structures in Ga are CV, V, and syllabic nasal consonants, which Dakubu (2002) represents as N. For the syllabic nasal consonants - N, they can belong to any place of articulation in onset position of a word, but it is only the velar nasal that can occur in word-final position of words in Ga (cf. Dakubu 2002: 48). So from this, it is obvious that unlike Akan, consonants such as m, n, w, r, are not permitted word- finally and therefore they are not syllabic in Ga. However, according to Dakubu (op cit), the lateral l and the retroflex r can bear tones in C1C2V2 syllable, which is derived from an underlying C1V1C2V2 through the elision of V1 similar to what happens in Akan in such a context. It is worth mentioning that just as it obtains in Akan, the syllable in Ga is a tone-bearing unit. 1.1.1.10. The tone in Ga Ga, just like Akan, is a tone language that operates two contrastive lexical tones, namely H and L as its basic tones (cf. Dakubu 2002). According to Dakubu (2002: University of Ghana http://ugspace.ug.edu.gh 15 6), this two-level tone property is, “…one of the major structural differences between Ga and its only close relative, Dangme, which has three phonemic tones…” Again, the pitch contrast can provide a contrast at the lexical as well as the grammatical level (Dakubu 2002). In the following examples, I present Ga monosyllabic words which exhibit the H and L contrast. (7). Ga Gloss a. i. lá sing, blood ii. là fire, dream b. i. bú hole ii. bù guard c. i. kp͡ ɛ ́ carve ii. k͡pɛ ̀ sew d. i. ká knock ii. kà stuck Aside from these basic tones, Dakubu (2002) posits that there can be compound tones in Ga, namely HL, and LH tones which are not contrastive, as obtains in other contour tone languages like Japanese, Mandarin Chinese, Mongolian, etc. Tone rules such as downstep and downdrift also apply in Ga. University of Ghana http://ugspace.ug.edu.gh 16 1.1.1.11. Overview of Dangme language Dangme is a member of the Kwa (Niger-Congo) family of languages. Dangme, together with Ga forms the only members of the Ga-Dangme group (cf. Dakubu 2002, Simons & Fennig 2017, among others). Speakers of Dangme can be found in the Eastern, Greater Accra, and some parts of the Volta regions of Ghana (cf. Simons & Fennig 2017). The language has six dialects, namely Ada, Gbugbla, Klo, Ningo, Osu (Osuwem), and Sɛ (cf. Apronti & Dakubu 1972). It is worth pointing out that the Ada, Ningo, and Gbugbla dialects have been described in the existing literature as the Coastal dialects, while the Klo, Osu and Sɛ form the so- called Inland dialects. The remaining two dialects are described as the Western dialect (cf. Dakubu 1987). According to recent estimates by Simons & Fennig (2017), which are based on the 2013 UNSD report, the language is spoken by approximately 1.02 million speakers in the above-mentioned regions in Ghana. 1.1.1.12. The Dangme vowel system In its vocalic inventory, Dangme has seven oral vowels and five nasal vowels, the same number as Ga, its sister language (cf. Apronti 1967, 1990; Dakubu 1987; Caesar & Adi 2014; Owulah 2014; Larweh 2017; among others). All put together, there are 12 phonemic vowels in Dangme which are listed as follows: /i, u, o, ɔ, e, ɛ, a, ũ, ĩ, ɔ͂, ɛ͂, ã/. In (8) below, I group the Dangme oral vowels in terms of two feature round. University of Ghana http://ugspace.ug.edu.gh 17 (8). The Dangme round sets Rounding set [+round]: u, o, ɔ [-round]: i, e, ɛ, a Table 6: Akan vs. Dangme vowel mapping Akan i ɪ u ʊ e ɛ o ɔ æ a Dangme i u e ɛ o ɔ a Again, as a brief comparison, we can see that the Dangme vocalic inventory does not contain the vowels /ɪ, ʊ, æ/. 1.1.1.13. The Dangme consonant system Dangme, a sister language to Ga, has about 23 consonantal phonemes (cf. Dakubu 1987, Caesar & Adi 2014, Owulah 2014, and Larweh 2017), which I present in Table 7. University of Ghana http://ugspace.ug.edu.gh 18 Table 7: Dangme consonantal phonemes (Adapted from Dakubu 1987:13) Bilabial Labio- Alveolar Pre- Labialised Velar Labial- dental palatal pre-palatal velar Stop p, b t, d k, ɡ k͡p, ɡ͡b Affricates ʧ, ʤ Fricatives f, v s, z h Nasals m n ɲ ŋ ŋ͡m Approximan j (w) w ts Laterals l Unlike Ga, Dangme shares fewer consonants with Akan. And just like Ga, some Akan source alveo-palatal consonants are missing from the Dangme inventory. In Table 8, I show how the Akan source alveo-palatals and the palatal consonants completely lack correspondents in the Dangme consonant system. Table 8: Akan vs. Dangme consonant mapping Akan ʨɥ ʥɥ ɲɥ ɥ ɕ Dangme From Table 8 above, it is observable that the Akan source alveo-palatals, palatal fricative, labial-palatals, and by extension, labial-palatalised consonants are not present in Dangme phonological system at all. In chapter four, I investigate how these non-native source consonants (both phonemic and phonetic) are adapted in Dangme. University of Ghana http://ugspace.ug.edu.gh 19 1.1.1.14. The syllable in Dangme Dangme, just like Ga, has an open syllable (cf. Dakubu 1987: 8, Apronti 1990: 10). Therefore, the basic syllable structures in Dangme are CV, V, and syllabic nasal. The syllabic nasal in Dangme is limited to /m/, which according to Dakubu (op cit), occurs in loanwords usually from Twi, and even that, an attempt is usually made to avoid it. When these syllabic nasals occur in word-final positions in native Dangme words, they are usually derived when, just as in Akan and Ga, the final vowel in the underlying representation is dropped in rapid speech. Apart from these syllabic nasals, there are also syllabic laterals which occur in C1C2V syllable structures where the C2 is either of the two variants; r or l. However, Dakubu continues (citing Apronti’s (1967) Ph.D. thesis in support) that phonetically, it seems sometimes when the C1 is voiceless, the C2, which has assumed nuclear status as a result of the dropping of preceding V in underlying C1V1C2V2 structures, then loses its tone and subsequently forms ‘cluster’ with C1 in this context (see also Apronti 1990). 1.1.1.15. The tone in Dangme Dangme has a three-level contrastive tone system comprising H, M(id) and L tones (cf. Dakubu 1987, 2002). Again, just like Akan and Ga, the syllable in Dangme, according to Dakubu (op cit), is the tone-bearing unit. In the example in (9), we show how tone in Dangme contrasts in monosyllablic words based on data from Dakubu (1987:19). University of Ghana http://ugspace.ug.edu.gh 20 (9). Tone levels in Dangme Monosyllabic words Gloss i. sá snatch ii. sā rot iii. sà sieve This study investigates how Akan’s two-level tone system is interpreted in Dangme’s three-level tone system. It again investigates how some active tone rules in Akan such as downstep (cf. Abakah 2000, Genzel & Kügler 2011b, Genzel 2013, among others, for Akan), are interpreted in the two target (i.e. recipient) languages. 1.2. Problem statement Although the phonological study of loanword phenomenon across languages has attracted much scholarly attention over the years, not much study has been carried out on Ghanaian languages, particularly on Akan words loaned into other Ghanaian languages. The closest attempts in this endeavour can be found in Owusu-Agyei (2007), which focused on both the phonology and semantics of some nativised Akan words in Ga. For the phonological nativisation in particular, she provides a description of how some Akan source segments are adapted in Ga, though she does not provide an account of what condition the nativisation process. Therefore, due to its limitation in scope, no adequate phonological generalisation University of Ghana http://ugspace.ug.edu.gh 21 could be made based on this particular study. Again, it considered only the segmental adaptation and did not touch on the suprasegmentals of nativisation at all. With regard to Dangme, there is not yet any such attempt made on this subject, as far as I know. Among the closest of such study are Dakubu’s (1973) general survey of some loaned words in Dangme, which preceded her draft dictionary in 2013 and Nartey’s (1976) article on English loaned words in Dangme. This study, therefore, seeks to fill this gap by providing a comprehensive account of the adaptation patterns employed by Ga and Dangme speakers in adapting Akan source words into their vocabularies. It also provides formal account of those phonological patterns within the Optimality Theory framework. 1.3. Objectives The main objectives of this project are: • To provides a comprehensive account of which segmental phonological conditions and processes are exploited by Ga and Dangme in the(ir) nativization of Akan words. • To investigates how prosodic units and structures (particularly, the syllable, tone and nasalization) of Akan source words have been adjusted in the borrowing processes to fit the permissible prosodic types and structures of Ga and Dangme. University of Ghana http://ugspace.ug.edu.gh 22 In the course of working towards achieving the above objectives, this study will also attempt to establish the nature of the symmetry and asymmetry, if any, between how Ga and Dangme adapt Akan source words into their vocabularies. 1.4. Research Questions/Hypotheses The following questions are to be addressed within this study: • How are Akan source words segmentally adapted in the Ga and Dangme vocabularies? • What is the interpretation of the prosodic structures of Akan source loanwords in Ga and Dangme? 1.5. Methodology 1.5.1. Description of research methods For the study project, I conducted fieldwork to elicit data for the investigation of how Akan source words are segmentally and prosodically adapted into Ga and Dangme. This fieldwork complemented the collection of such Akan loaned words in both the Ga and Dangme dictionaries. The following methodology was employed to guide the present study. 1.5.2. The Data source The main source of data for this study has been a secondary source. a. Two editions of the Ga-English dictionary with English-Ga index by Dakubu (1999, 2009b) for the Ga data, and University of Ghana http://ugspace.ug.edu.gh 23 b. The draft version the Dangme-English dictionary with English-Dangme glossary by Dakubu (2013). Also, I consulted J. Abedi-Boafo’s (1971/80) Dangme Nyaii (Classical and Idiomatic Dangme) 2nd edition, and E. O. Apronti’s (1990). Dangme Ngmami bɔ (The writing of Dangme) revised edition. These two books were consulted also by Dakubu in her preparation of the draft version of the Dangme dictionary. This secondary source was complemented by a primary source of data comprising recordings of one Ga Asafo war song on 31stMay, 2017 and one Dangme Asafo war song by Asafotsengua (Kudzragbe Division) of Ada Traditional Council on 5th August, 2017. I was assisted by one native speaker linguist each for each of the two target languages to identify the possible Akan loaned words in them. 1.5.3. Data The size and the nature of the data used for this study were as varied: for Ga, I collected approximately 959 entries in Dakubu’s (2009b) dictionary that are of Akan origin (see Appendix 4 for the list). These targeted entries have already been transcribed in the dictionary, and I had to use my Akan native speaker’s intuition together with the help of Christaller’s (1933) dictionary of Akan to do the transcription of the Akan source words. The total number of Ga entries of possible Akan source constitutes about one-tenth (i.e. 10%) of the entire entries in the dictionary, which is significant in that for every ten ‘Ga’ words, one is possibly of Akan origin. This number of entries was complemented by the few words I elicited University of Ghana http://ugspace.ug.edu.gh 24 from the Ga Asafo war songs. I purposively selected and used sixty English words out of the number of entries collected that have syllabic nasals and also of certain tone patterns in when translated into Akan, which will be the focus of my discussions in chapters 5 and 6. I then asked each of the Ga respondents to provide their equivalent words in Ga in isolative style to trangulate the transcriptions and the tone markings on the loaned word entries. For the Dangme data, I collected a total of 333 entries of possible Akan source, also from Dakubu’s (2013) draft dictionary (see Appendix 5 for the list). This number of entries collected constitutes about seven percent (7%) of the total entries in the draft dictionary. Just as in the Ga dictionary, the targeted entries had already been transcribed, but here I had to use my knowledge in Akan sounds to transcribe some of the selected entries since they were still in their orthographic forms, which is expected of a draft work. Here too I used my native speaker’s intuition together with the help of Christaller’s (1933) dictionary to transcribe the Akan source entries. Moreover, I observed that there are far more cases where the Akan source words are not provided at all for many entries though the author alludes to the fact that those entries are of Akan origin. For Dangme too, I purposively selected and used sixty English words out of the number of entries collected. The selection was informed by English words which when translated into Akan have syllabic nasals and certain tone patterns. The Dangme respondents were then asked to provide the equivalent words for the six sampled entries in Dangme. It is worth pointing out that some of the adapted forms in both Ga and Dangme have become obsolescent, while others have undergone semantic shift. I have indicated all these against the entries in the lists in Appendices 4 & 5. University of Ghana http://ugspace.ug.edu.gh 25 Now judging from the fact Akan, Ga, and Dangme belong to the same language family, namely Kwa, the obvious question then is whether these loaned words of Akan source are cognate words or not? I have derived my conviction that those words, as shall be discussed from the chapter three onwards of this thesis, are indeed loaned or ‘nativised’ words of Akan origin for the following four reasons: First, it is worth pointing out that from the dictionaries, virtually all these loaned words of Akan source have their indigenous equivalents already in the dictionaries. It is, therefore, probably for the purpose of prestige that words of Akan source have also been nativized in the vocabularies of both Ga and Dangme as a result of the several contact situations mentioned in the introduction. It must be emphasised that both Ga and Dangme speakers do not lack vocabularies for all the Akan source loaned words that they nativise. Some of the concepts borrowed from Akan into Ga and Dangme co-exist with their native counterparts in the languages. For examples, an Akan loaned word in Dangme; òpléú ‘squirrel’ from the Akan word òpúró co-occurs with ɡ͡blɛḿtsu,̄ the native word for the same animal. Second, the entries in the dictionaries have been subjected to scrutiny and validations by both the author and some native speakers who are language experts over several decades (cf. Dakubu 2013, Introduction). Therefore, it is assumed that those entries, without doubts, have been validated as borrowed or loaned words of Akan origin. University of Ghana http://ugspace.ug.edu.gh 26 Third, the revision that resulted in the 2009 edition of the Ga dictionary, in particular, benefitted from the inputs of native speaker Ga language experts who formed the participants for the two workshops held for that purpose (see Dakubu 2009b: 5, for the list of the participants of the two workshops). The Dangme dictionary has not gone through such a revision yet as the author has explicitly stated in that dictionary. However, there is a committee of Dangme language experts in place that is working currently for that purpose. Fourth, the morphology of some of the words identified in the dictionary easily makes them identifiable as of Akan origin. For example, the use of the suffix morpheme –fo [fʊ] ‘person marker’ which exists only in Akan morphology in such Akan words as [òhià̃ ̃́ꜝ fʊ́] ‘poor person, pauper’. This word is adapted in both Ga and Dangme as [òhià̃ ́ꜝ̃ fó] and [òhià̃ f́̃ ō] respectively with the same meaning. 1.5.4. Data collection instruments and techniques The recordings involved the elicitation of words and simple sentences, as well as of spontaneous social events such as oral texts (e.g. war songs, see Appendix 3 for the instrument used) by some asafo group members in both Ga and Dangme speaking communities for the Ga and the Dangme songs respectively. I used a headset, which had a microphone attached to it, and a laptop for all the recordings of the elicited words and sentences in the Praat software. I complemented these devices with a voice recorder (Olympus digital voice recorder VN-3100PC) in the recording exercise. University of Ghana http://ugspace.ug.edu.gh 27 To test how speakers of the two borrowing languages adapt the segmental and tonal properties of the Akan source words into their lexicons, I provided first the lists of English words which I purposively sampled from the two dictionaries with the respondents of the two target languages. The selection of the stimuli/tokens for the elicitation was guided by one of my objectives for this study; to account for how non-native Akan source segments are adapted in the two target languages. I then asked the respondents to translate them first into their respective languages on the sheets I gave them which contained the list of English words. When I crosschecked the translations they did with the selected entries from the dictionaries, I realised that for some few words, specifically seven and five entries representing approximately 12% and 8% of Ga and Dangme entries respectively, the native Ga and Dangme words were preferred and were written down as the translated words instead of the anticipated nativised words of Akan origin. I then provided the indigenised variants which I had already taken from the dictionaries and asked for their acceptability judgments. Those indigenised words were generally accepted by the respondents as variants. This exercise preceded the actual recording session. During the recording sessions, I used the recording menu in the Praat software to record directly the translated words onto my laptop. I placed a headset with an attached projected microphone over the head of each of the respondents at a time. After adjusting the microphone and testing it, I then asked the respondent to repeat each word (the translated Ga/Dangme word lists) three times while I recorded the repeated words. I did this for the elicitation of Ga and Dangme words of possible Akan origin. The two Ga respondents were given the same wordlist for University of Ghana http://ugspace.ug.edu.gh 28 the recording. Similarly, the Dangme respondents were given the same wordlist for the elicitation exercise (see Appendix 1 & 2 for the samples). The same recording procedure, as has been described above, was followed in the recordings of the Ga and Dangme simple sentences that were constructed using the same elicited words by the respondents. Before the recordings began, I made the respondents relax by taking some deep breaths to ensure they were in their right mood in order not to have a negative effect on the recordings. The elicitation of the Ga wordlist of Akan source lasted an average of 05:19 minutes (311.4 seconds), while the Ga sentences lasted for 06:16 minutes (369.6 seconds) average for the two respondents. For Dangme, on the other hand, the elicitation of the wordlist lasted an average of 05:38 minutes (323.33 seconds), while that of the sentence lasted for 07:10 minutes (approximately 426 seconds). The duration of the recordings of both the wordlists and the sentences by each respondent is presented in the following table. Table 9: Duration of recordings of Ga/Dangme words & simple sentences Ga Dangme Respondent Wordlist Simple sentence Respondent Wordlist Simple sentence (Duration) (Duration) (Duration) (Duration) 1. MO 353 sec. 441 sec. 1. RA 413 sec 437 sec. 2. NA 285 sec. 311 sec. 2. FO 277 sec. 390 sec. 3. MC 471 sec. 580 sec. Though the transcriptions have been done in the dictionaries, as I have already indicated, I did the recordings, to triangulate the earlier works and also to be able to provide acoustic evidence to support my analyses in this study. My acoustic analyses corroborate the transcriptions done in the two dictionaries. University of Ghana http://ugspace.ug.edu.gh 29 The various recordings of the events mentioned above have been subjected to Praat analyses with the objective of establishing segmental qualities and tonal patterns on the adapted forms, which constitute the basis of my study. In addition to verifying the entries collected from the two dictionaries, I set up a Multiple Forced Choice (MFC) listening experiments to test how both Ga and Dangme speakers perceive all the 10 vowels of Akan with particular focus on the 3 non-native vowels. In the experiment, the listeners were tasked to choose from the vowels that appeared on the experiment screen that was most similar to what they heard. The stimuli were vowel sounds. The number of the stimuli was 10 vowels representing the 10 oral vowels of Akan, namely [a, æ, e, ɛ, i, ɪ, o, ɔ, u, ʊ]. The listeners were to judge what they listened to against 7 responses vowels in CV words that appeared on the screen. For the Ga listeners, the responses were: ti, bu, gbe [ɡ͡be], ho, lɛ, nyɔ [ɲɔ], fa, while the Dangme listeners had the following as their responses: fi, pu, te, mo, wɛ, tɔ, kpa [k͡pa]. All these responses are real words in each of the target languages. The number of replications per each stimulus was 4, making 40 stimuli for each listener. There was break after every 10 stimuli. These stimuli were randomised after every break. Each listener had the opportunity to repeat any stimulus they did not perceive well up to a maximum of 5 replays. For the Dangme listeners, there were 15 females and 5 males. The Ga listeners, on the other hand, comprised 8 females and 2 males. My choice of number of respondents for each language was informed by the fact Dangme has University of Ghana http://ugspace.ug.edu.gh 30 six dialects as against one dialect by Ga. Similarly, Dangme has comparative more speakers. The results and the analyses thereof of the listening experiment are presented in chapter 3 of the thesis during the data presentations on the adaptations of Akan source vowels in Ga and Dangme. 1.5.5. Sample/metadata I engaged five respondents for the word elicitation exercises, two males and three females. These respondents are divided between the two languages as follows. For the Ga recordings, a male and a female respondent who were aged between twenty-five and thirty years were engaged. Both respondents were born and bred in the communities where Ga is predominantly spoken as L1 (first language), and have been speaking the language regularly. Their highest level of education is first degree in Ga Education. For the MFC experiment, on the other hand, ten Ga listeners were involved in the listening test. For the Dangme recordings, three respondents aged between twenty-five and forty-five were engaged. Of the respondents, one was a male and two were females. Because of the dialectal variations in Dangme, I selected one respondent from the Klo (Odumase Krobo), one from Ada, and the other one from Gbugbla (Miotso) dialects. All the respondents lived in the communities where the Dangme language is predominantly spoken as L1. Again, all the three respondents have been speaking the language as their L1 themselves. Their highest level of education is a University of Ghana http://ugspace.ug.edu.gh 31 first degree in Dangme Education. One of the respondents was doing his national service in the Department of Ga-Dangme at the University of Education, Winneba. Another respondent, who was in her final year of the degree programme, is a professional teacher. The remaining two of the respondents were also in the final year. For the war song, I had one recording for Ga by a group of different singers, and one recording for Dangme. Both songs were sung by Asafo groups and have been transcribed in appendices 6 and 7 respectively. I have summarised these pieces of information about the respondents in the Table 10 below. For the MFC experiment in Dangme, twenty respondents were engaged in the listening test. These comprised fifteen females, and five males. Just as I did for the elicitation exercise, all the respondents were students from the same department. University of Ghana http://ugspace.ug.edu.gh 32 Table 10: Summary of information about the respondents for the triangulation exercise Respon Sex Age Hometo Place of L1 L2 Level of Langua Professi Comment dent range wn abode proficiency ge of on in L2 parents 1. MO M 18-25 Teshie Teshie Ga EN Good Ga Student s/he has TW Average lived his 2. NA F 18-25 La La Ga EN Good Ga Student She’s TW Good lived her entire life in 3. FO F 36-45 Krobo Krobo Dan Ga V. good Dangme Student/ She’s Odumas Odumas gme TW Good Professi lived her e e FR Good onal entire life EN Good teacher at KO 1. MC F 26-35 Dawhen Dawhen Dan Ga V. good Dangme Student She’s ya ya gme TW V. good lived her EN Good entire life in Dawhenya 2. RA M 26-35 Dan EN Good Dangme Nationa gme TW Good l Service person In order to ensure the validity of the data used for the present study, whenever I was in doubt about any issue concerning the data for verification, I consulted some identified language experts who are native speakers of the two target languages, especially those who teach the languages at the College of Languages Education, University of Education, Winneba at Ajumako. University of Ghana http://ugspace.ug.edu.gh 33 1.5.6. Selection of respondents In selecting my respondents, I employed purposive sampling technique based on social factors such as age, sex, educational background and professional background since these demographic factors could also affect acoustically how Akan source words are adapted into the borrowing languages. I chose the younger respondents due to the fact that the respondents for the two dictionary projects seem to be the older generation. The objective was to verify if age could significantly affect tone adaptation. From the data analysis in this study, I realised that tone markings on the adapted words generally do not change with age. For my choice of a male and a female pairing for each of the two target languages, the objective was to establish whether gender could affect the pitch levels of the source words in the adaptation process. 1.6. Ethical considerations 1.6.1. Consent procedures I stated explicitly to the respondents that the purpose/objectives of the research study are as follows; to gather data of possible Akan source words in Ga and in Dangme in order to provide a comprehensive account of the general adaptation patterns employed by Ga and Dangme speakers in terms of the segment structure of the Akan source words, and also to investigate how Ga/Dangme speakers syllabify those possible Akan source words and also mark tone on those words. University of Ghana http://ugspace.ug.edu.gh 34 I explained to the respondents that the data collection exercise was expected to last for about 3 working days. And each recording session lasted for about 2 hours. For the Asafo war songs recordings, the respondents (the Asafo groups) sung two Asafo war songs, and those songs were audio-recorded. For the elicitation of some Ga/Dangme words, I provided a wordlist of some purposively selected English words and asked the respondents to say the Ga/Dangme equivalents. Their responses were audio-recorded by me. After the elicitation of the wordlist, each of the elicited words was used to form a simple sentence by the respondents. 1.6.2. Confidentiality Although the data collected so far do not border on confidentiality of the records of the respondents, I assured the respondents that based on their wishes; I will protect their identity in the data they give out. The research records of the respondents, which include their names, may be accessible to interested linguists, particularly phonologists and sociolinguists, historical linguists, and typologists. 1.6.3. Privacy The data collection exercise had nothing to do with the respondents’ privacy. However, I intend to respect and accord all of the respondents their privacy in this research study. University of Ghana http://ugspace.ug.edu.gh 35 1.6.4. Risks and benefits I did not anticipate any identifiable risks associated with this study in terms of physical, social or psychological risks. Similarly, I did not anticipate any direct benefits accruing to the respondents. There were no hazards to respondents. 1.6.5. Compensation Respondents were given prior information that each of them would receive a specified amount of money agreed upon between the respondents and myself as compensation package for their participation in this study. They were made aware that they would receive this package only at the end of the study. 1.6.6. Withdrawal from Study Respondents were made aware that their participation in the study was voluntary; therefore, they could withdraw from participating in the study at any point in time without penalty. Similarly, they were informed that they would not be adversely affected if they declined to participate or later stop participating in the study. Again, they were made aware that their legal representative would be informed in a timely manner if information became available that may be relevant to their willingness to continue participation or withdraw. They were informed their participation might be terminated if they showed lack of commitment to the study. 1.7. Thesis Overview In chapter one, I present the general introduction to the thesis. I also present the information about the languages under consideration in this study. In the same University of Ghana http://ugspace.ug.edu.gh 36 chapter, I provide the problem statement that informed and served as the basis for this research study and also lay out the research questions this study seeks to answer. In chapter two, I review some relevant literature on the subject matter of the present thesis, and also literature on the Optimality Theory, the theoretical framework that this thesis adopts. I discuss relevant literature on the debate over the formal approach or model for loanword adaptation in this chapter. The chapter three looks at the vocalic adaptations of Akan source words in Ga and Dangme. The vocalic adaptations focus on how the three non-native vowels of the Akan source words are adapted in Ga and Dangme. I formalise the discussions within the OT framework in this chapter. In chapter four, I investigate the adaptations of Akan source non-native consonants in Dangme. I focus, specifically, on how source non-native alveo- palatals, palatals and labial-palatalised consonants are adapted in Dangme. Additionally, I discuss how Ga and Dangme speakers realise the labial-velar consonant in a non-permissible word-final position when adapting it from Akan in this chapter. In chapter five, I look at how some Akan source prosodic structures such as the syllable and the nasal feature are adapted in Ga and Dangme. The focus of the chapter is on discussing how Akan source syllable and nasality is realised in Ga and Dangme. For nasalisation, it considers how the source nasalisation rule manifests itself in the borrowing language. The sixth chapter, on the other hand, discusses the adaptations of tone, as prosodic feature from Akan to Ga and Dangme in the adaptation process. In University of Ghana http://ugspace.ug.edu.gh 37 addition to discussing how the Akan source regular tonal melodies of nouns are interpreted in Ga and Dangme, the chapter looks at how Akan’s downstepped H syllable is incorporated into, for example, the three-level tone system of Dangme. In chapter seven, I summarise and conclude all the discussions and analyses done in the present thesis. In the chapter, I present the recommendations to be made in this thesis. 1.8. Summary of the chapter This chapter has provided a general introduction to the whole thesis. It has given overviews of the background information about Akan, Ga, and Dangme in terms of their sound systems (vowels and consonants), syllable structure, and tone. The problem statement, research objectives, and research questions have been presented in this chapter. The research methodology, which includes the data, data source, data collection instruments and techniques, the sample/metadata, and the techniques for the sampling have also been discussed. Another issue that has been talked about in this chapter is the ethical considerations that underpin this study. Finally, issues bordering on the ethical considerations for this study have also been discussed in the chapter. University of Ghana http://ugspace.ug.edu.gh 38 CHAPTER TWO LITERATURE REVIEW 2.1. Introduction In this chapter, I review some relevant existing literature on the subject matter of the thesis. I do a similar review on the theoretical framework to be employed for my analysis, which is Optimality Theory. Before I begin reviewing relevant literature on my theoretical framework, I do a quick overview of how the generative phonology evolved. Phonology, as an aspect of the study of grammar, has evolved over the years. Pāṇini’s 5th century (BC) study of the phonology of Sanskrit is credited as the known maiden attempt towards what is described in present-day terms as generative phonology (McCarthy 2003b, Goldsmith & Laks 2010). In his work, according to McCarthy, Pāṇini posited “explicit rules that relate an abstract representation of words to their actual pronunciation” (McCarthy 2003b: 328). This, according to McCarthy, is what greatly influenced structuralism, pioneered by Leonard Bloomfield, Edward Sapir, Roman Jakobson and Nikolai Trubetskoy in the 1930s, which became the foundation of modern linguistics. In structuralism, the phoneme was recognised as the fundamental unit of the organisation of sound systems. He continues that: “…this unit is a contrastive element by which two words can be distinguished. The phoneme abstracts away from detailed rules of pronunciation to provide a representation of language from which University of Ghana http://ugspace.ug.edu.gh 39 the fully predictable aspects of phonological structure are absent” (McCarthy 2003b: 328). However, it was not until 1968 that contemporary phonology began with the groundbreaking work towards formalising linguistic/phonological phenomena by Chomsky & Halle’s (1968) The Sound Pattern of English (SPE). They also took inspirations from the previous ideas in Pāṇini’s Sanskrit and in structuralism, particularly ones by Sapir. The four main tenets of the SPE, as it was developed, according to McCarthy (2003b), Goldsmith & Laks (2010), among others, are as follows: (i) The theory is generative in the sense that it requires explicit formulation of the rules that relate the underlying representations of the lexicon to the surface representations of actual pronunciation. (ii) The theory must be capable of describing all and only the sound systems and rules that actually occur in languages. Thus, a generative phonology is a model of the mental representation by adult native speakers of the sound structure of a language. (iii) The theory of phonological processes in SPE is a universal theory of how to express rules in terms of a metalanguage. (iv) The idea that phonological rules and representations are characterised in phonetic terms. That is, every speech sound is composed of a set of two-valued distinctive features. (McCarthy 2003b: 328-329, Goldsmith & Laks 2010: 7-8). University of Ghana http://ugspace.ug.edu.gh 40 Some of these principles of the SPE were not without controversies when the programme was first introduced. The most important ones, according to McCarthy were those about abstractness of underlying representations, conspiracies among rules, and rule naturalness. In their attempt to separately address each of these controversies, subsequent phonologists came up with several ideas that gave birth to a number of post-SPE generative phonological theories. These phonological theories are Paul Kiparsky’s Lexical Phonology, which sought to address the abstractness controversy found within the SPE model. One of the main strengths of the Lexical Phonology, according to Goldsmith & Laks (2010: 14), was that it was to offer, “… an effort to synthesise elegantly solutions to a large number of problems, many of which had not hitherto been viewed as directly related”. Later on, two theories, namely Conspiracy by Charles Kisseberth and Natural Phonology by David Stampe also emerged. All these successive phonological theories continued with the tradition of linear representations of phonological phenomena. The quest for non-linear representation of such phonological phenomena as stress and tone later on led to the development of such phonological theories as the Metrical Phonology. This theory was to specifically account for stress phenomena, and the Autosegmental Phonology, which was initially meant to handle specifically tonal phenomena, also emerged. These two theories depart from the predecessor SPE’s linear representation approach. Again, another practice that these two theories broke with their predecessors was that they employed constraints in place of the rules that were used on the representations in the SPE (McCarthy 2003b: 330). Since the feature will be used in the discussions in this University of Ghana http://ugspace.ug.edu.gh 41 present thesis, it is expedient I discuss briefly the Distinctive Feature (henceforth DF) theory in this section. Finally, the Optimality Theory (henceforth OT), also a non-linear and constraint-based theory, emerged to address virtually all the lingering challenges its predecessors had, which each of the post-SPE theories had sought to address separately with varying degree of success. OT espouses the view that all of phonology can be reduced to the interaction of constraints (cf. McCarthy 2003b). In this chapter, I focus my discussions on the last two theories, namely DF and OT in section 2.4. In subsection 2.4.1, I discuss the DF and in 2.4.2, I focus the discussion on the OT. 2.2. What is loanword adaptation? Loanword adaptation refers to the “transformations of foreign words into forms that better conform to the phonotactics of the borrowing language” (Peperkamp & Dupoux 2001: 1, Peperkamp 2004: 341). To throw more light on this, Uffmann (2013) postulates that borrowed words are, “… adapted to the phonological system of the borrowing language, both segmentally, identifying native phoneme categories to express the sounds of the original form, and suprasegmentally, making the borrowings conform to native phonotactics and syllable structure constraints, and adapting them to the native stress or tone system.” (Uffmann 2013: 1). The borrowing phenomenon has been grouped into two main types in phonology. These are what is termed (1) adaptation by Paradis (1996) or on-line adaptations by Kenstowicz & Sohn (2001), among others and (2) loanword (cf. University of Ghana http://ugspace.ug.edu.gh 42 Paradis 1996, Kenstowicz 2003a, Peperkamp 2004) or historical loanword (cf. Peperkamp & Dupoux 2001). Peperkamp (2004), on the other hand, refers to the second type as integrated loanwords, and she defines those words as words that have entered the lexicon of the borrowing language, that have been studied most often. And because of how closely-knitted these forms have been integrated into the lexicon of the L1, she continues, the monolingual speakers who use these loanwords never hear their source forms. There is thus no reason to postulate an underlying form that differs from the surface form in their grammar (Peperkamp 2004: 342). This type of borrowing is what is often termed nativisation or indigenisation, among others. Regarding on-line adaptation, on the other hand, Peperkamp (2004) posits, “…on-line adaptations are foreign words that are borrowed ‘here-and-now” (Op cit 2004: 342). On-line adaptations refer to the borrowed words in their early stages being integrated into the target language. In this study, I classify the phenomenon of Akan source words borrowed into Ga and Dangme as the historical or integrated loanwords, though there is every reason to assume that there are also on-line adaptations as long as the languages involved keep their contacts. This decision is informed by Dakubu’s (2002) opinion that these borrowed words from Akan have been “indigenised” in the borrowing languages. 2.3. Overview of literature on loanword among some Ghanaian languages The mentioning of the loanword phenomenon or linguistic borrowing often induces the idea of incorporating ‘foreign’ words into a target language. In the University of Ghana http://ugspace.ug.edu.gh 43 Ghanaian context, consideration attention is often paid to investigating the borrowing of some European words, particularly English, into one Ghanaian language or another (for Kwa languages, see Nartey (1976) (Adanme; Dangme); Sey (1990), Addo (2002), Abakah & Tenteh (2005), Adomako (2008, 2013), Apenteng (2013), Apenteng & Amfo (2014), among others in Akan; Dzameshie (1996), Agbedor (2006), among others in Ewe; for Gur languages, see Hudu (2002) in Dagbani, Awedoba (1980) in Kasem, among others. This might explain why relatively less attention has been paid to discussing the loanword phenomenon among the two major language sub-families in Ghana, namely Kwa and Gur. Even in those studies, discussion of loanwords from one Ghanaian language into another, often such discussions are not detailed in content. For examples, Dorvlo (2008) briefly talks about the form of some loanwords from Akan (Twi) and Ewe in Logba (Ikpana) when describing the grammar of Logba. There are few exceptional cases, however, where a considerable amount of time has been devoted to discussing the loanword phenomenon among languages from the same genetically-related languages in Ghana. Of such exceptions are studies by Heine (1968) who looked at the influence of Akan and Ewe among some Togo Remnant languages. Koffi (1998) has carried out a similar study though he focused on Akan and Ewe loanwords in Akebu, also a Togo Remnant language. It is not only among the Togo Remnant languages that such studies have been done. Owusu-Adjei (2007) too has looked at some Akan loanwords in Ga. This thesis seeks to contribute to existing literature on borrowings between one Kwa (Tano Central) language, Akan, on the one hand, and Ga and Dangme, also Kwa (Ga-Dangme) languages, on the other hand. University of Ghana http://ugspace.ug.edu.gh 44 2.4. Theoretical Framework There are various phonological theories within the generative phonology that could be employed for the analysis in this study. I, however, have opted for the OT for the simple reason being that it is an attested phonological theory in formalising segmentals as well as suprasegmentals especially prosodic structures such as the syllable, tone, and nasality, which are the focus of this thesis. Since the discussions of Akan source segmental (both vocalic and consonantal) adaptations, and those of the syllable and nasalisation form the core of the chapters three, four and five of the thesis, it will be appropriate to employ this analytical tool in this thesis. I briefly outline some basic tenets of both DF and OT in the following subsections. 2.4.1. Distinctive Feature (DF) theory The feature theory, acording to Hayes (2009), is “…part of a general approach in cognitive science which hypothesizes formal representations of mental phenomena” (Hayes 2009: 70). Hayes (2009) continues that in representing features in phonology, phonological rules of segments are organised according to the phonetic properties of the segments (see also Ladefoged 1993). Therefore, each segment is represented as a bundle of features that collectively defines it (cf. Hayes 2009:71). It is worth noting that the proposal for the feature set has its roots in system of features developed by Roman Jakobson and co. in the 1950s and was later on further developed by Chomsky & Halle’s (1968) SPE. By employing this feature system, we are able to characterise natural classes, i.e. groupings of sounds University of Ghana http://ugspace.ug.edu.gh 45 that appear in phonological rules (Hayes 2009: 74). The features are based mainly on the acoustic and articulatory properties of segments. There are categories of features, namely major class features, place features, manner features, and laryngeal features. The major class features are used to categorise all sounds- vowels and consonant. It is generally assumed that features are binarily valued. That is, either a segment has a particular feature, which is indicated by “+” sign, or it does not have it, which is marked by “-” (cf. Hayes 2009). Since the core of the analyses of this study is about the adaptations of one Akan source feature or another, I will employ the distinctive feature theory. Following Hayes (2009), I employ the major place features labial, coronal and dorsal for consonantal segments in this thesis. For example, for all consonant segments in whose articulation the lips are involved, including labial-palatals, labial-palatalised consonants, labialised consonants are specified for [labial]. 2.4.2. Optimality Theory (OT) OT is a general framework for modelling human linguistic competence that attributes linguistic generalisations to the interplay among ranked and violable constraints (cf. McCarthy 2003a, 2003b, 2007). The theory was originally developed by Prince & Smolensky (1993) and later on McCarthy & Prince (1993a, b). Its primary objective was to address the main challenges in the SPE programme, which had given rise to the emergence of separate phonological theories each of which was meant to address each of these problems, namely conspiracy of rules, naturalness in phonology, among others. University of Ghana http://ugspace.ug.edu.gh 46 It is a linguistic theory that has been applied successfully to aspects or components of linguistics, namely morphology, acquisition, semantics, variation, sociolinguistics, etc. in addition to the primary domains of phonology and syntax (cf. McCarthy 2003a). Again, OT differs from other generative linguistic theories in that it is comparative; that is, it has ranked, violable constraints; and also it is inherently typological, asserting that all variations in languages can be attributed to differences in rankings of sets of constraints (cf. Kager 1999, McCarthy 2003a). Representations in OT, unlike in other non-linear theories, are non- derivational. As has been mentioned earlier, the OT framework employs violable constraints in its approach and according to Kager (1999: xi), “the central idea of OT is that, “… surface forms of language reflect resolutions of conflicts between competing demands or constraints”. The well-formedness of a linguistic expression is determined comparatively in the sense that an expression is well- formed because it is the best among a set of competing candidates (McCarthy 2003a: 211). Furthermore, such a well-formed expression or candidate is ‘optimal’ in the sense that it incurs the least serious violations of a set of violable constraints, ranked in a language-specific hierarchy” (Kager: xi). In talking about ‘hierarchy’, McCarthy observes that, “candidates are compared by a hierarchy of ranked, violable constraints. When two constraints disagree in their assessment of competing candidates, the constraint that is ranked higher is the one that takes precedence.” (op cit 2003a: 211). So, though the optimal candidate or the most harmonic candidate can violate constraint(s), it must, as a matter of necessity, avoid violating higher-ranked constraints, which violation is fatal. Constraints in OT, as Kager (1999: xi) posits, “… directly encode markedness statements and University of Ghana http://ugspace.ug.edu.gh 47 principles enforcing the preservation of contrasts”. This statement clearly indicates the two major types of constraints, namely markedness and faithfulness constraints, which I shall elaborate at the end of this subsection. Some basic assumptions or principles that underlie the OT frameworks are summarised as follows. a. Universality b. Violability c. Domination d. Optimality. OT has a standard architecture that defines the stages it operates in, generating surface forms, which I present below in Figure 1. Input GEN Candidates EVAL Output GEN EVAL Figure 1: The Standard OT Architecture (Adapted from McCarthy 2003a: 211) From Figure 1 above, McCarthy (2003a) explains that the Input component consists of an underlying representation (from phonology perspective) or something from another aspect of the grammar. The GEN(erator) receives the Input and then generates a set of candidates or possibilities with recourse to the Input. These candidates are then supplied to the EVAL(uator) function to select the optimal candidate among the lot generated, which becomes the output forms. In the selection of the optimal candidate, the language-specific constraint hierarchy is University of Ghana http://ugspace.ug.edu.gh 48 applied. In addition to these, we have already mentioned that the constraints; CON in OT, are also universal. It is therefore assumed in the architecture of OT that the Input, the GEN and the CON are all universal, while the Output is language- specific. This assumption is an affirmation of how typological OT is (McCarthy 2003a: 211). In this regard, a language-specific ranking hierarchy when re-ranked can produce the ranking hierarchies of other grammars. This phenomenon is what has been termed in existing literature on OT as factorial typology. McCarthy (2003a) continues that it is assumed in OT that the CON function contains two types of constraints, namely markedness constraints, and faithfulness constraints. Markedness constraints are constraints that ensure structural well-formedness, and, as McCarthy (2003a: 212) posits, they “…assign zero violation-marks to a candidate based on aspects of its output structure, with reference to the input from which it is derived.” That is, in the evaluation of candidates, the EVAL is blind to the Input for the markedness constraints. The faithfulness constraints, on the other hand, try to preserve contrasts in the input in the output forms, and by so doing, according to McCarthy (op cit), “…assign zero violation-marks to the candidate that is identical to the input, they penalise candidates that differ from the input in various respects.” In other words, faithfulness constraints ensure that, as much as possible, the candidates preserve maximal identity with the input. These two types of constraints are always in conflict to derive the output forms. It is my hope that the OT framework would be efficient enough to provide a formal account for the descriptive generalisations that will be made about the University of Ghana http://ugspace.ug.edu.gh 49 adaptation strategies employed by both Ga and Dangme in nativising Akan source words in their lexicons. 2.5. Models/Approaches to loanword adaptation The study of loanword phonology cross-linguistically has led to the development of phonological theories. Though there is a consensus among specialists regarding the assertion that the loanword domain has contributed significantly to linguistic theorisations, yet there seems not to be a consensus as to which theoretical approach or model best accounts for the patterns observed cross-linguistically. Currently, there are two main schools of thought that have emerged in this regard, namely perception-based model, and production-based model and a third model, which is a mixture of these two main approaches, which I have termed the ‘hybrid’ model. In this section, I briefly take an overview of the central ideas of these schools of thought in the theoretical modelling of loanword adaptation phenomenon. I first discuss the production-based model, followed by the perception-based view, and then conclude with the tenets of the hybrid approach. 2.5.1. The production-based model The production-based theorists generally assumed in generative grammar that, “… the input to loanword adaptations is constituted by the surface form of the source language, and that the adaptations are computed by the phonological grammar of the borrowing language” (Peperkamp & Dupoux 2003: 367; Peperkamp 2004: 341). In other words, an output of the donor language (L2) becomes the input to the borrowing language which has to go through some phonological well- University of Ghana http://ugspace.ug.edu.gh 50 formedness adjustments, before it is incorporated into the lexicon of the target languages (L1). Proponents of this view include Silverman (1992), Yip (1993), Paradis (1995), Paradis & LaCharité (1997, 2005); Broselow (2000), Jacobs & Gussenhoven (2000), and Uffmann (2001, 2006, 2007)2. The central idea of this approach, according to Uffmann (2013), is that it is not perceptual similarity that counts in the adaptations process, but phonological equivalence. By phonological equivalence, the phonemes of the target language are identified by the speakers, and these phonemes, which are matched with those of the donor language, are analysed as equivalent. With regard to the relevant role that perception might play in phonological adaptations, Uffmann posits that “in many cases, the perceptual approximation account and the phonological equivalence account converge since phonologically equivalent structures should also be phonetically similar…” (Uffmann 2013: 656). It is not always the case that perception (phonetics) may play any significant role in the adaptation process and, as has been pointed out by the phonological approach, in many cases the adaptations do not go for the closest phonetic match, but rather seem to map sounds that are instead phonologically equivalent (cf. Paradis & LaCharité 1997; LaCharité & Paradis 2002, 2005; Paradis & Tremblay 2009; Uffmann 2013). However, the opponents of the production-based model cite, among other cases, the instances where monolingual speakers of the borrowing languages who use the loanword never hear their source forms, hence positing an underlying form (i.e. the ‘source’ form) that differs from the output form misses psychological 2It is worth pointing out that Uffmann does not explicitly contribute to this theoretical modelling debate per se, however, his studies from cross-linguistic perspective are mostly given purely phonological explanations, hence his inclusion in the proponents of the production-based approach. University of Ghana http://ugspace.ug.edu.gh 51 reality (cf. Peperkamp & Dupoux 2001: 1). The question at this point then is, should monolingual speakers always directly perceive from the source, even when they do not understand the source? Could it also not be possible that bilingual speakers can introduce the borrowed words into the borrowing language, and from there monolingual speakers pick them as their ‘source’? We shall see how this question will be addressed as we proceed further with the reviews. Some of the canonical cases that the proponents of the phonological approach cite to discount perceptual role, are Paradis & LaCharité’s (2002) study of vowel adaptation. They observed that the English lax vowels [ɪ, ʊ], which are phonetically closer to /e, o/ than to /i, u/ in languages that do not have a tense-lax contrast, are rather consistently adapted as /i, u/. This adaptation reflects their phonological status as [+high] vowels, instead of what would have been expected; /e, o/ if perception were to be main factor in the adaptations (see also Uffmann 2013). However, as we shall see in chapters three and four of the present thesis, for example, empirical evidence from the adaptations of non-native vowels in Akan loans in both Ga and Dangme point to the fact that it is possible that both perceptual and phonological accounts might equally be needed to comprehensively account for the model for loanword adaptation. The adapters in both L1 languages seem to go for both the perceptual closeness and the phonological height match, and even in addition to these two strategies, consider the ATR match. As Uffmann (2013) explains about the production-based approach, “a necessary condition for adaptations to reflect the phonological rather than the phonetic status of a segment is that borrowers are aware of the phonological status University of Ghana http://ugspace.ug.edu.gh 52 of the borrowed segment” (Uffmann 2013: 657). This raises the question of who introduces loanword into the L1. To this, it has been argued, especially by Paradis & LaCharité (2002) that loanwords are first introduced by competent bilinguals who have profound knowledge of both phonological systems (i.e. of their L1 and the L2). This view is contrary to the one held by the perceptionists, particularly by Peperkamp & Dupoux (2003) that loanword adaptations require “a relatively low level of competence in the L2 in order for perceptual deafness to play a role” (cf. Uffmann 2013: 657), though a perceptual study by Flege (2003) on fluent L1 and L2 Italian bilinguals shows that their role might be marginal as the perception- based approach claims. Based on this, it is concluded that though the bilingual speakers may borrow words into a speaking community, it is rather the monolingual speakers who adapted those words in the L1 (cf. Uffmann 2013). Having said these, the question that Uffmann (2013: 657-658) then asks, which I paraphrase here, is, if loanword adaptations actually involve the mapping of phonologically corresponding segments, how do speakers, even bilingual speakers, establish this correspondence? Especially if phonologists themselves cannot agree even on a universal set of segment specifications, how can linguistically naive speakers set up correspondence relations between the segments of two languages, if not phonetically? 2.5.2. The perception-based model The perceptionist’s view on loanword adaptation and the role perception plays has been summed up by Peperkamp (2004: 346) that, “concerning the role of perception, it has long been known that the way in which we perceive speech University of Ghana http://ugspace.ug.edu.gh 53 depends upon phonological properties of our native language .... Accordingly, it has been argued that certain loanword adaptations take place during perception, due to the difficulties that listeners have in perceiving non-native sound patterns” (see also Silverman 1992; Yip 1993; Rose 1999; Kenstowicz 2001; etc.). In effect, the perception-based approach (and its offshoot psycholinguistic approach) holds the view that the source words are not the actual inputs to the borrowing languages to which native phonological rules have to apply for their transformation. Transformation of the source words in the adaptation process rather takes place at the exact time perception is taking place (cf. Peperkamp & Dupoux 2001, 2003; Peperkamp 2004; among others). In an attempt to defend the perception-based view of the loanword adaptation, Peperkamp (2004), Peperkamp & Dupoux (2003) argue against the idea that loanword adaptations are computed in the phonology of the borrowing language’s grammar, and rather proposes that, “… a principled solution lies with the hypothesis that all (emphasis theirs) loanword adaptations are phonetically minimal transformations that apply during speech perception” (Peperkamp & Dupoux 2003: 368, Peperkamp 2004: 341-342). In furtherance of her case against postulation of phonological module for loanword adaptations, Peperkamp (2004) posits that: “Indeed, loanword adaptations do not involve synchronic alternations, but rather consist of transformations that are applied only during the introduction of the loanword. Once they have made their way into the borrowing language, there is no reason to keep University of Ghana http://ugspace.ug.edu.gh 54 the corresponding forms in the source language as the underlying forms in the lexicon of the borrowing language. It therefore makes no sense to postulate rules or constraints that apply to loanwords only”. (Peperkamp 2004: 345) According to Uffmann (2013), there seems to be two types of proposals for modelling perceptual adaptation in the phonological grammar, both of which employ OT. One is Steriade’s (2001) P-map proposal. In this proposal, speakers have so-called perceptual maps at their disposal, which encode the perceptual similarity between sounds. Uffmann (2013) continues that speakers then draw upon this knowledge when the grammar requires an unfaithful parse of the input and choose a candidate form that deviates perceptually the least from the input form. The other one by Boersma & Hamann (2009), on the other hand, analyses loanword adaptation within their bidirectional model of phonology. This bidirectional model by Boersma & Hamann then unifies both the production and the perception grammars in one constraint ranking. There are, however, different constraints for perception and those for production, and that loanword adaptation occurs in the perception component of the grammar. This approach, according to Uffmann (op cit), can thus explain mismatches between loanword adaptations and native processes. In pointing out the weakness of both phonological approach and mainstream perception approach, Peperkamp (2004) posits that, “… it appears that arguments in favour of perceptual transformations are sought for adaptations …, which cannot be accommodated within the native (production) grammar. The University of Ghana http://ugspace.ug.edu.gh 55 distinction between perception and production in loanword adaptations is, therefore, ad hoc”. (2004:346). In other words, transformations in loanword adaptations have been considered perceptual because the phonologies of the native grammars could not explain or account for them without empirical evidence from the perception point of view to that effect. She, therefore, provides an alternative explanation in psycholinguistic account in perceptual assimilation. In explaining the role of psycholinguistics and the point at which psycholinguistics comes into adaptation, Peperkamp & Dupoux (2003) postulate that at the point of perceiving those structures: “… all aspects of non-native phonological structure, including segments, suprasegments, and syllable phonotactics, are systematically distorted during speech perception. That is, non- native sound structures are assimilated to ones that are well-formed in the native language, both by monolinguals and by bilinguals” (Peperkamp & Dupoux 2003: 346). Basing the argument on the empirical evidence from results of experiments on how native speakers perceive non-native words to draw parallelism between ‘phonological deafness’ and loanword adaptations, Peperkamp & Dupoux (2003) posit that, “… loanword adaptations are not due to the phonological grammar, but rather, to perceptual processes involved in the decoding of nonnative sounds.”(2003: 367). The argument is then concluded after pointing out the weakness of production-based accounts by Peperkamp (2004) that: University of Ghana http://ugspace.ug.edu.gh 56 “… there is a third phenomenon that is driven by the requirement to respect native phonological structure: during speech perception, the process of phonetic decoding maps nonnative forms onto forms that are in accordance with the native phonology. This process is thus influenced by but not identical to the phonology of the listener’s native language. The perceptual assimilations that result from it are completely automatic and apply beyond the listener’s awareness. Moreover, they are based upon phonetic rather than phonological distance, and in the cases studied so far they correspond to the transformations that take place in loanword adaptations” (Peperkamp 2004: 349, see also Peperkamp & Dupoux 2001). Peperkamp & Dupoux (2001) further point out the problems associated with the phonological view of the model of loanword adaptations. These problems include the learnability of non-native segments and ill-formed syllable structures. That is, the fact that some adaptations are rather phonetically driven, and finally the fact the perception of source words is not faithful as phonological view of the adaptation would assumes that the input to loans is the phonetic representation of the source word. They posit that though perception in adaptation is not faithful, it stands to reason that it is influenced by the speaker’s knowledge of the phonology of the borrowing language. In the subsequent chapters of the present thesis, empirical data from Akan source borrowed words in both Ga and Dangme will be used to test against this claim. University of Ghana http://ugspace.ug.edu.gh 57 2.5.3. The ‘hybrid’ model There is a third intermediate view, which I term ‘hybrid’ in the present thesis. It refers to proposals that draw on explanations from more than one of the main grammar components. The explanations for loanword adaptations posited by the proponents of the view have been attributed to combined factors or facts of either phonetic and phonology, phonetic, phonology and sociolinguistic, etc. This position has been proposed by scholars such as Kenstowicz (2003b, 2005), Heffernan (2005), Rose & Demuth (2006), Yip (2006), Kenstowicz & Suchato (2006), Dohlus (2010), Uffmann (2013), among others. It argues that both phonology and perception (phonetics) play equally important roles in explaining loanword adaptation. Among the proponents of this hybrid view, one school of thought holds the position that all explanations for loans adaptations might not be grounded solely in either of the two main views above, but it could also be from non-grammatical factors. For example, Uffmann (2013) posits additional sociolinguistic factors to play significant role in the adaptation process of loanwords. This is in spite of the fact that the entire loanword adaptation phenomenon fundamentally results from sociolinguistic experiences such as contact situations. The third view, which advocates a combination of factors whereby each of those factors to some extent plays equally important roles in the adaptation process, is becoming the most appealing opinion among loanword researchers. A critical look at the claims by both opposing sides clearly point to the fact neither of the two opposing sides can completely discount the role(s) either side play. University of Ghana http://ugspace.ug.edu.gh 58 However, the main issue of contention seems to be the relative role played by the other model. There are some extremists, however, among each of the sides that will discount completely any significant role by the other side. There have been several instances in loanword adaptations for which comprehensive explanations have required both phonological and perceptual, and sometimes other factors. One of such cases is a recent study by Dohlus (2010) of how the same source target front rounded vowels /ø, œ/ found in both German and French loans are adapted differently by Japanese adapters. According to Dohlus (2010), vowels are adapted as /e/ in loans of German source, while they are adapted as /u/ in those of French source. Though perceptually these vowels are closer to /u/ in native Japanese phonology, there are extralinguistic explanations for the variations in the adaptations of same source segments which has to do with social class, period of entry of the loans, and the domain of use of the loans, among others. Uffmann (2013), on his part, also recommends additional factors such as sociolinguistic (conventionalisation) facts when explaining some loanword adaptation phenomena. After reviewing several instances of loanword adaptations where either phonological approach or perceptual approach or even a combination of both have failed to comprehensively explain them, Uffmann (op cit) concludes with a proposal following Paradis & LaCharité (1997), that the loanword adaptation process or computation “… requires competent bilinguals as agents of change, speakers who are able to assess phonological equivalence between segments as a measure of both phonetic similarity and similar systemic properties, including systems of oppositions and natural class behaviour.” (Uffmann 2013: University of Ghana http://ugspace.ug.edu.gh 59 661). This does not mean the role of perceptual deafness has been relegated to the background, as Uffmann cites empirical evidence from perception experiments to justify itsimportant. From the data on Akan source loans in both Ga and Dangme, I conclude that some of the explanations might transcend the two main approaches and would require additional non-grammatical accounts as some aspects of the hybrid approach seem to suggest. In this study, I test each of these approaches against how loanword corpora of Akan source are borrowed into Ga and Dangme in order to provide evidence for one or the other of these models. 2.6. Summary of the chapter This chapter has reviewed some literature relevant to the present study. In the review, I have briefly talked about phonology in relation to its history and the emergence of its major theories. The chapter has also discussed the positions researchers in the loanword phenomenon have taken regard the best model or approach to account for the phenomenon where I have concluded that for the adaptations of the Akan source loanword in two sister languages, Ga and Dangme, a comprehensive account would require both perceptual and phonological accounts as well as even non-grammatical explanations. I have also briefly given an overview on literature on loanwords among Kwa and Gur languages in Ghana in this chapter. University of Ghana http://ugspace.ug.edu.gh 60 CHAPTER THREE AKAN VOCALIC ADAPTATIONS IN GA AND DANGME 3.1. Introduction In this chapter, I discuss how some Akan source vocalic segments are adapted in Ga and Dangme. This chapter specifically looks at how Akan source non-native vowels are realised in Akan borrowed words in Ga and Dangme. These discussions to be made in the chapter will be formalised within the OT framework. I show that for the adaptations of the Akan source non-native vowels in Ga and Dangme, the speakers of the two target languages consider perceptual closeness, as well as height and ATR equivalences. The implication of this is that both perceptual and phonological factors come into play in the adaptations of those vowels. On the comparison of the patterns of adaptation by both Ga and Dangme, I show that the two languages share a lot in common in their adaptation patterns than they differ in. The rest of the chapter is organised as follows: Section 3.2 discusses the vocalic adaptations from Akan in Ga. In this section, I consider the adaptations of Akan source non-native vocalic segments in Ga. In section 3.3, I discuss vocalic adaptations from Akan in Dangme. In this section, the discussions concentrate on the strategies Dangme speakers employ in adapting Akan source non-native vowels. Finally, I conclude the chapter with a summary of all discussions in section 3.4. University of Ghana http://ugspace.ug.edu.gh 61 3.2. Adaptations of Akan source non-native vowels in Ga It will be recalled that in chapter one, we indicated that there are three vowels present in Akan which cannot be found in the vocalic inventory of Ga. These vowels are [ʊ], [ɪ], and [æ]. Out of these three vowels, two, namely [ʊ] and [ɪ] do not occur in word-initial position. The discussions in this section focus on the context of adaptation, frequency of occurrence of the target vowel, and the model that might have informed the adaptation process. In this subsection, we look at how these non-native vowels are realised in loaned words in Ga that are of Akan source. But before I discuss and analyse the data in this subsection, I show the results of the listening experiment and subsequently discuss how Ga speakers perceive Akan source vowels with focus on the three non-native vowels. This discussion will serve as the basis for the main discussions in this section. University of Ghana http://ugspace.ug.edu.gh 62 Figure 2: Results of Ga speakers listening experiment3 From Figure 2 above, we observe the results of a listening experiment carried out to test how Ga listeners perceive all the ten vowel sounds of Akan each of which was articulated in isolation. On the horizontal (i.e. x) axis are the ten Akan vowels, and the figures on the vertical (i.e. y) axis represent the number of times (frequency) a particular source vowel is perceived as a vowel in the target language. With our focus exclusively on the three Akan source non-native vowels, it can be observed that for an Akan source stimulus [æ], 23 out of the 40 times the Ga listeners perceived [æ] as [ɛ]. This represents approximately 58% of the responses for the perception of [æ]. The other figures are as follows: 8 out of 40 times, representing 20% that the stimulus [æ] was perceived as [e] by the Ga speakers. The remaining figures show that 15% of time (i.e. 6 times), the stimulus 3 In the present study, the following letters of the alphabet represent the following phonetic sounds in the charts: ae = [æ], ec = [ɛ], ic = [ɪ], oc = [ɔ], uc = [ʊ]. University of Ghana http://ugspace.ug.edu.gh 63 was perceived as [u]. The implications of these results are that given any Akan source input word that contains [æ], it is very likely that this vowel will be perceived and hence, realised as [ɛ] by Ga speakers. For the front high stimulus, [ɪ], it is observable from the Figure that 35 out of 40 times, which represents approximately 88% of the responses, the Ga speakers perceived it as [i]. Based on this result, we would expect the vowel [ɪ] in an Akan source word to be realised as [i] most of the times. However, as we shall see later in this chapter, the Ga speakers rather adapt an input [ɪ] as [e] in most cases in the adaptation process. 10% of the respondents perceived the front high vowel stimulus as a back mid [o]. This is unexpected, and as will be shown later in this chapter, such cross frontness adaptation does not actually occur when adapting a loanword of Akan in Ga. The last stimulus perception to be discussed is the high back non-native [ʊ]. From the perception results, 36 out of the 40 responses, the Ga speakers perceived the stimulus vowel as [o]. This represents an overwhelming 90% of the responses to the stimulus. For the remaining 10%, the stimulus was perceived as [u]. From this outcome, it is expected that for an input [ʊ] in an Akan loanword, it is very likely it will be realised as [o] in Ga. With this foreknowledge, it is hypothesised that the adaptations of the three non-native vowels of Akan source in Ga will be based on the following. University of Ghana http://ugspace.ug.edu.gh 64 (10). Assumptions based on Ga speakers’ vowel perception Akan Ga a). [ʊ] → [o] b). [ɪ] → [i] c). [æ] → [ɛ] The assumptions in (10) are based on how the Ga speakers perceive the three source non-native vowels in Ga. Based on the results of the perception experiment, I assume that the Akan source non-native [ʊ], [ɪ], and [æ] will in most cases be realised as [o], [e], and [ɛ] by the Ga speakers when adapting words from Akan. In the rest of this subsection, I test the above assumptions against natural data on how the Ga speakers adapt some words from Akan to their vocabulary. 3.2.1. The adaptations of source [ʊ] It has been observed cross-linguistically that speakers of languages that do not have the tense-lax contrast in their vocalic systems tend to perceive the vocalic segment [ʊ] as [o] when adapting it (cf. LaCharité & Paradis 2002, Flege 2003, Flege & McKay 2004, among others). With Ga identified among such languages that lack this contrast as was seen in chapter one, I examine, in this subsection, to know whether Ga speakers also follow this perception cue when adapting loaned words of Akan source that have this vocalic segment or they do something different. As a convention throughout this thesis, all the segments targeted for the University of Ghana http://ugspace.ug.edu.gh 65 discussions (both vocalic and consonantal) have been underlined in both the source and the target words. 3.2.1.1. Adaptation of source [ʊ] as [o] In the following examples, we see that the Akan source [ʊ] is adapted as [o] by the Ga speakers. This adaptation strategy, it will be explained, is influenced by the speakers’ perception of the source non-native vowel as showed in the results of the perception experiment discussed earlier in this chapter. (11). Akan Ga Gloss i. ɔb̀ʊ̃̀ǹsaḿ́̃ àbòǹsáḿ, àbòǹsáŋ ́ devil ii. àdʊ̀bɛ ́ àdòbɛ ́ raffia palm iii. fʊf́ ʊ́rɔ ̀ àfòfòrò, àfòflò̀ flower, blossom, bud iv. àhʊ́rá àhórà a very serious insult, taunt v. àsàfʊ̀ àsàfò military company; group vi. pʊ ́ k͡pòô, k͡pòó reject, refuse to accept; disown vii. ɔp̀àpʊ ́ k͡pàkp͡ ó male goat, he-goat From the examples in (11) above, it is observable that Ga speakers’ adaptation of the non-native [ʊ], which is specified for [+high, +round, -ATR] in Akan source words follows the cross-linguistic perceptual pattern. Since Ga lacks the tongue University of Ghana http://ugspace.ug.edu.gh 66 root contrast for the source vowel, the speakers go for the phonetically closest vowel, which is [o], in place of [ʊ]. The only phonological similarity between the source [ʊ] and the target [o] is that both share the same value for the feature round, i.e. both are [+round]. Apart from this, the target vowel is a mid vowel and also produced with the tongue root in an advanced position compared to a high and an unadvanced tongue root property of the source vowel. With regard to the context of occurrence, it can be observed that [ʊ] can be adapted as [o] in word-medial position as in (11i) – (11iv), and in word-final position as in (11v) – (11vii). From the data collected, there are 144 instances of the adaptations of Akan source [ʊ]. Out of these instances, on 108 occasions, the source non-native vowel is adapted as [o], constituting 75% of frequency of adaptation. This high frequency of occurrence is in consonance with the outcome of the listening perception test in Figure 2 above, where in most cases the Ga listeners perceived the non-native [ʊ] as [o]. From the discussion above, it can be claimed that wherever a source [ʊ] occurs, it is very likely it will be adapted as [o] in Ga. However, as we shall soon see, this is not the only strategy employed by Ga speakers in the adaptation of this source non-native vowel. In addition to [ʊ] being perceptually adapted as [o], Ga speakers can also adapt the same source vowel with two other vowels as follows. University of Ghana http://ugspace.ug.edu.gh 67 3.2.1.2. Adaptation of source [ʊ] as [u] In the example (12) below, I discuss one of the other vowels with which [ʊ] can be adapted in Ga. That is, the realisation of source [ʊ] as [u] in Ga. We will observe that the source [ʊ] is replaced with a phonologically equivalent high back vowel [u] in Ga. In terms of context, this adaptation strategy occurs only in word-medial position with the exception of only one example of its occurrence word-finally as shall be seen in (12). This exceptional example has even been labelled ‘obsolescent’ in the Ga dictionary. (12). Akan Ga Gloss i. àhʊ̃̀m̀ka ́̃ àhù̃ŋk̀á joy, satisfaction ii. àkʊ̃̀ǹtá àkù̃ǹtá brother-in-law iii. àwʊ́sú àwùsú, birth waters iv. kʊ̀rá kùrá, kùlá preserve, look after v. kʊr̀ àfa ́̃ kùràfa ́̃ brother/sister-in-law vi. ɔb̀ʊ́sʊ̃ḿ́ ákɪt́ɪẃ́ àbúsúmaḱ̃ útŕé chameleon vii. sʊ̀rʊ̀ŋk̀ʊ ́̃ srò̀ŋk̀ṹ extraordinary (obsolesce.) From the frequency count of the data collected for the present study, the source non-native [ʊ] is adapted as [u] on only 17 out of the 144 occasions, representing University of Ghana http://ugspace.ug.edu.gh 68 approximately 12% frequency rate. It can be observed that phonologically the target differs from the source vowel only in its value for the feature [ATR] i.e. it is [+ATR] as against the source [-ATR]. It is worth pointing out that there were very few exceptions in the data in which a source [ʊ] was adapted as [a] as in àjʊ́wá ‘copper, brass; a metal container’ realised as àjáwá in Ga. Hence, positing perceptual factors in this particular adaptation strategy will lack support as empirical evidence from vowel perception experiments across languages with such lack of contrast (cf. LaCharité & Paradis 2002, among others) predicts otherwise. The most logical explanation, therefore, is to postulate phonological factors. As has been mentioned already, the source vocalism is specified for [+high], Ga speakers, therefore, replace this vowel, which is not present in their vowel system with the only back high vowel present in their phonological system; [u]. From acoustic perspective, it can be concluded that the Ga speakers maintain the First Formant (F1), i.e. the height value of the source vocalism in addition to preserving redundantly the Second Formant (F2), i.e. the frontness, of the same in the target vowel. I formalise the phonological adaptation based on height equivalence within the OT framework. As has been discussed above, the back high vowel specified for [-ATR] is absent in Ga, hence whenever the vowel occurs in any word of Akan source, it has to be repaired in Ga in the adaptation process. The constraint that bans this source vowel from occurring in the output is *[+high, -ATR]. As it shall be seen later in this chapter, the non-specification of the frontness property or the value for the feature [round] in the constraint will serve to account for all University of Ghana http://ugspace.ug.edu.gh 69 unadvanced high vowels that are banned in Ga. This markedness constraint is defined as follows: (13). *[+high, -ATR]: A vowel segment specified for [+high, -ATR] is prohibited This constraint when ranked highly will penalise any output form that faithfully preserves the feature values of the non-native input vowel. It has also been observed that the Ga speakers faithfully preserve the value for the feature [round] of the input vowel. Thus, a [+round] source vowel is adapted as [+round] in the output. I define this specialised constraint from the faithfulness constraint family IDENT as follows: (14). IDENT-IO(Rd): Correspondents in input and output have identical value for the feature [round] This constraint, just like *[+high, -ATR], has to be undominated since as we have seen in the two examples discussed so far (and also in the examples to follow), the F2 of the source vowel is most often maintained in the target vowel. Hence, by ranking this constraint higher, it becomes fatal for any output that violates it. Again, as has been seen already and will be discussed further later in this section, preserving the frontness property of the source vowel is to be respected more than preserving either the height or the ATR value. University of Ghana http://ugspace.ug.edu.gh 70 Another specialised constraint from the Identity family to be used in this analysis is IDENT-IO(high). This constraint ensures that the height of the source vocalic segment is maintained in the adapted form. This reflects the phonological match in terms of height between the source and the target vowels. It is defined as follows: (15). IDENT-IO(high): Correspondents in input and output have identical value for the feature [high] The structural well-formedness constraint *[+high, -ATR] dominates IDENT- IO(high) since it is more costly to maintain the prohibited segment specified for [+high, -ATR] than to alter the height of the source vowel in the output forms. However, IDENT-IO(high) outranks MAX-IO(F), a constraint that bans deletion of the non-native segment. MAX-IO(F) is defined as follows: (16). MAX-IO(F): Every segment of the input has a correspondent in the output in terms of a feature (F) From the discussions of the example (12), it has been observed that Ga speakers adapt the source [ʊ] by altering its [ATR] feature value in the output. Therefore, MAX-IO(F) has to rank low so that its violation will have minimal consequence for the optimal candidate, which has to necessarily violate it. University of Ghana http://ugspace.ug.edu.gh 71 Tableau 1: The adaptation of the Akan source [ʊ] as [u] in Ga / kʊ̀rá / *[+ high, - IDENT-IO IDENT-IO ‘preserve’ ATR] (Rd) (high) MAX-IO(F) a.[kʊ̀rá] *! b. [kùrá] * c. [kòrá] *! * d. [k]̀rá] *! * e. [kàrá] *! * * Constraint ranking: *[+ high, -ATR], IDENT-IO (Rd) » IDENT-IO (high) » MAX- IO(F) From Tableau 1, candidate (a), though the most faithful output form to the input becomes the least harmonic candidate for fatally violating the highly-ranked *[+high, -ATR]. Candidate (e) is the next candidate that is not preferred in this Tableau for violating another highly-ranked constraint, IDENT-IO (Rd) for adapting the input rounded vowel as a non-rounded vowel. Also by altering the input high vowel to a low vowel, it additionally violates IDENT-IO (high). This crucially- ranked constraint is also fatally violated by candidates (c) and (d) for adapting the source high vowel as low vowel [o] and [ɔ] respectively in addition to both violating MAX-IO(F). The optimal candidate, candidate (b) is also penalised minimally for adapting the input [-ATR] value as [+ATR] in violation of the lowly-ranked MAX-IO(F). University of Ghana http://ugspace.ug.edu.gh 72 3.2.1.3. Adaptation of source [ʊ] as [ɔ] The discussions so far have focused on an example of what seems to be perceptual adaptation and another example of a phonological adaptation of height. There is yet a third strategy by which the Ga speakers can adapt Akan source non-native [ʊ]. This strategy also seems more phonological than perceptual. I present an illustration of this below in (17). (17). Akan Ga Gloss i. àhʊ́rɔ ́ àhɔĺé sputum, phlegm ii. àkʊ́kɔb́ɛśa ́̃ àkɔḱɔb́ɛśa ́̃ a thorny shrub iii. ànʊ̀kʷádɪ ́ ànɔk̀̃ ʷálé truth iv. ɛm̀ʊ̃ òmɔ ́̃ rice; small waist beads for babies v. ɛs̀ʊ ́̃ sɔ ́̃ blame; responsibility vi. mań́̃ ꜝsʊ ̃́ mań́̃ sɔ ̃ quarrel, strife, ill-feeling; litigation From the examples in (17) above, we observe that the source non-native [ʊ] is adapted as [ɔ] in Ga. The adaptation is observed in word-medial position as can be seen in (17i) - (17iii), and in word-final position as in (17iv) – (17vi). In terms of the frequency of occurrence, from the data collected, Akan source [ʊ] is adapted as [ɔ] by the Ga speakers on 19 out of the 144 of such adaptations, representing University of Ghana http://ugspace.ug.edu.gh 73 approximately 13%. Again, from the perception perspective, though in the vowel space of Ga [ɔ] is closest to [o], the vowel with which we would have expected to be adapted on perceptual grounds, the former differs from the latter in terms of the activities of the tongue root. While the former is produced with an unadvanced tongue root, the latter is articulated with the tongue root in an advanced position. In terms of the feature theory, [ɔ] is specified for [-ATR], the same as the source vowel [ʊ]. Could this, i.e. ATR, have been the main factor that influenced the Ga speakers in the adaptation process? If yes, then the explanation is that there is a mapping of the phonological feature ATR between the source vowel and the target vowel. Incidentally [ɔ] is the only back vowel in Ga that is specified for [- ATR], so it is the only back vowel that matches with the source vocalism in terms of its [ATR] value. In addition to the ATR, both the source vowel and the target vowel share the same F2 property of being back vowels. Based on the explanations above, it can be concluded that the adaptation strategy employed in this pattern is based on phonological (ATR) equivalence. In formalising this adaptation strategy, we employ the entire set of constraints used in 3.2.1.2, save one, that is IDENT-IO(high), which needs to be slightly modified. The obvious reason for the slight modification of this faithfulness constraint stems from the fact that vowel height is irrelevant to the present adaptation strategy. What is relevant is the ATR value. Therefore, we simply change the feature [high] in the constraint to the feature [ATR], and subsequently modify the definition of the constraint in (18) as follows: University of Ghana http://ugspace.ug.edu.gh 74 (18). IDENT-IO(ATR): Correspondents in input and output have identical value for the feature [ATR] In terms of the ranking of constraint in (18), just as we saw in Tableau 1, IDENT- IO(ATR) has to be dominated by both *[+high, -ATR] and IDENT-IO(Rd), but it, in turn, has to outrank MAX-IO(F). This ranking is necessitated by the fact that it is not expected of the optimal output form to be faithful to the input form in preserving all of its features; we observed that the height of the source vowel mismatches that of the target vowel. By its low ranking, the optimal candidate will not be severely penalised when it eventually violates it. Tableau 2: The adaptation of an Akan source [ʊ] as [ɔ] in Ga / mań́̃ ꜝsʊ̃́ / IDENT-IO *[+ high, - IDENT-IO MAX- ‘litigation’ (Rd) ATR] (ATR) IO(F) a.[mań́̃ sʊ̃́] *! b. [mań́̃ só] *! * c. [mań́̃ sú̃] *! d.  [mań́̃ sɔ]́̃ * e. [mań́̃ sa]́̃ *! * Constraint ranking: *[+ high, -ATR], IDENT-IO (Rd) » IDENT-IO (ATR) » MAX- IO(F) From Tableau 2, candidate (e), which alters the roundness property of the input vowel, hence fatally violating the highly-ranked IDENT-IO(Rd), is the least University of Ghana http://ugspace.ug.edu.gh 75 candidate preferred as the output form. It incurs additional violation mark for the lowly-ranked MAX-IO(F) for changing [high] to [low]. Candidate (a), the most faithful output form is also ruled out of the competition for violating fatally the highly-ranked well-formedness constraint against [ʊ], albeit that is the only constraint it violates. Both candidates (b) and (c) follow similar trajectory in losing out on the competition for fatally violating the crucially-ranked IDENT-IO(ATR). Both candidates have altered the input [-ATR] value into [+ATR] in [o] and [u] respectively. Their main competitor, candidate (d), which satisfies this constraint at the expense of minimally violating MAX-IO(F) emerges as the optimal candidate. From the above examples, it could be observed that the Akan source [ʊ] can be adapted as any vocalism specified for the feature [+round]. The adaptations, as we have just seen, could be influenced by either perceptual or phonological factors. 3.2.2. The adaptations of source [ɪ] In this subsection, we discuss how the Akan source non-native [ɪ] is adapted in Ga. Just like [ʊ], Ga does not have the high front unadvanced vowel [ɪ] in its vowel system, and as such this vocalic segment has to be adapted. University of Ghana http://ugspace.ug.edu.gh 76 3.2.2.1. Adaptation of source [ɪ] as [e] From the results of the listening experiment above, it would be predicted that Ga speakers will mostly adapt this non-native vowel as [e], which also follows a cross-linguistic prediction based on perception. In the following examples in (19), this prediction is confirmed in how Akan source words with the non-native [ɪ] are adapted in Ga. (19). Akan Ga Gloss i. dàdɪs̀ɛŋ́̃ ́ dàdèsɛŋ́́ iron pot ii. kòŋk̀òhɪń̃ ɪ ́̃ kòŋk̀òhéné chief fish-monger iii. àɥɪr̀ɛh̀ʊ́ẃ àwèrɛh̀ó grief, sadness iv. àtààdɪ ́ àtàdé, àtàlé dress, garment v. kùmaá̃ ś̃ ɪ ́ kũ̀maá̃ ś̃ é the city of Kumasi vi. kʷáꜝɪ ́ kʷáe thick forest From the examples above, it is observable that Ga speakers adapt the source non- native [ɪ] as [e] when it occurs in word-medial position, as evident in (19i) – (19iii), and in word-final position as found in (19iv) – (19vi). In 90, representing 53% of the occurrences, 171 occurrences of the source non-native vowels in the entries collected for this study, [ɪ] was adapted as [e]. This finding contradicts the University of Ghana http://ugspace.ug.edu.gh 77 results of the perception test carried in which most Ga listeners perceived the stimulus [ɪ] rather as [i]. This is also supported by the findings from the perception test carried out in which an overwhelming 88% of Ga listeners perceived [ɪ] as [e]. This adaptation pattern by Ga speakers would be predicted on perceptual grounds, following what has been found out among languages that do not have this vocalic segment in existing literature. Again, as can be predicted, the adapted vowel shares F2 (i.e. frontness) with the source vowel. That is, both the source and the target vowels are front. Phonologically, the two vowels contrast in terms of height (F1) and ATR, thus while the former is a high vowel, the latter is a mid vowel. Similarly, while the former is specified for [-ATR], the latter has a positive value for the feature. This clearly strengthens the claim that the adaptation strategy employed here seems more perceptual than phonological. In addition to the above adaptation strategy, there are two vowels which the non-native source vocalic segment can be adapted as. I discuss the two adaptation strategies as follows. 3.2.2.2. Adaptation of source [ɪ] as [i] The Akan source non-native [ɪ] can be also adapted by the Ga speakers in some loaned words of Akan source as [i]. I exemplify this strategy as follows. University of Ghana http://ugspace.ug.edu.gh 78 (20). Akan Ga Gloss i. àbɛt̀ɪŋ́,́ àbɛt̀ɪńɪ ́ àbɛt̀íŋ,́ àbɛŋ̀t̀íŋ ́ a very old, tall palm tree ii. àɕɥɪɕ̀ɥɛ ́ àʃʷìʃʷɛ ́ mirror iii. àhɪŋ̀̃ k̀ʷàá àhiŋ̀̃ k̀ʷrà́ King’s servant, attendant iv. ɔk̀ʷàsɪà́ kʷàʃìá stupid person v. àfɪ ́ áfí year vi. ɔb̀àŋʨ̀ɪ ̀ baǹ̀̃ ʧí cassava From the examples in (20) above, it can be observed that [ɪ] is adapted is as [i] in Ga in the context of word-medial position, as in (20i) – (20iv), and in word-final position, as in (20v) – (20vi). Again, from the frequency count, in 54 (i.e. 31%) occasions of out of the 171 instances of the adaptations of [ɪ], it is realised as [i] in Ga. The 31% frequency rate is a marked improvement on the very low rate of 3% frequency Ga listeners perceived the stimulus [ɪ] as [i] in the listening experiment. From the discussions above, one can posit that the explanation for the adaptation strategy employed by Ga speakers will be to phonological factors rather than perceptual ones. That is, the fact that both the source and the target vowels are high, and in feature theory, both share the feature [high]. From acoustic perspective, Ga speakers preserve the F1 of the source vowel in the target vowel, in addition to preserving redundantly the F2 of the same source vowel. The University of Ghana http://ugspace.ug.edu.gh 79 preservation of the height of the source vocalism can be captured and explained within OT as follows. With regard to the set of constraints to be employed in the present analysis, we recall the same set of constraints and the constraint ranking used in accounting for a similar height adaptation used in 3.2.1.2. The only difference between the adaptation process analysed in Tableau 1 and Tableau 3 below is that while the input vocalic segment in Tableau 1 was [+round], the one for Tableau 3 is [- round]. And, despite the contrastive values for the feature round, the constraint IDENT-IO(Rd) can account adequately for the formalisation of the two input vowels in the two Tableaux. Tableau 3: The adaptation of an Akan source [ɪ] as [i] in Ga / àfɪ ́/ *[+ high, - IDENT-IO IDENT-IO ATR] (Rd) (high) MAX-IO(F) ‘year’ a. [àfɪ]́ *! b. [àfí] * c. [àfé] *! * d. [àfá] *! e. [àfú] *! * Constraint ranking: *[+ high, -ATR], IDENT-IO(Rd) » IDENT-IO(high) » MAX- IO(F) From Tableau 3 above, candidate (a) and candidate (e) both become suboptimal candidates for their fatal violations of the highly-ranked constraints *[+high, -ATR] and IDENT-IO(Rd) respectively. Candidate (a) violates the constraint because it University of Ghana http://ugspace.ug.edu.gh 80 preserves the prohibited input vowel segment [ɪ], while candidate (e) also incurs violation mark for altering the frontness property of the input from front to back, which also violates MAX-IO(F). Candidate (c) and candidate (d) equally fatally violate the crucially-ranked constraint that demands the input vocalic segment and the output vocalic segment to share the feature [high]. While candidate (c) adapts the input high vowel as a mid vowel, candidate (d) adapts the same input as a low vowel. The former candidate incurs additional violation mark for MAX-IO(F) for deleting the input [-ATR] feature and replacing it with [+ATR], a constraint also minimally violated by candidate (b), the optimal candidate. 3.2.2.3. Adaptation of source [ɪ] as [ɛ] There is a third strategy by which this source non-native vowel [ɪ] can be adapted by Ga speakers. This is through adapting it as [ɛ], though it is the vowel least expected to adapt [ɪ] as. This assertion is supported by the loan corpus gathered for this study, in which most of the time the source vowel is adapted as [e], followed by [i] ahead of [ɛ] in frequency of occurrence order (S. Rose, personal communication, 28th September, 2017). In the following examples in (21), I exemplify this adaptation strategy. University of Ghana http://ugspace.ug.edu.gh 81 (21). Akan Ga Gloss i. ɔt̀ɪm̀̀̃ pɔŋ̀̃ ́ tɛḿ́pɔŋ́̃ ́ path, way ii. ʨɥɪm̀̀̃ ʧʷɛm̀̀ all together iii. òɡùàǹtɪŋ́̃ ́ ɡʷaǹ̀̃ tɛŋ́̃ ́ sheep; fool iv. àpɪśɔẃ́ àk͡pɛśɛó́, àpɛśɛó́ a small animal, the potto v. àʨɥɛr̀ɪ ́ àʧʷɛr̀ɛ,́ àʧʷɛl̀ɛ ́ fist, blow with the fist vi. ]̀ʨɪà̃ ḿ̃ ɪ ́̃ òʧàámɛ ́ linguist, spokesman for a chief In the examples above, the non-native source [ɪ] is adapted as [ɛ] in word-medial position in words, as seen in (21i) – (21iv), and as in word –final position in (21v) – (21vi). In 27 out of the 171 occurrences of [ɪ] in loaned words of Akan source, it was adapted as [ɛ]. This represents approximately 16% of the occurrences of the vowel. This figure is not too different from the 10% that perceived this source vowel as [ɛ] in the listening test. To explain the adaptation strategy in (21), we observe that the source vowel [ɪ] is closer to [i] in terms of F1 in the acoustic space, however, it is perceptually closer to [e] than it is to [ɛ] on both phonological (height) and perceptual grounds respectively. University of Ghana http://ugspace.ug.edu.gh 82 Therefore, adapting the source high [-ATR] vowel as [ɛ], which is some phonetic distance away from the input, could be motivated by ATR factors, and obviously not by either height or proximity as is possibly with the other two vowels discussed already, namely [e] and [i]. That is, both the source and the target vocalic segments, in this example, have one phonological feature in common, that is, they are both [-ATR], apart from the obvious fact that both are also [+front]. Additionally, a cursory look at the data collected points to the fact that the target vowel [ɛ] very often (i.e. in 21 out of the 27 instances), occurs in the environment of [+nasal] segments. Could this possibly have effect on the adaptation strategy, as it is generally accepted in acoustics that nasals have a suppressing effect on vowels so that nasal or nasalised vowels fall below their oral counterparts in the acoustic space? Hence, could an intended perceptual adaptation of the source [ɪ] by Ga speakers with [e] have been suppressed further down to [ɛ] due to the effect of the neighbouring nasal segments? If yes, then how do we account for the context in which there are no nasal segments in the neighbourhood? Or is it the case of the speakers matching the source front [-ATR] vowel with the only front [-ATR] vowel in Ga in the adaptation process, just as was explained for the adaptation of source [ʊ] as [ɔ] in Ga? In either of the two, the effect of phonology is clearly significant to assume that the adaptation strategy employed here is phonological. This pattern of adaptation can be formalised in the OT as follows. University of Ghana http://ugspace.ug.edu.gh 83 For the analysis in this sub-subsection, we employ the same set of constraints and the same constraint ranking in 3.2.1.3 to account for the phonological adaptation that is driven by ATR equivalence. This decision has been necessitated by fact of the argument made for the immediately preceding analysis. The Tableau analysis is thus presented as follows. Tableau 4: The adaptation of source [ɪ] as [ɛ] / aʥɥɪŋ́ ́/ IDENT-IO *[+ high, - IDENT-IO (ATR) MAX-IO(F) ‘mud-fish’ (Rd) ATR] a. [aʥɥɪŋ́]́ *! b. [aʥɥɛŋ́]́ * c. [aʥɥíŋ]́ *! d. [aʥɥéŋ]́ *! * e. [aʥɥɔŋ́]́ * *! Constraint ranking: *[+high,-ATR], IDENT-IO(Rd) » IDENT-IO(ATR) » MAX-IO(F) From Tableau 4, both candidate (a), the most faithful candidate, and candidate (e) become the least harmonic output forms due to their fatal violations of the two highly-ranked constraints IDENT-IO(Rd) and *[+high,-ATR] respectively. While candidate (a) fatally preserves the prohibited [ɪ], thereby violating *[+high,-ATR], candidate (e) adapts the front vowel as a back vowel in violation of IDENT- IO(Rd). By so doing, the latter also violates MAX-IO(F) in addition. Both candidate (c) and candidate (d) satisfy the two highly-ranked constraints by preserving the front value and also changing the prohibited input vowel, yet they University of Ghana http://ugspace.ug.edu.gh 84 fatally violate the crucially-ranked constraint that ensures correspondence in ATR value between the input and the output forms. Candidate (d) incurs extra violation mark for adapting the input high vowel as a mid vowel in violation of MAX-IO(F). Candidate (b), the optimal candidate also violates MAX-IO(F), albeit minimally, also for adapting the input high vowel as a mid vowel. This analysis seems to be a matter of phonology rather than perception. If it were perceptual, one would have expected the candidate with [e], like candidate (d), to emerge, because as it has been established cross-linguistically, [e] sounds perceptually closest to [ɪ] among the three possible candidates, namely [e], [i], and [ɛ]. The speakers of Ga go for ATR equivalence, and among the three front vowels in Ga, it is only [ɛ] that shares similar ATR value i.e. [-ATR] with the source segment. 3.2.3. The adaptations of source [æ] So far we have discussed the adaptations of two of the three non-native vowels of Akan source by Ga speakers. The third non-native vowel of Akan source whose adaptation we will discuss in this sub-subsection is the low vowel [æ]. This vowel is adapted systematically as [a] in Ga irrespective of the context within which it occurs. 3.2.3.1. Adaptation of source [æ] as [a] In the following examples in (22), I show how this vowel is realised in Akan source words borrowed into Ga. University of Ghana http://ugspace.ug.edu.gh 85 (22). Akan Ga Gloss i. àkǽ̃ńsí àkaŋ́̃ ʃ́iì̃ ̀̃ competition ii. m̀pæ̀̃ǹtú àk͡paǹ̀̃ tú a kind of plantain, for fufu. iii. kǽꜝʥá káʤà the root of a plant growing on ant-hills iv. pæ̀tú pàtú a bird, the owl; the unexpected, sudden v. dæ̀æ̀bí dààbí no vi. dæ̀bídà dàbídà no, emphatic denial vii. kʷæ̀dú àkʷàdú banana As noted above, the source non-native vowel [æ] is invariably adapted as [a], the closest vowel in the vowel space in terms of height and frontness. The only exception in the data is the adaptation of the word bǽɡʷúá ‘any meeting of elders’, which is adapted as bɛɡ̀ʷà *bàɡʷà. Unlike what obtained in example (22), interestingly only 2% of Ga listeners perceived this source vowel as [a] in the listening test. Majority of them (i.e. 58%), rather perceived it as [e], though it is adapted only as [a] by Ga speakers in the adaptation process. It is interesting to note that this source vocalism is sometimes perceived confusingly as [ɛ] even among some native speakers of Akan (especially students study the Akan University of Ghana http://ugspace.ug.edu.gh 86 language) to the extent that they can even represent it orthographically as such. Thus, it would not have been out of place to expect Ga speakers to confuse [æ] as [ɛ] (see Flege & McKay’s (2004) similar report on how Italian speakers confuse English vowel /æ/ and /ɛ/) if their adaptation of this source vowel is perceptually motivated. Indeed, from the listening experiment conducted, few Ga listeners perceived the source [æ] as [ɛ]. However, as seen in the examples in (26) above, this source vowel is systematically realised as [a] in Akan loaned words in Ga. The most plausible explanation for this adaptation strategy seems to reside in the phonology. Phonologically, both the source and the target vowel are equivalent in sharing the feature [+low]. The only feature value that separates the two is the [ATR]; Ga has only the [-ATR] counterpart in its phonological system. This adaptation strategy can be straightforwardly accounted for within the OT theory as shown in the next sub-subsection. We established from the discussion of the examples in (22) above that the illicit source vowel [æ] is invariably realised as [a] in loaned words of Akan source at word-medial position. Since this vowel is not realised at word-final position in the Twi dialect of Akan, no such examples were found in the data. The most plausible explanation for this adaptation strategy seems to reside in the phonology. Phonologically, both the source and the target vowels are equivalent in sharing the feature [+low]. The only feature that separates the two is the [ATR] feature; Ga has only the [-ATR] counterpart in its phonological system. In other words, the University of Ghana http://ugspace.ug.edu.gh 87 source low [+ATR] vowel is adapted as the [-ATR] counterpart available in the Ga phonological system. From the brief discussion above, it seems therefore, that speakers’ adaptation is phonological rather than perceptual. They maintain height equivalence to the source segment. This adaptation strategy can be straightforwardly accounted for within OT framework as shown below. To account for the prohibition of the vocalic segment [æ] in the output, the well-formedness constraint *[+low, +ATR] is introduced which is defined as follows: (23). *[+low, +ATR]: A vocalic segment specified for [+low, +ATR] is prohibited By ranking *[+low, +ATR] constraint highly, it will severely penalise any candidate that preserves this prohibited segment. As we observed from the data in (22), the source [æ] is invariably adapted as [a], both of which differ only in [ATR] value. Therefore, *[+low, +ATR] has to outrank the constraint that demands correspondents between input and outputs in terms of ATR i.e. IDENT- IO(ATR). The lower ranking of IDENT-IO(ATR) will ensure that the imminent violation of it by the optimal candidate will have minimal effect on it. Again, just as has been done in the previous Tableaux analyses, two more constraints are adopted for the present analysis, namely IDENT-IO(Rd) and IDENT-IO(ATR). The former constraint is modified as IDENT-IO(low) to reflect University of Ghana http://ugspace.ug.edu.gh 88 the that height, and not frontness, is the key factor in the adaptation process. This modified constraint is defined as follows: (24). IDENT-IO(low): Correspondents in input and output have identical value for the feature [low] Before we formalise how Ga speakers adapt Akan source non-native [æ], it will be expedient to establish the fact that this vowel in itself is not the input in Akan, but derived in the context of the environment of [+ATR] vowel. In the input in Akan, it is specified for [-ATR] but due to the higher ranking of the constraint AGREE(ATR) over IDENT-IO(ATR) and *[+low,+ATR], candidate with [æ] emerges as the optimal candidate. The AGREE constraint is defined as follows: (25). AGREE(ATR): Segments agree in terms of the laryngeal feature [ATR] AGREE(ATR) dominates IDENT-IO(ATR) and *[+low,+ATR] so that candidates that fail to apply ATR harmony rule where the context is ripe fatally the highly-ranked constraint. This is exemplified in Tableau 5 below. University of Ghana http://ugspace.ug.edu.gh 89 Tableau 5: The realisation of [æ] in Akan / daabi / IDENT-IO AGREE(ATR) *[+low, +ATR] ‘no’ (ATR) a. [daabi] *! b.  [dææbi] * * Constraint ranking: AGREE(ATR) » IDENT-IO(ATR), *[+low, +ATR] Having discussed the realisation of the [+low,+ATR] vowel in Akan, we move to discussing the formalisation of the same vowel in Akan words borrowed into Ga in Tableau 6 below. Tableau 6: The adaptation of the source [æ] as [a] in Ga / dææbi / IDENT-IO *[+low, IDENT-IO (ATR) ‘no’ (low) +ATR] a. [dææbi] *! b. [daabi] * c. [dɛɛbi] *! * d. [deebi] *! Constraint ranking: IDENT-IO(low), *[+low, +ATR] » IDENT-IO(ATR) Tableau 6 shows the adaptation of an illicit source [æ]. As has been discussed in (22) above, the only option in repairing this source illicitness is through faithfully adapting the source height and only altering its ATR value. Therefore, the attempt by candidates (c) to adapt the source vocalism with the nearest vowel on the Ga University of Ghana http://ugspace.ug.edu.gh 90 vowel space, which is [ɛ], fatally violates the highly-ranked IDENT-IO(low). The same candidate violates IDENT-IO(ATR) for changing the source [+ATR] to [- ATR]. The other candidate applies similar strategy of adapting the illicit source [æ] to the nearest vocalic segment is candidate (d). It also fatally violates IDENT- IO(low) just like candidate (c), however, it is better than candidate (c) for remaining faithful to the input by preserving its ATR value. Thus it satisfies the low-ranked IDENT-IO(ATR), though this does not improve its chance for optimality in any way. This strategy, however, results in the candidate fatally violating the highly-ranked constraint that demands that the output shares correspondence in height with the input, that is, IDENT-IO(low). This constraint is also fatally violated by a similar output, candidate (c) that additionally changes the ATR value of the input vowel in violation of IDENT-IO(ATR). Though candidate (b) also violates the latter constraint, its violation is insignificant in deterring it from emerging as the optimal candidate. 3.2.3.2. Summary of discussions From the discussions made thus far, the following interim conclusions are made based on the adaptation strategies employed by the Ga speaker. (26). Source Ga a). [ʊ] → [o], but + [u] and [ɔ] b). [ɪ] → [e], but + [i] and [ɛ] University of Ghana http://ugspace.ug.edu.gh 91 c). [æ] → [a], but never [ɛ] From the discussions, it is clear that neither phonological factors alone nor perceptual factors alone can comprehensively account for the strategies that Ga speakers adopt in the adaptation of non-native vowels of Akan source. For each of [ɪ] and [ʊ] perceptual as well as phonological motivations have been observed to inform how the speakers adapt these vowels to the phonological system of their language. Acoustically, the F2 of the source vowel (i.e. frontness property) is anticipated and perceived by Ga speakers, hence the feature value [+round] is systematically preserved in all the three vocalic options that the source non-native segment is adapted as. A symmetrical interpretation also holds for the adaptation of source non-native vowel [ɪ] as [e], [i] or [ɛ] (S. Rose, personal communication, 28th September, 2017). The perceptual adaptations, as we have discussed, has been from the psycholinguistic point of view, which is espoused by Peperkamp (2004), following Best (1994), stating that “... non-native segments are assimilated to the closest available phonetic category by a phonetic decoding module that is part of the speech perception system” (Peperkamp 2004: 347, see also Peperkamp & Dupoux 2003). Peperkamp & Dupoux (2003: 368) explain further what the ‘closest’ implies by positing that, “… ‘closest’ is defined in terms of either acoustic proximity or proximity in the sense of fine-grained articulatory gestures…”. They continue to discount the relevance of phonological factors by University of Ghana http://ugspace.ug.edu.gh 92 observing that, “… thus, phonological proximity as reflected in the featural structure of segments is irrelevant.”(Peperkamp & Dupoux 2003: 368). The first part of the argument for the perceptual motivations in the adaptation process has been attested in the present thesis with the discussion of how Akan source non-native high vowels; front and back, are adapted as [e] and [o] respectively in Ga. Cross-linguistic studies on perception of non-native vocalic segments have also lend credence to this perception. However, evidence from the adaptation of Akan source non-native vocalic segments in Ga does not seem to support the second part of the argument that seeks to claim that all perception does in the model for the adaptation is by relegating the role phonology can play in the process. As has been discussed already, a source non-native vowel [ɪ] is adapted as three different vowels, namely [e], [i], and [ɛ]. Putting [e] aside, it makes complete phonological sense to postulate that adapting the source [ɪ] as [i] is a straightforward process; it is a matter of faithfully adapting the entire feature matrix of the source save the value for the feature [ATR]. The only contrast (phonological difference) between the source vowel and the target vowel is that, while the former is [-ATR], the latter is [+ATR]. Again, how does perception (‘closest available category’) explain why, say, a source [ɪ] is adapted as [ɛ]. Could [ɛ], a mid vowel, be the closest available category to the source non-native high vowel in Ga? The answer is obviously no! In the vowel space, [ɪ] occurs between [i] and [e] in the source Akan, and it has been claim already in this section that the University of Ghana http://ugspace.ug.edu.gh 93 source being adapted as the latter could well be perceptually defined. However, there seems to be no perceptual motivation for it to be adapted as a vowel further away from it in the vowel space as the perceptionists, especially Peperkamp & Dupoux (2003), Peperkamp (2004), seek to claim. In feature theory, both [ɪ] and [ɛ] differ in the feature [high], as has been explained already. 3.3. Adaptations of Akan source non-native vowels in Dangme In this section, we look at how Dangme speakers also adapt the three non-native Akan source vowels. The section specifically looks at the strategies Dangme speakers employ in adapting non-native vowels. The discussions on Dangme also focus on the context of adaptation, frequency of occurrence, and the model that might have informed the adaptations. In a similar fashion, we will formalise the discussions within the OT. In the discussion of the segmental inventories of the targeted languages, it would be recalled that we realised there are three vowels present in Akan which are not found in the vocalic inventory of Dangme. These non-native vowels, we listed as [ʊ], [ɪ], and [æ]. In this subsection, we discuss in detail how these non- native vowels are realised in some Akan loaned words in Dangme. Just as we did in the previous section, we precede the main discussions with results of the listening experiment conducted and their implications for the analyses. Following the universal perception of the three non-native vowels of Akan source, we would expect them to be adapted in Dangme as follows: University of Ghana http://ugspace.ug.edu.gh 94 Figure 3: Results of Dangme speakers listening experiment From Figure 3 above, just as was done for Figure 2, we observe the results of a listening experiment that was carried out to test Ga listeners’ perception of all the ten vowels of Akan. Here too, in Figure 3, the ten Akan vowels are represented on the horizontal axis, while the number of times an Akan stimulus vowel is perceived as one of the seven oral vowels present in Dangme is on the vertical axis. Again, our focus is on Dangme listeners’ perception of the three Akan source vowels, [ɪ], [ʊ] and [æ]. I begin the analysis with [æ]. It came out from the results that on 38 out of 76 occasions, [æ] was perceived as [ɛ]. This represents 50% of the responses for the perception of [æ]. Surprisingly, on only once, which represents an underwhelming 1.3%, was [æ] perceive as [a]. It came out rather surprisingly that 25 out of the 76 responses was the stimulus perceived as [i]. This figure represents an unexpected 33% frequency rate. As we shall see later in this section, an Akan University of Ghana http://ugspace.ug.edu.gh 95 source [æ] is never adapted as [i] in the adaptation process. I could not immediately explain why such an unusual perception. Perhaps further experimentation has to be carried to ascertain whether or not this pattern will be repeated. With regard to the stimulus [ɪ], it was perceived as [e] on 64 out of 76 occasions. This represents 84% frequency rate. On 7 occasions, which represents 9.2%, [ɪ] was perceived as [i]. The same stimulus was perceived as [ɛ] only once (i.e. 1.3% frequency rate). The stimulus [ʊ] was perceived as [o] on 67 out of 75 times. This represents 89% frequency rate. On the remaining 8 occasions, representing 11%, [ʊ] was perceived as [u] by Dangme listeners. The pattern of the perception of the stimuli by Dangme speakers is similar to that by Ga speakers in the earlier analysis of the perception test. One major difference is that unlike by Ga listeners, 33% of the time, the stimulus [æ] was unexpectedly perceived as [i] by Dangme listeners. The same stimulus [æ], on the other hand, was perceived as [a] rather only once by the same speakers. Following this pattern of perception, the following assumptions are made regarding how Dangme speakers would adapt the stimuli in the process of adapting Akan source words in Dangme. University of Ghana http://ugspace.ug.edu.gh 96 (27). Assumptions based on Dangme speakers’ vowel perception Akan Dangme a). [ʊ] → [o] b). [ɪ] → [e] c). [æ] → [ɛ] In the rest of the subsection, the assumptions above are tested with some empirical data from Akan loanwords in Dangme as follows. 3.3.1. The adaptations of source [ʊ] Just as we did in the preceding section with the hypothesis about the adaptations of non-native vocalic segments based on cross-linguistic patterns (cf. LaCharité & Paradis 2002, Flege 2003, Flege & McKay 2004, among others), we test these hypotheses against the strategies Dangme speakers employ in adapting some words of Akan source to their vocabulary. 3.3.1.1. Adaptation of source [ʊ] as [o] In the following examples, I show that the source [ʊ] can be realised as [o] in Dangme in consonance with the results of the perception experiment among Dangme listeners. University of Ghana http://ugspace.ug.edu.gh 97 (28). Akan Dangme Gloss i. àdʊ̀wá àdòwá royal antelope ii. ànʊ̀kʷárɪ ́ ànòkʷálé fact, truth iii. ɔb̀ʊ̃̀ǹsaḿ́̃ abòsiá̃ ḿ́̃ devil iv. ɔp̀àpʊ̃́ àk͡pàk͡pó, k͡pàk͡pó male goat, he-goat v. àsàfʊ̀ àsàfò group, company, congregation vi. àɥɪr̀ɛh̀ʊ́ àjùrɛh̀ó, àwìlɛh̀ò, grief, sorrow, sadness vii. bàbàsʊ̀ bàbàsò, bàbàsò venereal disease, gonorrhoea From the examples in (28), it can be observed Dangme speakers, just like we saw with Ga speakers, adapt the non-native [ʊ] as [o] in Akan loaned words. The contexts of this adaptation strategy are in word-medial position, as seen in (28i) – (28iii), and in word-final position, as in (28iv) – (28vii). With regard to the frequency of occurrence, out of the 46 occurrences, in 30, which represents approximately 65% of the time, [ʊ] is adapted as [o]. This is similar to the 67 out of 75 occurrences realised in the listening test. The adaptation strategy here follows the cross-linguistic perceptual pattern claimed for similar adaptation in Ga in the preceding section. The explanation, here and there as well, is that since Dangme speakers do not have the source [ʊ] in their sound system, the go for the perceptually closest vocalic segment in their vowel space, [o], in place of [ʊ]. University of Ghana http://ugspace.ug.edu.gh 98 Aside from [ʊ] being adapted as [o], there are two more strategies thorugh which Dangme speakers can adapt this non-native Akan source vowel. 3.3.1.2. Adaptation of source [ʊ] as [u] In the following examples, we discuss how the adaptations are done. I begin the discussions with the example in (29) below. (29). Akan Dangme Gloss i. ɔb̀ʊ̃̀ǹsaḿ́̃ àbùsáḿ, àbùsiá̃ ḿ́̃ devil ii. æ̀ʥìnam̀̃ ʊ̃á̀ àʤlàmũ̀á domestic cat iii. bʊ̀á búá help iv. sìàdɪb̀ʊ́á sìàlèbúá hedgehog v. ɔk̀úḿbɔm̀̀mɔf̀ʊ́ɔ ́ òkúḿ-mɔm̀ɔf̀ūɔ poisonous twig/vine snake vi. fʊt̀ ʊ́ fútŕú treasure In (29) above, we observe that the non-native source [ʊ], which is specified for [+high, -ATR] is replaced with an equivalent high back vowel [u], which is present in Dangme. Unlike in the previous examples, from the data collected, this adaptation strategy occurs in non-final position only. And for the frequency of occurrence, it occurs in 12 out of 46 instances, which represents 26% frequency University of Ghana http://ugspace.ug.edu.gh 99 rate. This is an improvement over the 11% occurrence that Dangme listeners perceived [ʊ] as [u] in the listening experiment. A closer look at the source and the target vowels points to the fact both differ only in their values for the feature ATR; while the former is [-ATR], the latter is [+ATR]. Therefore, positing phonological factors for the adaptation strategy as has been done already that the speakers adapt [ʊ] as the phonological equivalent vowel [u] becomes more straightforward than perceptual factors. In the following Tableau analysis, we formalise this adaptation strategy within the OT. From the discussion above, one can observe striking semblance to the analysis that was made in 3.2.1.2. I, therefore, employ the set of constraints and the ranking that was used for the Ga examples in Tableau 7 below. Tableau 7: The adaptation of the Akan source [ʊ] as [u] in Dangme / bʊ̀á / *[+ high, - IDENT-IO IDENT-IO ‘help’ ATR] (Rd) (high) MAX-IO(F) a. [bʊ̀á] *! b. [ búá] * c. [bóá] *! * d. [b]́á] *! e. [báá] *! * Constraint ranking: *[+ high, -ATR], IDENT-IO (Rd) » IDENT-IO (high) » MAX- IO(F) University of Ghana http://ugspace.ug.edu.gh 100 For preserving the prohibited input vocalic segment [ʊ], candidate (a) incurs fatal violation of the highly-ranked well-formedness constraint *[+high, -ATR], a constraint that bans this segment in the output forms. All the other candidates avoid violating this constraint by altering some feature of the input vowel in some way. However, for adapting this rounded vowel as an unrounded vowel, candidate (e) also incurs fatal violation mark for IDENT-IO(Rd), another highly-ranked constraint in Tableau 6. Also adapting a high vowel as a low vowel, candidate (e) gets extra punishment for violating IDENT-IO(high). The IDENT-IO(high) is also violated by candidate (c) and candidate (d) due to their replacement of the input non-native high vowel with mid vowels [o] and [ɔ] respectively. Their violation of this constraint is fatal due to its high ranking. The former receives extra violation mark for violating the lowly-ranked MAX-IO(F), for adapting a [-ATR] vowel as a [+ATR] vowel. This lowly-ranked constraint is also violated by the eventual winner of the competition, candidate (b), also for the same reason. This constraint being the only constraint candidate (b) violates, its violation does not significantly affect its optimality in the Tableau. 3.3.1.3. Adaptation of source [ʊ] as [ɔ] In the following discussion, I look at the last strategy that Dangme speakers employ in adapting the non-native [ʊ] aside from the two previous strategies. In (30) below, I provide examples of this strategy. University of Ghana http://ugspace.ug.edu.gh 101 (30). Akan Dangme Gloss i. àkʊ́kɔʥ́ań́̃ àkɔḱɔʤ́áí a kind of domestic fowl ii. àkʊ́kɔb́ɛśa ́̃ àkɔḱɔb́ɛśa ́̃ sweet smelling shrub iii. ànʊ̀̃kʷárɪ ́ ànɔk̀̃ ʷálé truth, fact From (30) above, we observe that Dangme speakers can adapt the source non- native [ʊ] as [ɔ] in some loaned words of Akan source. Surprisingly, however, in the listening experiment carried out, none of the 20 Dangme listeners perceived the non-native [ʊ] as [ɔ] as can be seen in Figure 2. In terms of frequency count, this adaptation strategy occurs only 4 times, representing 9% of the total occurrence in data collected. The target [ɔ] differs from the source [ʊ] only in terms of height. In other words, aside from their roundness property, the two vocalic segment share value for the ATR feature. It is therefore concluded that the speakers map the non- native source [-ATR] vowel to the only [-ATR] back vowel in Dangme during the adaptation process. Hence, the model for this adaptation is phonological, i.e. to attain equivalence in the value for the phonological feature [ATR]. This conclusion is in line with a similar pattern observed with the Ga case considered in the preceding subsection. Just as was done for the analysis of the preceding Tableau 7, we use the same set of constraints and the same ranking we employed in the analysis of a symmetrical strategy in Ga in 3.2.1.3 in the formalisation of the pattern discussed above. University of Ghana http://ugspace.ug.edu.gh 102 Tableau 8: The adaptation of the Akan source [ʊ] as [ɔ] in Dangme / ànʊ̃k̀ ʷárɪ ́/ IDENT-IO *[+ high, - IDENT-IO MAX- ‘truth, faith’ (Rd) ATR] (ATR) IO(F) a. [ànʊ̃̀kʷálé] *! b. [ànòkʷálé] *! * c. [ànũ̀kʷálé] *! d.  [ànɔk̀̃ ʷálé] * e. [ànak̀̃ ʷálé] *! * Constraint ranking: IDENT-IO(Rd), *[+high, -ATR] » IDENT-IO(ATR) » MAX- IO(F) So far from the data discussed in both Ga, in the previous subsection, and in Dangme, presently, it has been generally observed that the non-native source [ʊ] is not adapted as [a]. This observation is supported by the outcome of the listening experiment conducted among the Dangme listeners where none of them perceived [ʊ] as [a]. Therefore, for candidate (e) to adapt this non-native input vowel as [a] results in fatal violation of the highly-ranked constraint that demands that an input [+round] vowel has a correspondent [+round] in the output. Similarly, preserving the banned [ʊ] in the output as is done in candidate (a) also results in the fatal violation of the equally highly-ranked*[+high,-ATR], a constraint which prohibits the non-native vowel. For adapting the source [-ATR] vowel as [+ATR] vowels, both candidate (b) and candidate (c) also incur fatal violation marks for crucially-ranked IDENT-IO(ATR). And adapting the input University of Ghana http://ugspace.ug.edu.gh 103 [+high] vowel as a [-high] vowel, candidate (b) is penalised additionally for this offence for violating MAX-IO(F). The MAX-IO(F), which is the lowest-ranked constraint in the present Tableau is also violated by candidate (d), albeit minimally, for also adapting the input [ʊ] as [ɔ]. Due to the low ranking of MAX- IO(F), candidate (d) emerges as the optimal candidate for this Tableau. 3.3.2. The adaptations of source [ɪ] in Dangme In the following subsection, I discuss the strategies Dangme speakers employ in the adaptations of the non-native Akan source [ɪ]. Just as was observed with Ga speakers in the preceding section, Dangme speakers employ different strategies in the process of adapting the Akan source non-native vowel [ɪ]. 3.3.2.1. Adaptation of source [ɪ] as [e] In the following discussions I talk about each of these strategies in the adaptation process. The discussions on Dangme also focus on the context of adaptation, frequency of occurrence, and the model that informed the adaptations. (31). Akan Dangme Gloss i. àdɪ ́ꜝbɔ ́ àdébɔ ́ creation, nature ii. àpɪḿ́̃ àpéḿ a kind of plantain, with thin fingers iii. ʨɪńaḿ́̃ kénaḿ̃ i ́̃ fried fish University of Ghana http://ugspace.ug.edu.gh 104 iv. àdɪ ́ àdé creation, the created world v. ŋk̀áꜝtɪ ́ ákáté groundnut vi. wɔf̀àsɪ ́ āfas̀̃ é sister’s child; matrilateral cousin vii. àmaǹ̃ ɪ ́̃ àmaǹ̃ é death, trouble, calamity viii. àpɪḿ́̃ àk͡pé thousand In (31) above, it is observable that the non-native source [ɪ] is adapted as [e] in word-medial position, as in (31i) - (31iii), and in word-final position, as seen in (31iv) – (31viii). In all, a total of 68 occurrences of such adaptations are identified in the data for this study. Out of the number, on 42 instances the non-native source is adapted as [e]. This represents approximately 62% frequency rate. From the examples above, it can be that [ɪ], being adapted as [e] by Dangme speakers, can be predicted on perceptual grounds as the experiment results have shown. In the listening experiment, an overwhelming 84% of the Dangme listeners perceived this non-native vowel as [e]. In terms of phonology, the only property that the source and the target vowel share is F2 (i.e. frontness). Aside from this property, they have nothing phonological in common. That is, the two vowels contrast in two phonological ways; while the source vowel is high, the target vowel is mid. Similarly, while the source is specified for [-ATR], the target is specified for [+ATR]. Based on these, it stands to reason that the model for this adaptation strategy is perception than it is phonology. University of Ghana http://ugspace.ug.edu.gh 105 3.3.2.2. Adaptation of source [ɪ] as [i] There are other strategies that Dangme speakers employ in adapting the non-native source [ʊ] to their phonological system. In the following examples, I illustrate one of these strategies. (32). Akan Dangme Gloss i. ŋk̀áꜝtɪɛ́ ́ káꜝtíɛ groundnut ii. àɕɥɪɕ̀ɥɛ ́ àhìhʷíɛ,́ àjìhúɛ ́ mirror iii. àɥɪr̀ɛh̀ʊ́ àwīlɛh̀ó grief, sorrow, sadness iv. àmaǹ̃ ɪɛ̀̃ ́̃ àmaǹ̃ íɛ ́ story, case, event, matter, news v. kàɪ ́ káì remind, remember, recollect vi. áḿpaǹ̃ ɪ ́̃ ápāní edible species of bat, esp. rosette bat vii. m̀mɔd̀ɪń́̃ mɔd́í effort, esp. successful effort From the examples in (32) above, we observe that the non-native source [ɪ] is adapted as [i] in both word medial and word-final positions, as can be seen in (32i) - (32iv) and (32v) – (32vii) respectively. From the data, out of the 68 occurrences of the adaptation of this non-native source vocalism, 22 of them, representing 32%, are instances of replacing it with [i]. This figure is an improvement on the University of Ghana http://ugspace.ug.edu.gh 106 only 9% of Dangme speakers who perceived the non-native vowel as [i] in the listening experiment reported above. Just as was done with the Ga adaptation, it can be explained that this adaptation strategy is motivated by phonological factors rather than perceptual ones. The source and the target vowels are both high and front. In acoustic terms, the two vowels share both the F1 and the F2 properties. They differ only in terms of their values for the [ATR] feature. In the following OT analysis, we account for the phonological adaptation of the non-native source vocalism. In formalising the vocalic height adaptation in Dangme, we employ the same set of constraints and the constraint ranking that were used for similar height adaptation strategy in the preceding section for Ga in 3.2.2.2 for the avoidance of repetitions. Tableau 9: The adaptation of Akan source [ɪ] as [i] in Dangme /kàɪ ́/ IDENT-IO *[+high, - IDENT-IO MAX-IO(F) ‘remember, remind’ (Rd) ATR] (high) a. [káɪ]̀ *! b. [káì] * c. [káè] *! * d. [káà] *! * e. [káù] *! * Constraint ranking: IDENT-IO(Rd), *[+high, -ATR] » IDENT-IO(high) » MAX-IO(F) From Tableau 9, candidate (d) and candidate (e) are suboptimal candidates due to their fatal violation of the highly-ranked IDENT-IO(Rd). This violation results University of Ghana http://ugspace.ug.edu.gh 107 from their altering of the input front vowel to central and back vowels respectively. In addition to this, the two candidates are penalised for adapting input high vowel as a low and changing input [-ATR] to [+ATR] respectively. Another candidate that fatally violates another highly-ranked constraint *[+high, -ATR]. The competition is then left between candidate (b) and candidate (c). The latter falls out of the competition due to its fatal violation of the deciding constraint; IDENT- IO(high) for adapting [+high] as [+low]. Candidate (b), the eventual winner avoids violating this crucially-ranked constraint by adapting the input high vowel as a high vowel. However, by altering the input [-ATR] as [+ATR]. Due to the low ranking of the MAX-IO(F), the optimality of candidate (b) is not affected. 3.3.2.3. Adaptation of source [ɪ] as [ɛ] It is not only the two strategies discussed above that Dangme speakers employ in to adapt the Akan source non-native [ɪ]. There is a third way by which the vowel [ɪ] can be adapted in Dangme. In the following examples, I illustrate this adaptation strategy. (33). Akan Dangme Gloss i. àjɪf̀(ʊ)ŕɔ ́ àjɛf̀ĺó, àjɛf̀ĺóò bride, newly married woman ii. àɥɪr̀ɛh̀ʊ́ àwɛr̀ɛh̀ó grief, sorrow, sadness iii. ʥɥɪŋ́̃ ́ ʤʷuɛ reason, think University of Ghana http://ugspace.ug.edu.gh 108 iv. ɔʨ̀ɪà̃ ̃́ꜝ mɪ ́̃ òʧìāꜝmɛ ́ spokesperson, linguist From the examples in (33) above, it is observable that [ɪ] can be adapted as [ɛ] in word-medial position, as seen in (33i) – (33ii), and in word-final position, as in (33iii) – (33iv). This adaptation strategy has a limited occurrence in the data (i.e. 5 out of 68), representing 6%. Just as was explained for a similar strategy by Ga speakers, the source [ɪ] would have been adapted as [i] if vowel height was the main motivation for the adaptation. Similarly, [ɪ] would have been adapted as [e] on perceptual grounds since [ɪ] should be perceptually closer to [e] than any other vowel in Dangme. Therefore, positing either height or perceptual adaptation to explain the strategy might not be as plausible as ATR adaptation would. In either of the two, the effect of phonology is significant enough to assume that the adaptation strategy employed here is phonological. This adaptation pattern can be formalised within OT as shown below. Just as has become a practice in this section, we fall on the constraints that were used for the analysis of a similar adaptation strategy in Ga in the preceding section in 3.2.2.3. In Tableau 9 below, I formalise this in OT. University of Ghana http://ugspace.ug.edu.gh 109 Tableau 10: The adaptation of source [ɪ] as [ɛ] /ɔʨ̀ɪà̃ ́ꜝ̃ mɪ/́̃ IDENT-IO *[+ high, - IDENT-IO MAX-IO(F) ‘spokesperson’ (Rd) ATR] (ATR) a. [òʧɪà̃ ́ꜝ̃ mɪ]́̃ *! b.[òʧìāꜝmɛ]́ * c. [òʧìāꜝmí] *! d. [òʧìāꜝmé] *! * e. [òʧìāꜝmɔ]́ *! * Constraint ranking: IDENT-IO(Rd), *[+high, -ATR] » IDENT-IO(ATR) » MAX- IO(F) From Tableau 10 above, we observe that both candidate (a) and candidate (e) fatally violate the two highly-ranked constraints in the Tableau for preserving a prohibited vowel and adapting a front vowel as a back vowel respectively. And by so doing, candidate (e) incurs violation mark for the lowly-ranked MAX-IO(F). Both candidate (c) and candidate (d) follow the two candidates in becoming suboptimal candidates for also fatally violating the deciding constraint; IDENT- IO(ATR). The violation is due to the fact that they have adapted the input [-ATR] vowel as a vowel with a [+ATR] feature value. For also adapting an input high vowel as a mid vowel, candidate (d) violates MAX-IO(F), a constraint also violated by candidate (b), the optimal candidate, albeit the only violation mark it incurs in the Tableau. University of Ghana http://ugspace.ug.edu.gh 110 3.3.3. The adaptations of source [æ] in Dangme In the following subsection, we look at how the source non-native [æ] is adapted in Dangme. Unlike what was observed in Ga, where this source vowel is systematically adapted as [a] in Dangme, it can be adapted as two different vowels which we discuss below. 3.3.3.1. Adaptation of source [æ]as [a] In the examples in this sub-subsection, I discuss one of the vowels with which the Akan source non-native [æ] can be adapted in Dangme. I show that this adaptation is motivated by phonological equivalence. (34). Akan Dangme Gloss i. àsímæ̀̃sí àsímas̀̃ í so-and-so, what’s-his-name ii. dæ̀æ̀bí dààbí no, negative response iii. dæ̀bídà dàbídà no, emphatic denial iv. m̀pæ̀̃ǹtú kp͡ at̀̃ ú big-fingered Musa paradisiaca v. pæ̀tú pàtú a bird, an owl vi. sæ̀̃ŋk̀ú saŋ̀̃ k̀ṹ hand-piano/organ University of Ghana http://ugspace.ug.edu.gh 111 From the examples above, the source non-native [æ] is adapted as [a] in word- medial position. In terms of frequency of this adaptation, out of the 40 such adaptation, 30 are done as [a], which represents 75% of the occurrences. This figure is in sharp contrast to the only 1 out of the 76 times that Dangme listeners perceived the Akan stimulus [æ] as [a]. The phonological explanation for this adaptation strategy could be that both the source and the target vowel are equivalent in sharing the feature [+low], they differ only in the value for the feature [ATR]. The source vowel is specified for [+ATR], while the target vowel has [-ATR] specification. In formalising the adaptation of [æ] as [a], I adopt the same set of constraint and the ranking employed in 3.2.3.1 for a similar segment in the present analysis. Tableau 11: The adaptation of the Akan source [æ] as [a] in Dangme / sæ̀̃ŋk̀ú / IDENT-IO *[+low, +ATR] IDENT-IO(ATR) ‘hand-piano’ (low) a. [sæ̀̃ŋk̀ú] *! b. [saŋ̀̃ k̀ṹ] * c. [sɛŋ̀k̀ṹ] *! * d. [sèŋk̀ṹ] *! Constraint ranking: IDENT-IO(low), *[+low, +ATR] » IDENT-IO(ATR) From Tableau 11, we observe that the four different candidates employ different strategies in adapting the prohibited input vocalism each of which follow the University of Ghana http://ugspace.ug.edu.gh 112 phonotactics of the language. However, due to the relatively high ranking of IDENT-IO(low), candidate (c) and candidate (d) which adapt input low vowel as mid vowels become suboptimal. Also, for candidate (c) adapting an input [+ATR] as [-ATR], it violates IDENT-IO(ATR) in addition. Candidate (a), which is the most faithful output form is also ruled out for the race of optimality due to its fatal violation of another highly-ranked constraint, for retaining the illicit input vowel [æ], though this is the only constraint it violates. Candidate (b) then emerges as the optimal candidate for minimally violating the lowly-ranked IDENT-IO(ATR) for adapting the input [+ATR] low vowel as [-ATR] low vowel. As it was with Ga, it has also been observed in Dangme that the speakers correctly anticipate the frontness property of the source non-native vowels; hence they almost invariably adapt a non-native front vowel as a front vowel. 3.3.3.2. Adaptation of source [æ] as [ɛ] Aside from the source non-native [æ] being adapted as [a], the same vowel can be realised as [ɛ] in some Akan loaned words in Dangme. I illustrate this in the following examples. (35). Akan Dangme Gloss i. dæ̀æ̀bí dɛb̀í no, negative response ii. dæ̀bídà dɛb̀ídà no, emphatic denial University of Ghana http://ugspace.ug.edu.gh 113 iii. kɔk̀ɔs̀æ̀ʨì kɔk̀ɔs̀ɛʧ̀ì a title for the vulture iv. pàtæ̀kú pɛt́ɛḱú hyena v. àkæ̀ǽtíá àklɛ̀t́īa chimpansee, baboon From the examples in (35), it is can be seen that at the word-medial position, the source [æ] can also be adapted as[ɛ], in addition to [a], in Dangme. Though this adaptation strategy is found in 5 out of the 40 occurrences, representing approximately 13%, it is significant to report that unlike in Ga, the source non- native vowel can be realised as the front mid vowel in Dangme. This adaptation can be explained acoustically that Dangme speakers perceive the low vowel more front than central in non-initial position. It is interesting to note from figure 2 that out of the 76 occasions Dangme speakers listened to the source vowel, exactly half of them perceived it as [ɛ] in the listening experiment conducted. Again, in the vowel space of Akan, [æ] is closer to [ɛ] than it is to [a] in both F1 and F2 values, and this even makes it harder for non-native speakers of Akan and other languages that do not have this vowel to perceive it differently from [ɛ]. It is so even with non-linguist native speakers. A more puzzling case is where the source non-native low vowel is realised as [e] in an exceptional example where akæætia ‘chimpanzee, baboon’ has another variant [àkeetià̃ ] in Dangme. As has already been pointed out, Dangme has five dialects which are divided into two major dialects groups, namely the coastal and University of Ghana http://ugspace.ug.edu.gh 114 the non-coastal dialects. Though it is yet to be confirmed, the source word (the input) could be the Fante dialect variant; akeeʦia, instead of the Twi variant. In that case, it is hypothesised here that the word could have entered Dangme through the coastal Dangme with the Fante variant (i.e. with [e]) as the entry word in those dialects, while the Twi variant (i.e. with [æ]) could have entered through the non- coastal dialects perhaps both happening at the same time. This might explain why the two variations exist in Dangme. In the following subsection, I provide an interim summary for the adaptations of the Akan source non-native vowels in Dangme. 3.3.4. Summary of discussions From all the discussions thus far, the following interim conclusions can be made based on the adaptation of phonological patterns employed by Dangme speakers when adapting segments of Akan source into Dangme. In the following examples, we observe that an Akan source non-native vowel which is rounded is adapted as any of the rounded vowels in Dangme. Similarly, a source non-native front vowel is realised as any of the three front vowels in Dangme. (36). Source Dangme a). [ʊ] → [o], but + [u] and [ɔ] b). [ɪ] → [e], but + [i] and [ɛ] c). [æ] → [a]/[ɛ] University of Ghana http://ugspace.ug.edu.gh 115 From the discussions above, just as was made about similar patterns observed in Ga, it is obvious that phonology or perception alone cannot comprehensively account for the strategies Dangme speakers employ when adapting non-native vowels of Akan source into Dangme. But rather a combination of both perceptual and phonological factors account for such adaptations. However, due to the high ranking of the well-formedness constraints that ban these non-native vowels, the speakers’ adaptations may be motivated by either perception or phonology. And in the latter, factors such as the height and the ATR value of the source vowel may inform the strategy the speakers employ in the adaptation process. And for the perceptual adaptations, I repeat, for emphasis, the explanation provided earlier in the preceding section from the psycholinguistic point of view following Peperkamp & Dupoux (2003), Peperkamp (2004), among others here that “... non-native segments are assimilated to the closest available phonetic category by a phonetic decoding module that is part of the speech perception system” (Peperkamp 2004: 347 following Best 1994; see also Peperkamp & Dupoux 2003). 3.4. Summary of the chapter This chapter has discussed the adaptation strategies employed in borrowing some Akan source vocalic segments by both Ga and Dangme speakers. The discussions have centred on the adaptations of the Akan source non-native vocalic segments in the two target languages. University of Ghana http://ugspace.ug.edu.gh 116 With regard to the model that influences the adaptations by speakers of the two target languages, it was concluded that both perceptual and phonological factors interplay in driving the adaptations. For the perceptual explanation, it was concluded that both Ga and Dangme speakers go for the perceptually closest vocalic segment available in their vocalic inventories in place of the Akan source non-native ones. The other adaptation patterns were also attributed to phonological factors, where the speakers considered the height or the ATR value of the source vowels in their adaptation process. In terms of OT accounts for the adaptations of the source non-native vowels, the conclusion is that it is the crucially high-ranking of the faithfulness constraint IDENT-IO(Rd) over MAX-IO(F), in addition to satisfying the markedness constraints *[+low,+ATR] and *[+high,-ATR], that ensures that the optimal candidate always maintains the source roundness though the height of the input vowel may be altered in the output. It was finally concluded, based on the discussions, that either phonology alone or perception alone cannot comprehensively account for the phonological adaptation patterns both Ga and Dangme speakers employ in incorporating some Akan source segments in their respective languages. As a comparison, it was also observed that there is much resemblance in the patterns that speakers of the two target languages employ in the adaptation processes than they differ in. I briefly compare and contrast the adaptation strategies for the Akan source non-native [ɪ], [ʊ], and [æ] in the following table. University of Ghana http://ugspace.ug.edu.gh 117 Table 11: Adaptations of Akan source non-native vowels in Ga and Dangme Akan source vowel Ga Dangme i. [ɪ] e i ɛ e i ɛ ii. [ʊ] o u ɔ o u ɔ iii. [æ] a a ɛ University of Ghana http://ugspace.ug.edu.gh 118 CHAPTER FOUR AKAN CONSONANTAL ADAPTATIONS IN GA AND DANGME 4.1. Introduction In this chapter, I discuss the adaptations of some Akan source consonants in Ga and Dangme. The chapter specifically looks at how the Akan source labial-palatals and labial-palatalised consonants, which are non-native in both Ga and Dangme, are realised in some Akan borrowed words in Ga and Dangme. The chapter also considers how some Akan source consonants, though present in both Ga and Dangme, yet are adapted due to the restrictions on the distributions in the target languages. The discussions that will be made in this chapter are formalised within the OT framework. In section 4.2, I discuss the adaptations of the labial-palatals and labial- palatalised consonants of Akan source in Ga and Dangme. I look at how and why the Akan source palatal fricative, though present in Ga, yet is still adapted in the same section. In section 4.3, on the other hand, I discuss how these source labial- palatals and labial-palatalised consonants are realised in Dangme. Section 4.4 discusses the adaptations of the illicit source word-final labial-velar fricative [w] in Ga. And in section 4.5, I discuss the how this labial-velar fricative is realised in Dangme at word-final position. It is shown that either delete or vocalic the word- final [w] when repairing this positionally ill-formed segment. Finally, I summarise the discussions of the chapter in section 4.6. University of Ghana http://ugspace.ug.edu.gh 119 4.2. Adaptations of Akan source non-native consonants in Ga In this section, we look at how some Akan source consonants that are not present in the consonantal system of Ga are adapted in Ga. These consonants, as we saw in chapter one, are the labial-palatals in Akan as well as the labial-palatalised consonants, which are realised at the phonetic level of representation. But before we discuss the examples of Akan loaned words with labial-palatal and labial- palatalised consonants, it will be appropriate we discuss first the labial- palatalisation process in Akan to serve as a background to the main discussion to follow. The process and the context of the labial-palatalisation in Akan have been discussed by Abakah (2012), following Ladefoged (1993), as follows: To begin with, when a sequence of [LABIAL] and [CORONAL] vowels follow a set of consonants in Akan, both vowels superimpose their C-Place [LABIAL] and [CORONAL] articulator features on the preceding consonant[,] provided the requisite conditions are met. For instance, when an anterior coronal or the C-placeless h, is followed by a V1V2 of a Ue/Ua sequence, the consonant, if it is t, is labial-palatalised in all the dialects of Akan. (Abakah 2012: 78). Abakah (2012) further proceeds to discuss the dialectal variations in the applications of this rule on some consonants in Akan. With regard to these variations, Abakah (2012) opines that the Asante labial-palatalises all plosives, University of Ghana http://ugspace.ug.edu.gh 120 nasals, and glottal fricative /h/ only when the U in the CUa/e strings is specified for [+ATR], otherwise, it does not labial-palatalises at all. For both the Akuapem and the Fante dialects, however, the process fails to apply only if the C in the string is a bilabial plosive. Unlike the Asante, the [-ATR] feature value of the U does not serve as an exception to the application of the rule. The following examples are taken from Abakah (2012: 78-79) for illustration. (37). UR Fante Akuapem Asante Gloss i. /due/ [dɥie] [dᶣie] [dᶣie] expression of condolence ii. /tue/ [tɥei] [tᶣie] [tᶣie] to puncture iii. /bue/ [bʷei] [bʷue] [bᶣie] to open iv. /pue/ [pʷei] [pʷue] [pᶣie] to go out v. /nuã/ [nᶣiã] [nᶣiã] [nᶣiã] sibling vi. /dua/ [dᶣia] [dᶣia] [dᶣia] tree vii. /tua/ [tᶣia] [tᶣia] [tᶣia] to apply suppository viii. /nʊã̃ / [nɥɪã̃] [nᶣɪã̃] [nʷʊ̃ã] to cook ix. /dʊa/ [dɥɪa] [dᶣɪa] [dʷʊa] to arrest x. /tʊa/ [tɥɪa] [tᶣɪa] [tʷʊa] to join xi. /sʊa/ [sɥɪa] [sᶣɪa] [sʷʊa] to carry University of Ghana http://ugspace.ug.edu.gh 121 From the examples above, we observe that the labial-palatalisation process is an active process in all the three major dialects of Akan. This process can be superimposed on the primary articulation of a consonant, as seen in the examples above, or it can assume a consonantal status as was briefly mentioned in chapter one of this thesis, where we observed from the Ga consonantal inventory that it does not have these “compressed” labialisation process i.e. [ɥ] or [ɥ], but rather it operates the “protrusion” labialisation process, which is represented by [w]. The subsequent subsections look at how these compressed labials or labialised consonants in some Akan source words are realised in Ga. 4.2.1. Adaptations of source non-native consonants in Ga This subsection discusses the strategies Ga speakers adopt in adapting the Akan source compressed labialisation process, which is lacking in their system, when adapting words of Akan source into their vocabulary. 4.2.1.1. Adaptation of source [ʨɥ] as [ʧʷ] The Akan source labial-palatal consonant [ʨɥ] is realised as [ʧʷ] in Akan forms that are adapted in Ga. This is shown in the data below. (38) Akan Ga Gloss i. àʨɥɪr̀ɪ ́ àʧʷèré ladder, staircase ii. òʨɥáǹsa ́̃ òʧʷàǹsa ́̃ drunkard iii. ǹʨɥʊ̀̃ma ́̃ ǹʧʷù̃ma ́̃ red clay University of Ghana http://ugspace.ug.edu.gh 122 iv. ʥɥàrɪ ́ ʤʷálé rinse, wash without soap v. ʥɥa ʤʷrà̀, ʤʷlà̀ break into pieces; sabotage vi. ʥɥòŋk̀ú ʤʷòŋk̀ú hip joint, thigh bone, loins From the examples above, it is clear that the Akan source voiceless labial-palatal affricate [ʨɥ] is systematically adapted as an equivalent labialised voiceless pre- palatal affricate [ʧʷ], as in as seen in (38i) – (38iii). Similarly, in (38iv) – (38vi), the source non-native voiced labial-palatal affricate [ʥɥ] is adapted as a corresponding voiced labialised pre-palatal sound which is present in Ga. It is worth noting that the labialisation rule in the adapted forms in Ga is not triggered by the lip posture of the following vowel. The following vowel can either be rounded, as seen in (38iii) and (38vi) or unrounded, seen in (38i) – (38ii) and (38iv) – (38v). In effect, the source compressed labialisation is straightforwardly realised as a simple protrusion in Ga. Ga speakers match the source labial-palatal segment with phonetically equidistant labialised segment present in their inventory. This segmental matching seems phonological, rather than perceptual. To account for this, we will need a constraint that will ban the labial-palatal [ɥ] in the output forms in Ga. This compressed consonantal segment is specified for [+cor, -ant, +cont]. A constraint to ban this segment in the output will, therefore, be formulated as *[+cor, -ant,+lab]. I define this structural well- formedness constraint as follows. University of Ghana http://ugspace.ug.edu.gh 123 (39). *[+cor,-ant,+lab]: A segment specified for [+cor,-ant,+lab] is prohibited Since the labial-palatal or labial-palatalised consonant is prohibited in Ga, it has to be adapted to some permissible consonant in the process of the adaptation. Therefore, *[+cor,-ant,+lab], the constraint that bans this consonant has to rank high so that any output form that preserves it is severely penalised. The primary articulatory gesture of the source consonant i.e. affrication is preserved in the adapted form. To account for this, there is the need for a constraint that ensures that the source labiality property is preserved in the output form. This faithfulness constraint is IDENT-IO (lab). (40). IDENT-IO(lab): Correspondents in input and output have identical value for the feature [labial] This constraint has to dominate a constraint that will block the deletion of some feature, say MAX-IO(F). So, any candidate that alters a labial-palatal or labial- palatalised consonant to a simple labial consonant violates this consonant for the simple reason that a change in some feature has to be effected to achieve this intended alteration. Once MAX-IO(F) was defined in the previous chapter, we will not define it here again to avoid repetition. University of Ghana http://ugspace.ug.edu.gh 124 Tableau 12: The adaptation of source [ʥɥ] as [ʤʷ] in Ga /ʥɥoŋku/ IDENT–IO *[+cor,-ant,+lab] MAX-IO(F) ‘hip joint’ (lab) a. [ʥɥoŋku] *! b. [ʤʷoŋku] * c. [ʤeŋku] *! * Constraint ranking: IDENT-IO(lab), *[+cor,-ant,+lab] » MAX-IO(F) In Tableau 11 above, we observe that each of the three candidates adapts the input non-native consonant differently. As a strategy to license the deletion of the illicit labial-palatal feature, candidate (c) alters the vowel quality of the input, i.e. the feature value changes from [+round] in the input [o] to [-round] in [e]. Again, it is a change from back vowel to front vowel through a phonological process commonly termed fronting. This change will license the deletion of the labial feature on the consonant. Other than that, dropping the labial diacritic on the consonant when it is preceding a rounded vowel will be unnatural despite the fact that labialised pre-palatals are phonemic in Ga. Though [ʤʷeŋku] is possible in Ga, yet it is not as predictable as [ʤʷoŋku] in terms of labialisation, as it is easier to attribute the rounded vowel [o] in [ʤʷoŋku] as the trigger of the labialisation process than to attribute it to [e] in [ʤʷeŋku]. By employing this strategy, candidate (c) fatally violates the highly-ranked IDENT-IO(lab) for missing the correspondent input feature [lab] in the output. By this same action the same candidate additionally violates the lowly-ranked constraint by deleting the feature University of Ghana http://ugspace.ug.edu.gh 125 [cor]. Candidate (b), on the other hand, faithfully preserves the input features. This preservation is in fatal violation of the equally highly-ranked constraint against the segment *[+cor,-ant,+lab]. The competition is then left with candidate (b). Though candidate (b) violates only one constraint, just as its competitors do, due to the low ranking of the violated constraint, it emerges the optimal candidate for this Tableau. 4.2.1.2. Adaptation of source [ɕɥ] as [ʃʷ] Another Akan source labial-palatal that is adapted in Ga is the voiceless labial- palatal fricative [ɕɥ]. As has already been discussed in the preceding subsection, Ga speakers adapt this source non-native labial-palatal as labialised counterparts. In the following examples I look at how this consonant is adapted in Ga. (41). Akan Ga Gloss i. ɕɥɛd́ɪ ́ ʃʷɛd́é binoculars ii. ɕɥɛɕ́ɥɛ ́ ʃʷɛʃ́ʷɛ ́ a fresh water fish iii. ɕɥìé ʃʷìê, ʃʷìé spill, pour something; fall, pour; to start iv. ɕɥíḿ ʃʷíḿ fast, in a flash v. àɕɥɪɕ̀ɥɛ ́ àʃʷìʃʷɛ ́ mirror vi. ɕɥɪá̃ ɕ́̃ ɥɪá̃ á̃ ́̃ hʷiá̃ h́̃ ʷiá̃ ,́̃ hʷiɛhʷiɛ thin, slender, slim University of Ghana http://ugspace.ug.edu.gh 126 In the examples in (41), it is seen, just as in the previous section, that source labial- palatal fricative is adapted as Ga labialised pre-palatal fricative; the closest consonant in the acoustic space available. This adaptation strategy is straightforwardly predictable based on Ga phonology. The only exception to this generalisation made so far is the exceptional example in (41vi) where contrary to the expected adapted form *ʃʷiá̃ ʃ́̃ ʷiá̃ ́̃ or ʃʷiɛʃʷiɛ, we observe that the source fricative is realised as glottal in the adapted forms though the [labial] feature is preserved. As was observed in chapter one of the present thesis and obviously in the examples above, Ga has the palatal fricative [ʃ]. It will be explained in the following discussion that this seemingly exceptional cases might be better explained by reference to the diachronic facts of Akan consonant system; the palatal fricative developed from the glottal fricative [h]. I offer a detailed explanation for this later in this section. To formalise the discussions in sub-subsection 4.2.1.3, observe the following Tableau for how this adaptation is accounted for. Tableau 13: The adaptation of source [ɕɥ] as [ʃʷ] in Ga /àɕɥɪɕ̀ɥɛ/́ IDENT –IO *[+COR,- MAX-IO(F) ‘mirror’ (LAB) ANT,+LAB] a. [àɕɥɪɕ̀ɥɛ]́ *! b. [àʃʷìʃʷɛ]́ * c. [àʃìʃɛ]́ *!* ** Constraint ranking: IDENT-IO(lab), *[+cor,-ant,+lab] » MAX-IO(F) University of Ghana http://ugspace.ug.edu.gh 127 From the Tableau 12, just as was seen in the preceding analysis, candidate (c), which completely deletes the input non-native labial-palatal feature incurs double fatal violation marks for the high ranking IDENT-IO(lab) for missing the corresponding labial feature. The candidate violates, in addition, the constraint against the deletion of the [cor] feature among the set of the three features that define the input [ɥ], though it satisfies the markedness constraint against this sound. Candidate (a), on the other hand, satisfies the high ranking IDENT-IO(lab) by preserving the input labial-palatal segment. However, it fatally violates equally highly-ranked *[+cor,-ant,+lab] by that action. Though this is the only constraint it violates, yet it loses in the competition for the optimal output form; candidate (b), minimally violates the lowly-ranked MAX-IO(F), for deleting the feature [cor] in the input and preserving the feature [lab]. 4.2.1.3. Adaptation of source [ɥ] as [w] So far we have looked at the adaptations of the labial-palatals [ʨɥ], [ʥɥ], and [ɕɥ]. In the existing literature on Akan, these consonants are assumed to have been derived from the underlying /k/, /ɡ/, and /h/ respectively in Akan through labial- palatalisation. This labial-palatalisation affects yet another consonant in Akan which is the labial-velar glide /w/. This rule changes an underlying /w/, which is a labial-velar approximant, to the labial-palatal approximant [ɥ], whenever the former occurs before a front vocalic segment at the phonetic level of representation. Focusing on their places of articulation, it can be observed that the University of Ghana http://ugspace.ug.edu.gh 128 underlying segment which is articulated at the velum (and also with the lips at the same time) /w/ is fronted to the palatal region, becoming palatal [ɥ], while preserving its labiality property at the phonetic level of representation. In feature theory, this can be expressed as a change in the value of the feature [coronal]. That is, the underlying segment with the feature specification [-coronal] changes to [+coronal] at the phonetic level. In the following examples, we observe that the source labial-palatal consonant is de-palatalised while maintaining the labial feature in the adapted forms. (42). Akan Ga Gloss i. àɥɪr̀ɛf̀í àwèrɛf̀í, wèrɛf̀í forgetfulness ii. àɥɪr̀ɛh̀ʊ́ẃ àwèrɛh̀ó grief, sadness iii. æ̀ɥísíaá̃ ́̃ àwísa,́̃ àwúsa ́̃ orphan iv. ɔɥ̀ɪŕɛ ́ òwélè revenge, vengeance v. àɥɪr̀ɛt̀ɔ ́ òwélètɔɔ̀ ̀ avenging In the examples above, we observe that the source labial-palatal fricative, which is not present in Ga is systematically adapted as the labial-velar consonant [w]. By this adaptation strategy, one account for the adaptation of the source labial–palatal as labial-velar could be that Ga speakers might have adapted the source words in their original forms, that is, when an underlying /w/ have not undergone University of Ghana http://ugspace.ug.edu.gh 129 palatalisation before front vowels in Akan at the time of borrowing them. That is, these words might have entered the Ga vocabulary at the time when, for example, the word for ‘forgetfulness’ was still àwɪr̀ɛf̀í in Akan. Therefore, Ga speaker might not have altered any source consonant in the adaptation process. An alternative explanation could also be that these words might have been borrowed into Ga in their present form. Ga speakers, therefore, had to repair the non-native labial-palatal fricative. In the process of repairing the fricative, they go for the closest consonant in their consonantal inventory taking into an account the place and the manner of articulations of the source segment. The labialised pre- palatal approximant [w] is selected in place of the source [ɥ]. This adaptation pattern is expected as we have already discussed that the source non-native labial- palatal segments are adapted simply as labial segments. In a similar way to how Ga speakers adapt the source labial-palatal affricates In providing a formal account of how Ga speakers adapt the source non- native [ɥ], I employ the same set of constraints and the corresponding ranking as used in the analysis of the source [ʨɥ] for the present one. I formalise the adaptation pattern in the following Tableau. University of Ghana http://ugspace.ug.edu.gh 130 Tableau 14: The adaptation of source [ɥ] as [w] in Ga / àɥɪr̀ɛf̀í / IDENT- *[+cor, -ant, MAX-IO(F) ‘forgetfulness’ IO(LAB) +lab] a. [àɥèrɛf̀í] *! b. [àwèrɛf̀í] * c. [àhèrɛf̀í] *! * Constraint ranking: IDENT-IO(lab), *[+cor,-ant,+lab] » MAX-IO(F) Following a similar pattern as the preceding analyses, we observe from Tableau 13 that candidate (a) faithfully preserves the prohibited labial-palatal segment in fatal violation of highly-ranked *[+cor,-ant,+lab]. In candidate (c), the illicit [ɥ], which is a fricative segment, is repaired through adapting it as [h], another fricative. However, since the target consonant is a non-labial consonant, candidate (c) incurs a fatal mark for violating the highly-ranked constraint that demands correspondent in the output in terms of the [labial] feature, IDENT-IO(lab). Also, for deleting the coronal feature of the input non-native consonant, candidate (c) additionally violates MAX-IO(F). Though candidate (b) also violates MAX-IO(F), it still emerges the optimal candidate due to the low ranking of this constraint, the only one it violates. 4.2.1.4. Adaptation of source derived [Cᶣ] as [Cʷ] The labial-palatalisation process can be superimposed on the primary articulations of consonants, particularly stops and some fricatives in Akan. These consonants, as could be predicted by now, are adapted as an additional [labial] feature University of Ghana http://ugspace.ug.edu.gh 131 superimposed on their primary places of articulation. In the following examples, I show how Akan source words with these labial-palatalised segments are adapted in Ga. (43). Akan Ga Gloss i. bòdᶣíá bòdʷúá horse-tail fly whisk ii. æ̀dᶣìàwá àdʷùàwá fruit iii. okᶣíá òkʷàá farming, agriculture iv. òkᶣìàfʊ́ òkʷààfóɲò farmer v. tᶣìé tʷúè gush out vi. æ̀bùsᶣìá àbùsʷũ̀a ́̃ family; the nuclear family From the examples in (43), it is observable that the Akan source non-native labial- palatalised consonants are adapted as their labialised counterpart present in Ga. This adaptation strategy can be explained that Ga speakers correctly perceive the primary articulations of the Akan source consonants [t, d, k, s], however the additional labial-palatal is rather simply realised as labial (-velar) by the speakers. This pattern is phonological in the sense that the labial-palatalised consonants do not in word neither can they be realised in speech by Ga speakers. Therefore, the speakers go in for the phonologically closest consonants in place of the Akan source non-native ones. In other words, the labial-palatalisation rule, which is active in Akan is absent in the phonology of Ga, and in its place, University of Ghana http://ugspace.ug.edu.gh 132 labialisation applies. For example, all the consonants that can undergo labial- palatalisation such as in /t, d, k, s/ in [tᶣ, dᶣ, kᶣ, sᶣ] respectively in Akan are rather systematically adapted as labialised consonants [tʷ, dʷ, kʷ, sʷ] respectively in Ga. This means these sounds are perceived more back than they are in Akan. Though here we are talking about the adaptations of the labial-palatalised consonants in Akan vis-à-vis those of the labial-palatals, both follow the same process. Therefore, following from what have been discussed in the preceding subsections in this chapter, I employed the constraint and the rankings used for Tableaux for the OT analyses. Tableau 15: The adaptation of source [Cɥ] as [Cw] in Ga / æ̀dᶣìàwá / IDENT- *[+cor,-ant, MAX-IO (F) IO(LAB) +lab] ‘’’fruit’ a. [àdᶣìàwá] *! b. [àdìàwá] *! * c.  [àdʷùàwá] * Constraint ranking: IDENT-IO(lab), *[+cor, -ant,+lab] » MAX-IO(F) From Tableau 14, for employing deletion of the non-native labial-palatal feature in the phonological adaptation process, thus missing a correspondent labial feature in the input, candidate (b) commits fatal violation of the highly-ranked IDENT- IO(lab). The same candidate incurs extra violation mark for MAX-IO(F) for the same offence. Though candidate (a), on the other hand, has the correspondent labial-palatal feature, thus satisfying IDENT-IO(lab), yet it commits fatal violation University of Ghana http://ugspace.ug.edu.gh 133 of *[+cor,-ant,+lab] for preserving the prohibited labial-palatal feature. Though this is the only constraint it violates, it loses out on the race for optimality to candidate (c) that violates minimally the low ranking MAX-IO(F), for dropping the coronal feature in the feature matrix that defines the prohibited input labial- palatal feature. 4.2.1.5. Adaptation of source [ɕ] as [h] Thus far the discussions have focused on how Akan source labial-palatals are adapted in Ga. In the following discussion, we are going to look at how source palatal fricative is sometimes adapted in some Akan loaned words in Ga. As mentioned earlier, source palatal fricative is usually adapted as an equivalent voiceless palatal fricative in Ga with few exceptions which I discuss as follows. (44). Akan Ga Gloss i. ɔd̀ɪɕ́ɪ ́ òdéhè noble, free born person ii. ɛɕ̀ɪ-́dᶣìà òhɛ ́ gum copal; gum copal tree; a bead iii. ɕɛŋ̀̃ ŋ̀ ̀ hɛŋ́̃ ŋ́ ́ flame, colour, bright From the examples above, it can be observed that source palatal fricative [ɕ] is adapted as the glottal fricative [h] though Ga has the palatal fricative [ʃ] in its consonantal system. This adaptation occurs at both the word-medial position, as in (44i) – (44ii) and word-final position, as seen in (44iii). The question then is, could University of Ghana http://ugspace.ug.edu.gh 134 this adaptation be motivated by perceptual factors as it were for the adaptations of some non-native vowels as discussed in chapter three? The answer to this question is that probably may not be. The explanation seems to lie in the diachronic account of the source language. As it has been noted in some existing literature on Akan consonantal phonology (cf. Schachter & Fromkin 1968, Dolphyne 1988/2006, Boadi 1988, Abakah 2012, among others), this palatal fricative is an allophone of an underlying consonant /h/. Before front vowels, this glottal consonant is drawn forward to the palatal region in the oral cavity, which results in the phonological process of palatalisation. For instance, Dolphyne (1988/2006) in particular reports of Stewart’s (1966) data that contains list of words that Koelle collected in some Akan communities in the 19th century that contained the non-palatalised consonants in present-day palatalising environment (cf. Dolphyne 2006: 21). For example, it contained Akan words like gia, present-day gya [ʥa] ‘leave behind’. From the foregone discussion, the above Akan source words are as follows in the underlying representation: ɔd̀ɪh́ɪ,́ ɛh̀ɪ(́-dᶣìà), and hɛŋŋ, which map onto Ga speakers’ òdéhè, òhɛ,́ and hɛŋ́̃ ŋ́ ́ respectively. Another question, therefore, is at what point in time did the adaptation of the glottal fricative take place in Ga? Did these Akan words with the glottal fricative, rather than its palatal allophonic variant find its way into the Ga lexicon? Was it at the time that this consonant had not yet undergone the palatalisation rule? Perhaps, future study may aid in finding answers to the above questions. University of Ghana http://ugspace.ug.edu.gh 135 4.2.1.6. Adaptations of source word-final [w] The last consonantal adaptation that will be discussed in this section is of source word-final [w]. Again, as was observed in sub-subsection 4.2.1.5, the source segment is present in the Ga phonology; however, it is not permitted in word-final position. To repair this ill-formedness, the main strategies employed include vocalisation, elision, and compensatory lengthening. (45). Akan Ga Gloss i. háẃ hàò worry, bother, annoy ii. ɡùàẁ ɡʷàò whip, beat, lash, flog iii. kàẃ káó cake, biscuit iv. tɛẁ̀ tɛò̀ transplant, plant trees v. æ̀níꜝháẃ àníꜝháó laziness vi. aǹ̀̃ tɛẃ́ ań́̃ ꜝtɛó́ ambush, waylay From the examples in (45) above, it can be observed that the word-final labial- velar glide, which is specified for [+lab] is vocalised into [o], also specified for [+lab], thus maintaining the [lab] feature of the source segment. This adaptation is necessitated by the fact that Ga does not permit the labial-velar glide word-finally. Though Ga speakers correctly anticipated the roundness property of the source word-final [w], due to its positional restriction, it is replaced with a rounded University of Ghana http://ugspace.ug.edu.gh 136 vowel. However, what is not clear is the question of why the speakers did not go for the acoustically closest rounded vowel, which is [u] instead of the [o] that it is adapted as?4 Is this adaptation perceptually-motivated as was concluded for the perceptual adaptations of the non-native rounded vowel [ʊ] in Akan source words? In formalising the vocalisation of Akan source word-final [w] in Ga within OT, we will need a constraint that will ban such structurally ill-formed segment in the output form. The constraint to ban the labial-velar glide from occurring at the final position of an adapted word is formulated as *w]σ, and it is defined as follows: (46). *w]σ: Segment [w] is prohibited as the word-final syllable By the relatively high ranking of this constraint, any candidate that faithfully preserves the input word-final [w] incurs fatal violation of it. This structural well- formedness constraint has to outrank a faithfulness constraint that will demand that the output shares correspondence with the input by blocking the consonantal feature [cons] of [w] from being neutralised in the output. The faithfulness constraint to ensure correspondents in both the input and the output for the feature [cons] is IDENT-IO(cons), which I define below. (47). IDENT-IO(Cons): Correspondents in input and output have identical 4 The only exception is in the adaptation of the tiw ‘to follow, pursue, catch up with somebody’ which is adapted as tiu/tuu, sìw ‘suppress; replace, displace;’ ʃìù, asitíw ‘making an error due to intentionally carelessness’ àʃìtíú. University of Ghana http://ugspace.ug.edu.gh 137 value for the feature [consonantal] This constraint will be violated by any candidate that alters the source [w] to [o] though the same candidate will satisfy the other Identity constraint, IDENT- IO(lab) by preserving the labial feature in the target segment. IDENT-IO(cons) has to rank low so that the optimal candidate that has to necessarily violate it will do so minimally. In the following Tableau, I present the formalisation of the adaptation of the illicit input word-final [w] through vocalisation. Tableau 16: The adaptation of source word-final [w] as [o] in Ga / hàẁ / IDENT-IO IDENT-IO MAX-IO ‘worry, bother’ (lab) *w]σ (Cons) a. [hàẁ] *! b. [hàò] * c. [hà] *! Constraint ranking: IDENT-IO(lab), *w]…σ » MAX-IO » IDENT-IO(Cons) From Tableau 15, the most faithful output form; candidate (a), becomes the least harmonic candidate in the present analysis due to its fatal violation of the highly- ranked constraint against word-final labial-velar approximant *w]σ. Candidate (c), on the other hand, avoids violating this position-sensitive constraint through the deletion of the prohibited input word-final [w]. However, by employing deletion as the repair mechanism, it in turn fatally violates the general constraint against University of Ghana http://ugspace.ug.edu.gh 138 deletion, MAX-IO. This violation becomes fatal due to the crucial ranking of the faithfulness constraint. Though candidate (b) is the only output form that violates the constraint against neutralising the consonantal feature of the final [w] through vocalisation, yet it emerges as the optimal candidate due to the lower ranking of the violated constraint. So in effect, Ga speakers will employ the vocalisation of the illicit source word-final [w] into [o] in this Tableau analysis as an adaptation strategy. Aside from vocalising the ill-formed source word-final [w], the same ill- formedness can be repaired by eliding the source segment. The elision seems to take place only and only if the remainder of the syllable of the adapted words is minimally disyllabic. In the following examples, I illustrate how the elision of the source word-final [w] takes place in the adapted forms. (48). Akan Ga Gloss i. èkúẁ àkú heap ii. àsɛm̀̀sùróẃ àsɛm̀̀̃ srò̀ fear of trouble iii. àsrà́àfʊ́kúẃ àsrà́àfóàkú regiment iv. hùrúẃ hrù́, hlù́ jump, jump about v. wòsóẃ hósò shake, quake vi. àɥɪr̀ɛh̀ʊ́ẃ àwèrɛh̀ó grief, sadness University of Ghana http://ugspace.ug.edu.gh 139 From (48), it is observable that source word-final [w] is dropped in the adapted forms, unlike in the previous examples in (45) where it was vocalised into [o]. It is worth noting that as was earlier pointed out, the condition for the elision is that the rest of the word (i.e. the adapted form) should have at least two syllables. There is an exception to this generalisation in the Akan source word ɛpɔẃ ‘lump, bump; knot, knot in a mesh’, which is adapted as k͡pɔ,́ and not *k͡pɔɔ́,́ *k͡pɔó; ʥàw ‘worship, revere’, which is realised as ʤá, and not *ʤáa or *ʤáo. The phonological implication of this is that in the constraint-based analysis, the constraint that bans final [w] is more important than the one that demands disyllabic minimality. Therefore, in constraint ranking, the constraint on the former should outrank that of the latter. To account for the deletion of the illicit word-final [w], we will employ the same set of constraints as used for the analyses in sub-subsetion 4.2.1.1. However, since the strategy is deletion, the anti-deletion constraint, MAX-IO has to be demoted so that the eventual winner of the analysis will violate it minimally. Therefore, the last two constraints of the preceding constraint ranking in Tableau 15 is re-ranked in Tableau 16 as follows. University of Ghana http://ugspace.ug.edu.gh 140 Tableau 17: The adaptation of source word-final [w] as [ø] in Ga / àɥɪr̀ɛh̀ʊ́ẃ / IDENT-IO IDENT-IO MAX-IO *w] ‘grief, sadness’ (lab) σ (Cons) a. [àwèrɛh̀óẃ] *! b. [àwèrɛh̀ó] * c. [àwèrɛh̀óó] *! Constraint ranking: IDENT-IO(lab), *w]…σ » IDENT-IO(Cons) » MAX-IO Tableau 16 shows that each of the three candidates employs different repair strategies for the illicit input word-final [w]. Candidate (a) preserves the illicit source word-final segment, which action incurs fatal violation mark for the highly- ranked position-sensitive *w]σ. Candidate (c), on the other hand, employs vocalisation of the illicit labial-velar glide at the final position. Therefore, it surfaces as [o] in the output. It is worth noting that as we shall see in the following discussion; this strategy could also be explained as resulting from compensatory lengthening. By so doing, candidate (c) also violates fatally IDENT-IO(cons) for adapting an input word-final [+cons] as a [-cons] in the output. For completely deleting the illicit input word-final glide, candidate (b) violates MAX-IO, though it still emerges the preferred output form due to the low ranking of MAX-IO in the present Tableau. Though there is no correspondence between the input final [w] and the output, candidate (b) is not penalised for this lack of correspondence. The final major adaptation strategy employed for the source word-final [w] is a two-step process, namely elision, and subsequent compensatory lengthening University of Ghana http://ugspace.ug.edu.gh 141 which is driven by the disyllabic minimality requirement discussed in the preceding examples. I exemplify this in the following examples. (49). Akan Ga Gloss i. dòẃ dòò be browned by roasting ii. ɔd̀áẁ dáà mouth, mouth cavity iii. sɔẁ̀ sɔɔ̀ ̀ catch from below; catch, with a flourish iv. kúẃ kúú group, council, society v. ɡʷùàẁ ɡʷàà whip, beat, lash, flog vi. bɪr̀ɛẁ̀ blɛ̀ò̀ò slow, soft, slight From the above examples, it can be observed that the primary strategy employed in repairing the ill-formed source word-final [w] is deletion or elision. However, due to the disyllabicity minimality requirement imposed on such words, the source vocalism seems to be lengthened to satisfy this requirement or constraint to avoid such ill-formed outputs as *dò < dòẃ, *dá < ɔd̀áẁ, and *kú < kúẃ. Below we consider how the elision and its subsequent compensatory lengthening are accounted for within the OT. In accounting for the minimum loaned form to be disyllabic after the deletion of the illicit source word-final [w], which triggers compensatory lengthening, we will need a constraint that will University of Ghana http://ugspace.ug.edu.gh 142 ensure that a disyllabic input form will map onto a disyllabic output form. This faithfulness constraint is IDENT-IO[σ], which is defined as follows. (50) IDENT-IO[σ] - There must be identity between I and O in terms of syllable structure This constraint has to dominate the anti-deletion constraint, MAX-IO so that deletion of the illicit source word-final will attract lesser punishment, however, for not applying the compensatory lengthening, the candidate will violate this crucially-ranked constraint. In the Tableau below, I present the formalisation of this adaptation strategy. Tableau 18: The adaptation of source word-final [w] through compensatory lengthening in Ga / kúẃ / IDENT-IO IDENT-IO IDENT-IO MAX-IO (lab) *w]σ [σ] ‘group, council’ (Cons) a. [kúẃ] *! b. [kúú] * c. [kú] *! * Constraint ranking: IDENT-IO(lab), *w]…σ » IDENT-IO[σ]» MAX-IO» IDENT-IO (Cons) From Tableau 17, candidate (a) preserves the input word-final [w], which is in fatal violation of the position-sensitive constraint *w]σ. Candidate (c), on the other hand, though employs the deletion of the illicit input word-final, yet it fatally University of Ghana http://ugspace.ug.edu.gh 143 violates the crucially-ranked IDENT-IO[σ] for missing a correspondent input syllable. This same candidate is again penalised for deleting the final [w] in violation of the anti-deletion MAX-IO. For altering the input word-final [w], candidate (b) incurs a violation mark for IDENT-IO(cons), a constraint that ensures input and output forms correspond in terms of value for the feature [consonantal]. In this section, I have discussed the adaptations of the Akan source non- native labial-palatals and labial-palatalised consonants in Ga. I have also discussed the strategies Ga speakers employ in adapting the labial-velar glide in word-final position in words that are borrowed from Akan. It has come out that there is a high-ranking constraint that prohibits the labial-velar glide [w] from occurring at word-final position in Ga. To repair Akan source word-final labial-velar glide in the adaptation process, word-final [w] is either deleted or it is vocalised into [o]. There can also be compensatory lengthening, which results usually to ensure that rest of the word after the deletion of the final [w] is minimally disyllabic, though there are few exceptions to this generalisation that were found in the data. 4.3. Adaptations of Akan source non-native consonants in Dangme In this section, I present the strategies that Dangme speakers also adopt when adapting non-native consonants from Akan. The section also discusses the adaptation of Akan source word-final [w] in Dangme. The discussions are formalised within the OT. University of Ghana http://ugspace.ug.edu.gh 144 4.3.1. Adaptations of Akan source labial-palatals in Dangme In the following sub-subsections, we take a look at the adaptations of the Akan source labial-palatals as well as the labial-palatalised consonants, which segments are not present in Dangme, as was shown in chapter one. With our foreknowledge of the labialisation process from the preceding section, I discuss how these labial- palatal and labial-palatalised consonants are adapted in Dangme. 4.3.1.1. Adaptation of source [ʥɥ] as [ʤʷ] In adapting the Akan source non-native [ʥɥ], I show that just as was observed in Ga, Dangme speakers preserve the source [+labial] feature in the adapted forms though the primary source consonant is not labialised. (51). Akan Dangme Gloss i. ʥɥòŋk̀ú ɡ͡bʷòǹɡú thigh bone, hip joint ii. æ̀ʥɥúma ́̃ àʤʷṹma ́̃ work, labour iii. æ̀ʥɥúmaʥ́̃ ãń àʤʷṹmaʤ́̃ aŋ́̃ ́ unprofitable labour, a wild goose chase iv. ʥɥɪń́̃ ʤʷũ̀ɛ ̀̃ reason, think v. æ̀ʥɥàmań́̃ àʤʷùam̀̃ a ́̃ prostitute; fornication University of Ghana http://ugspace.ug.edu.gh 145 From the example (51), it can be observed that the Akan source non-native voiced labial-palatal affricate is adapted as phonetically close equivalent sounds in Dangme, namely the voiced pre-palatal affricate [ʤ] as in (51ii) – (51v), and as a voiced labial-velar stop [ɡ͡b] as in (51i). As we saw in chapter one on the consonantal inventory of Dangme, unlike in Ga, Dangme does not have labialised consonants. However, once the speakers anticipate lip protrusion, they create the protrusion labialisation by ensuring that both the target pre-palatal and the labial- velar are immediately followed by rounded vowels, though that may not be the case as can be seen in (51iv) – (51v). In these two examples, the labial-palatals in the Akan source are followed by front and central vowels respectively vis-à-vis the back (rounded) vowel [u] in Dangme. In section 4.2, we observed that Ga speakers adopt the strategy in adapting the source non-native voiced labial-palatal affricate which is very similar to how Dangme speakers adapt this same source consonant. Therefore, we employ the same constraint set and the ranking used for the similar analysis as previously (in 4.2.1.1) in the present analysis. Tableau 19: The adaptation of the source non-native [ʥɥ] as [ʤʷ] in Dangme / æ̀ʥɥúma ́̃ / *[+cor, - IDENT- MAX-IO(F) ‘work, labour’ ant,+lab] IO(lab) a. [àʥɥúma]́̃ *! b. [àʤʷṹma]́̃ * c. [àʤíma]́̃ *! * Constraint ranking: IDENT-IO(lab), *[+cor,-ant,+lab] » MAX-IO(F) University of Ghana http://ugspace.ug.edu.gh 146 In Tableau 18, each of the three candidates repair the source non-native labial- palatal differently. Candidate (c) employs deletion of the non-native segment as the repair strategy in fatal violation of the highly-ranked IDENT-IO(lab), for adapting the source [+lab] as a [-lab] consonant. By deleting the labial feature, the same candidate violates the lowly-ranked MAX-IO(F) in addition. Candidate (a), on the other hand, preserves the [lab] feature by preserving the input prohibited consonant which leads to its fatal violation of the crucially-ranked constraint against the input labial-palatal, *[+cor,-ant,+lab], though this is the only constraint it violates. Candidate (c) preserves the [lab] feature, but rather drops the [cor] feature in the input, thereby satisfying the two highly-ranked IDENT-IO(lab) and *[+cor, -ant,+lab] respectively. However, for deleting the [cor] feature, it incurs violation mark for the anti-deletion constraint, MAX-IO(F). Despite this violation, candidate (c) emerges the harmonic output form in this Tableau due to the low ranking of MAX-IO(F). 4.3.1.2. Adaptation of source [ʨɥ] Aside from the adaptation of the Akan source non-native voiced labial-palatal affricate, the voiced labial-palatal affricate can also be adapted in Dangme, just as is done in Ga. In the following very limited examples, I show the adaptation of this source non-native consonant. University of Ghana http://ugspace.ug.edu.gh 147 (52). Akan Dangme Gloss i. àʨɥɪr̀ɪ ́ àʧré ladle ii. ʨɥɪǹ̃ ɪs̀̃ iń́̃ tʷōìsíń a kind of drum and its music, used for chiefs From the examples above, it can be observed that the Akan source non-native voiceless labial-palatal affricate is adapted as two different consonants, namely a voiceless pre-palatal affricate as seen in (52i) or with an alveolar stop as in (52ii). Again, whereas the source [lab] is preserved in the latter, it is dropped in the former. These adaptation strategies differ from what we observed in Ga where the source was systematically adapted as the labialised pre-palatal affricate present in the language. Since the examples for these two adaptation strategies are very limited, I will not formalise them as have been done in the previous discussion thus far. 4.3.1.3. Adaptation of source [ɕɥ] It is not only the Akan source non-native labial-palatal affricatives that are adapted in Dangme, but the speakers adapt also the source voiceless labial-palatal fricative [ɕɥ]. Just as was indicated in the preceding subsection, very few instances of this adaptation were observed in the data which I discuss in the following subsection. University of Ghana http://ugspace.ug.edu.gh 148 (53). Akan Dangme Gloss i. àɕɥɪɕ̀ɥɛ ́ àhìhʷíɛ,́ àhʷùhʷúɛ,́ mirror ii. àɕɥɪŕɪẃ́ àhĺéú sugarcane From (53), it is observable that the source non-native labial-palatal fricative [ɕɥ] is adapted rather as the glottal fricative [h] instead of the phonetically closer labialised pre-palatal fricative [ʃʷ], as was observed with the Ga. The explanation for this adaptation strategy is that unlike Ga, Dangme does not have the pre-palatal fricative in its inventory. Therefore, Dangme speakers seem to go for acoustically closest consonant available, which in this case is [h] as its replacement. The accompanying labialisation rule seems to apply selectively before both rounded and unrounded vowels as seen in (53i), and it fails before unrounded vowel and when preceding [l] as seen in (53i) and (53ii). It would be recalled that in the preceding section, it was explained that for the source labial-palatal fricative to be adapted as labialised glottal fricative, then one needs to investigate some diachronic facts about Akan, which was postulated that the source labial-palatal is assumed to have developed from an underlying glottal fricative /h/. Could it be the case that since this sound is not present in the Dangme inventory, in the adaptation therefore, there was a phonetic minimality mapping to [h], which could be closer? Or is it the case that this source sound was adapted at the time when it was still /h/ in the source language just as was explained for the adaptation of the source [ɥ] in Ga? University of Ghana http://ugspace.ug.edu.gh 149 A parallelism for the second question could be drawn from Peperkamp & Dupoux’s (2003) assumptions based on empirical evidence from Japanese that, “… older loans might have entered the language during a stage in which certain sounds in either the source or the borrowing language had different phonetic characteristics.” (Peperkamp & Dupoux 2003: 369). Just as was explained for the limited examples of the adaptation process, formalisation of this adaptation strategy will not be done for the obvious reason of the paucity of the examples. 4.3.1.4. Adaptation of source non-native [ɥ] as [w] (54). Akan Dangme Gloss i. àdàɥɪr̀ɪk̀ɔs̀ɔ ̀ ādàwrì̀kɔs̀ɔ ̀ iron bell ii. àɥɪǹ̀̃ dàdɪ ̀ àwèǹdàdè a plant iii. àɥɪr̀ɛh̀ʊ́ àwɛr̀ɛh̀ó, àwīlɛh̀ó, grief, sorrow, sadness iv. àkàtàɥìá àkàtàwìá umbrella From the examples above, it is observable that the Akan source non-native labial- palatal fricative is realised with the labial-velar glide [w] in many of the examples in Dangme just as was the case with Ga with very few exceptions as in the adaptation of the source word æ̀ɥísiá̃ á̃ ́̃ as àhʷũsa ́̃ ‘orphan, orphanage’, àɥɪr̀ɛh̀ʊ ́ as àjʷùrɛh̀ó ‘sorrow’ in which [ɥ] is realised as [hʷ] and [jʷ] respectively or even [ʤʷ] as in àʤʷùrɛh̀ó. With a similar diachronic explanation to the adaptation of University of Ghana http://ugspace.ug.edu.gh 150 the source [ɕɥ], the Akan source [ɥ] is an allophonic variant of /w/ whenever it occurs before front vowels in Akan. Therefore, it is phonologically predictable when the non-native source is usually adapted as [w] in Dangme. I formalise this adaptation in Dangme within the OT as below. In formalising within OT this adaptation strategy, which was termed de- labial-palatalisation in the previous section, I employ the set of constraint and the subsequent constraint ranking which were used for the analysis in 4.3.1.1 as shown below. Tableau 20: The adaptation of the source non-native [ɥ] as [w] in Dangme / àkàtàɥìá/ IDENT–IO *[+cor, - MAX-IO(F) (lab) ant,+lab] ‘umbrella’ b. [àkàtàɥìá] *! b. [àkàtàwìá] * c. [àkàtàjìá] *! * Constraint ranking: IDENT-IO(lab), *[+cor, -ant,+lab]» MAX-IO(F) In Tableau 19, the input is realised in three ways as reflected in the three candidates. Candidate (c) repairs the input illicit labial-palatal fricative through adapting it as a palatal glide/ approximant thus, fatally violating the highly-ranked constraint that demands correspondents between inputs and outputs in terms of the feature [lab]. For deleting the features of the input, the same candidate is also penalised by the low-ranking MAX-IO(F). In an attempt to satisfy these two constraints, candidate (a) faithfully preserves the input segment, which results in University of Ghana http://ugspace.ug.edu.gh 151 the violation of the crucially-ranked well-formedness constraint against [ɥ], *[+cor,-ant,+lab]. Though this is the only constraint candidate (a) violates, it is outcompeted by its co-competitor candidate (b), which also incurs a single violation mark for MAX-IO(F) for dropping the [cor] feature in the input, due to the relatively lower ranking of MAX-IO(F). 4.3.1.5. Adaptations of source non-native [Cᶣ] as [Cʷ] So far the discussions have focused on the adaptations of the Akan source labial- palatals in Dangme. In this subsection, we shift our focus to the discussion of how Akan source labial-palatalised consonants are adapted in Dangme. (55). Akan Dangme Gloss i. æ̀tᶣià̃ ́̃ àtʷùã ́̃ rebellion, defiance ii. bɛr̀ɛd̀ᶣìá bɛl̀ɛd̀ʷùá a tall straggling shrub iii. àbèdᶣìá bɛʤ̀ʷùá a palm tree, when slightly taller than tamlaa and still untrimmed From the data in (55), we observe that the source labial-palatalised alveolar stops [tᶣ] and [dᶣ] are adapted as the labialised alveolar [tʷ] and [dʷ] respectively. This can be straightforwardly accounted for as change in a feature value of the feature matrix that defines the superimposed [ɥ] while the primary consonant is University of Ghana http://ugspace.ug.edu.gh 152 maintained, as has been explained up to this point. However, as it can be seen, in (55iii) the source primary alveolar stop is adapted as a pre-palatal affricate, this time preserving the palatality property of the source at the expense of modifying the manner and place of articulation. Additionally, the labiality of the source is also preserved through the superimposition of the labialised diacritic on the target affricate. Just as was done before, examples of these adaptation strategies are few in the data collected for the present study. Therefore, making a generalisation in the form of formalisation based on these limited cases will not be appropriate. 4.3.2. Adaptations of source word-final consonants in Dangme So far we have discussed the adaptations of the Akan source non-native labial- palatal and the labial-palatalised consonants in Dangme. In the final discussions of this chapter, we look at the adaptations of the Akan source word-final [w] in Dangme, discussing the strategies, which Dangme speakers employ in the adaptation of words of Akan source that end in the labial-velar glide. In the discussions to follow, we will observe that the adaptations follow similar patterns as were observed for Ga in the preceding chapter. The three main repair strategies include vocalisation, elision, and compensatory lengthening. 4.3.2.1. Adaptations of source word-final [w] as [o] This sub-subsection looks at the process of vocalisation that is employed as a strategy for repairing an illicit Akan source word-final [w] in Dangme. University of Ghana http://ugspace.ug.edu.gh 153 (56). Akan Dangme Gloss i. bɪr̀ɛẁ̀ blɛò̀ slowly, carefully, gently ii. æǹ ih́̃ áẃ ànih́̃ áó laziness iii. ɡúáẃ ɡʷáò whip, beat iv. fɛẃ ́ fɛó́ beauty From the examples in (56), it is clear that the Akan source word-final [w] is vocalised into [o] in Dangme with the feature [+round] been preserved in the adaptation process. This adaptation has been necessitated by the fact that Dangme bans the labial-velar glide at word-final position. Dangme speakers’ choice of the mid vowel [o], instead of the acoustically closer vocalic segment high vowel [u], is what requires explanation, however. Does the word-final [w] perceptually sound like [ʊ] in the ears of the speakers, hence informing its adaptation as [o], which was observed to be one of the target vowels when adapting the source non-native [ʊ] on perceptual grounds? This adaptation strategy falls in line with what was observed about Ga in section. I formalise this adaptation of the Akan source word-final [w] through vocalisation in Dangme, and to do this, I employ the set of constraints and the constraint ranking that were used for a similar analysis in Ga in 4.2.1.6 (Tableau 15). University of Ghana http://ugspace.ug.edu.gh 154 Tableau 21: The adaptation of the source word-final [w] as [o] in Dangme / fɛẃ́ / IDENT-IO IDENT-IO MAX-IO ‘beauty’ (lab) *w]σ (Cons) a. [fɛẃ́] *! b. [fɛó́] * c. [fɛ]́ *! Constraint ranking: IDENT-IO(lab), *w]…σ » MAX-IO » IDENT-IO (Cons) From Tableau 20, we observe three different strategies, which each of the three candidates employs in repairing the illicit input word-final [w]. Candidate (a) is the output form that has perfect segmental match to the input through preserving faithfully all the input segments. By being faithful to the input, it fatally violates the highly-ranked position-sensitive constraint *w]σ that ban [w] from occurring at word-final position. Candidate (c), on the other hand, satisfies *w]σ by completely deleting the prohibited [w] at the final position. However, by so doing it incurs fatal violation mark for the crucially-ranked anti-deletion constraint, MAX-IO. For adapting a vocalic segment as a non-vocalic segment, candidate (b) violates IDENT-IO(cons), which demands both the input and the output segments correspond in terms of the feature [cons]. Candidate (c) still emerges the optimal candidate despite this violation due to the low ranking of IDENT-IO(cons). There is another strategy by which Dangme speakers repair the Akan source word-final [w]. This strategy is also through vocalisation. However, unlike in the immediately preceding discussion, the illicit source word-final [w] is vocalised into [u]. The following examples illustrate this adaptation strategy. University of Ghana http://ugspace.ug.edu.gh 155 (57). Akan Dangme Gloss i. àɕɥɪŕɪẃ́ àhĺéú sugarcane ii. àsɛẃ́ àsɛú́ misfortune iii. fɛẃ́ fɛú́ beauty From the examples above, it can be observed that the illicit word-final [w] is adapted as [u]. This adaptation seems to answer the question raised in the previous discussion as to why the target vowel is [o] instead of the expected [u], if the adaptation takes into an account the closest vocalic segment in the vowel space. Just as was explained for such similar strategy in Ga, Dangme speakers seem to adapt the positionally ill-formed [w] as the closest vocalic segment in Dangme, which is [u]. However, from (57iii), it is evident that Dangme speakers make a choice between [o] and [u] as the replacing vowel for the illicit source word-final [w]. In (56iv), the same source word fɛẃ́, which final [w] was adapted as [o] as in fɛó́ is realised as fɛú́ in which [w] changes to [u] in (60iii). It is therefore explained that in the adaptation of the Akan illicit source word-final [w] through vocalisation, Dangme speakers have two options, namely adapting as [o] as in (56), or altering [w] into [u] as seen in (57). It is not yet clear to me whether this variation is dialectal or not. In formalising this form of vocalisation, it seems to suffice to employ the Tableau 20 since it could be predicted that any of the examples in (57) as an input is likely to surface with either [u] or [o] in word-final position. University of Ghana http://ugspace.ug.edu.gh 156 Another repair strategy for the illicit source word-final [w] aside from the vocalisation is deletion. Just as in Ga, Dangme speakers can also employ deletion when adapting word-final [w]. This adaptation strategy cannot be discussed in detail since only one instance of such adaptation has been identified in the data for the present study. In this example, the Akan source word títíríẃ ‘mainly, chiefly, especially’ is adapted as títlī, títrī in Dangme. It can be seen that the source word- final [w] is completely dropped in the adapted forms in Dangme. 4.4. Summary of the chapter This chapter has discussed the adaptations of consonants in Akan source words in both Ga and Dangme. In the discussions of the consonantal adaptations, the main issued were focused on the adaptations of Akan source non-native consonants and the adaptations of Akan source illicit word-final labial-velar glide [w]. For the discussions of the non-native consonants, we looked specifically at the adaptations of the labial-palatals and the labial-palatalised consonants, all of which are articulated with compressed labialisation. With regard to the adaptations of the non-native consonants, it was concluded that speakers of both target languages usually adapt these source consonants to the perceptually closest ones that are present in their language. The non-native labial-palatal feature is achieved by the speakers through imposing the labialisation feature on the acoustically closest consonant. And where this acoustic closeness strategy cannot work, as was the case of the adaptation of the Akan source palatal fricative, speakers seem to incidentally replace such consonants with University of Ghana http://ugspace.ug.edu.gh 157 their underlying counterparts in Akan. This led us to pose the question as to whether the adaptation of such instances might have taken place prior to the source consonants undergoing the palatalisation process or it is due to other phonetic factors. The phonological adaptation pattern has been accounted for in OT by positing a higher-ranked IDENT-IO(lab) and also markedness constraint *[+cor,- cont,+lab] over MAX-IO(F). This ranking ensures that in avoiding violating the prohibited source non-native consonant, Ga and Dangme speakers preserve the labiality of the same source consonant where equivalent labialised sounds are available in the consonantal inventories of the two target languages. The other consonantal adaptations that engaged our attention was that of the adaptations of Akan source word-final [w], in which it came out that two main strategies apply, namely vocalisation and deletion. In the former strategy, the illicit word-final [w] vocalises into the rounded [o] in both Ga and Dangme, and additionally into [u] in Dangme alone. For the latter strategy, we found out that the application of the deletion strategy could lead to compensatory lengthening in Ga, for instance, usually when the rest of the source word is minimally smaller than disyllable after the final [w] is deleted. This compensatory lengthening was observed to apply exclusively in Ga. I tabulate the major points of the discussions in the following showing the similarities and the differences between the adaptation patterns of Akan source consonants by both Ga and Dangme speakers. University of Ghana http://ugspace.ug.edu.gh 158 Table 12: Adaptations of some Akan source consonants in Ga and Dangme Akan source consonants Ga Dangme Non-native consonant: ʨɥ ʧʷ ʧ/tʷ ʥɥ ʤʷ ʤʷ ɕɥ ʃʷ hʷ (has no ʃ) Cᶣ Cʷ Cʷ ɥ w w ɕ ʃ, h h Word-final w: i. o i. o, u ii. deletion ii. N/A iii. compensatory iii. N/A lengthening after the deletion University of Ghana http://ugspace.ug.edu.gh 159 CHAPTER FIVE AKAN PROSODIC ADAPTATIONS IN GA AND DANGME 5.1. Introduction In this chapter, I discuss the phonological patterns of the adaptations of some Akan source prosodic features by both Ga and Dangme speakers. The prosodic properties to focus our discussions on include vocalic nasality and the syllable. The discussions focus on how the [+nasal] feature of some Akan source vowels, particularly how the non-native vowels are realised in both Ga and Dangme in the adaptation process. Regarding the syllable, I discuss the adaptations of the Akan source syllabic nasal, in the two target languages. I look at the main strategies the speakers of both target languages employ in adapting this syllabic nasal and attempt to answer the questions of the how and the where this adaptation occurs in the adapted forms of Akan source words in the adaptation process. In the discussions of the syllable structure, I briefly look at the adaptations of CV monosyllables as well as CVN and VV disyllables in both Ga and Dangme. Finally, the outcomes of the discussions are also be formalised within the OT framework as have been done in the two previous chapters. The rest of the chapter is organised as follows: section 5.2 considers the strategies that Ga speakers employ in adapting the nasal feature on some of the Akan source words in the adaptation process. The section specifically looks at the realisations of the nasal feature on some non-native vowels in Ga. It also discusses University of Ghana http://ugspace.ug.edu.gh 160 how some Akan source nasalised vowels are denasalised in Ga as well as sometimes how some source non-nasalised vowels are adapted as nasalised vowels. Section 5.3 also discusses the adaptations of nasality on some Akan source non-native vowels in Dangme. It specifically considers how Akan source nasalised and non-nasalised vocalic segments are realised in Dangme during the adaptation process. The section attempts to draw parallels between the strategies both languages adopt in adapting this prosodic feature in Akan source loaned words. In section 5.4, we take a look at how the Akan source syllabic nasal, at different positions, is realised in Ga. These positions in the word, as discussed below, will include word-initial, word-medial, and word-final positions. The same subject matter of the adaptations of source syllabic nasal, but this time in Dangme is addressed in 5.5. The chapter is summarised in section 5.6. 5.2. Adaptations of Akan source vocalic nasality in Ga This section discusses the strategies Ga speakers employ in adapting the nasal feature on some of the Akan source vowels, both native and non-native vowels. We first consider the adaptations of the nasal feature on the Akan source non- native vowels in Ga. In chapter three, it was observed that the upper peripheral non-native vowels of Akan source, namely [ɪ] and [ʊ] are adapted as [e, i, ɛ] and [o, u, ɔ] respectively in Ga. The two source non-native vowels are adapted to six vowels in Ga, and out of the six, four are mid vowels. Two of the four mid vowels, namely [o] and [e] are not nasalised in Ga. Therefore, except when replacing a non-native vowel as a vowel from the periphery stock, the prosodic property of University of Ghana http://ugspace.ug.edu.gh 161 nasality of the target vowel is dropped in the adapted forms. In the following subsections, I present the adaptations of the nasal feature of each of these Akan source non-native vowels in Ga. The discussions are also extended to the cases where the Akan source non-native is nasalised but the target vowel is denasalised not because the target vowel cannot be nasalised. We also discuss instances where the source vowels are oral, but they are nasalised in the target Ga language. Some of the discussions are formalised within the OT framework. 5.2.1. Adaptations of Akan source nasalised [ɪ ̃] With our knowledge of the target vowels with which the non-native [ɪ] is adapted, our focus here is to answer the question of whether or not the environment can play any significant role in deciding the preservation or otherwise of a source nasal feature on the target vowel. (58). Akan Ga Gloss i. àhɪŋ̀̃ k̀ʷàá àhiŋ̀̃ k̀ʷrà́ King’s servant, attendant ii. ɕɥɪá̃ ɕ́̃ ɥɪá̃ á̃ ́̃ hʷiá̃ h́̃ ʷiá̃ ́̃ thin, slender, slim iii. m̀pɪm̀̀̃ ma ́̃ am̀̀̃ pèma ́̃ a type of plantain iv. àpɪḿ́̃ àk͡pé thousand, one thousand v. àmaǹ̃ ɪ ́̃ àmaǹ̃ é trouble, adversity vi. ʨɥɪm̀̀̃ ʧʷɛm̀̀ all together University of Ghana http://ugspace.ug.edu.gh 162 vii. àʥɥɪŋ́̃ ́ àʤʷɛŋ́ ́ a bird, any wagtail, mud-fish viii. ɔʧ̀ɪ̀̃aḿ̃ ɪ ́̃ òʧàámɛ ́ linguist, spokesperson for a chief From the examples in (58i) – (58ii), it is observable that whenever the source non- native nasalised [ɪ] is adapted as [i], the target vowels is systematically nasalised into [i ̃] even without a nasal consonant in the neighbourhood in (58ii). However, whenever the same source vowel is adapted as [e] as seen in (58iii) – (58v) or [ɛ] as in (58vi) – (58viii), the nasal feature is invariably dropped on the target vowels despite the presence of a nasal consonant in the context. This pattern of adaptation can be explained with recourse to Ga phonology. That is, in Ga, /e/ is one of the front vowels that cannot be nasalised (cf. Dakubu 2002, Campbell 2017). However, what is not clear is why oral [ɛ], which can also be nasalised in native Ga phonology, is adapted without the source nasal feature. Therefore, it seems Ga speakers apply a rule on vocalic nasalisation when adapting Akan source vowels into their vocabulary. The rule can be stated as: in adapting the Akan source [ɪ ̃] as a non-high front vowel, the [+nas] feature is systematically dropped in the target segment, though the nasalisation context is provided. There are few exceptions to this general rule such as àtɛŋ́̃ ́ ‘judgment’, which is adapted as the source nasal feature intact as àtɛŋ́̃ ́ and not the expected *àtɛŋ́.́ It is worth pointing out that [ɛ] can be nasalised in Ga in native words such University of Ghana http://ugspace.ug.edu.gh 163 as aʦrɛlɛ [àʧr̀ɛl̀̃ ɛ]́̃ ‘money-doubling’, just as it is nasalised in Akan. However, Ga speakers may adapt it without the nasal feature though the nasalising environment may be provided, as seen in (58). From the data, it has been observed that in the adaptation process, when [ɛ]̃ is nasalised in source words, the nasal feature is most likely to be maintained in the adapted form in Ga. However, whenever a source [ɪ ̃] is adapted as [ɛ], the nasal feature is usually dropped in the adapted form in Ga. I now attempt to formalise the dropping of the nasal feature on the Akan source vowels by Ga speakers in the adaptation process. In formalising the adaptations of the nasalised front vowel [ɪ], I introduce a constraint that penalises a candidate with a non-high nasalised vowel. This constraint has to rank above a constraint, which is against deletion; MAX-IO. I define this well-formedness constraint as (59): (59). *[-high,+ATR,+nas]: A vocalic segment specified for [-high, +ATR] must not be nasal The constraint is formulated after Kager’s (1999: 28) context-free *VNASAL, which completely prohibits all vowels becoming nasal in a language. However, as has been observed about Ga phonology so far, nasalising vocalic segments is allowed in both the native and the loanword phonologies. There are few vowels in Ga that cannot be nasalised. These vowels, as have been shown, include [e]. The advanced University of Ghana http://ugspace.ug.edu.gh 164 mid vowels are systematically not nasalised in all words borrowed from Akan, which is in consonance with the nasalisation rules in the native Ga phonology. One faithfulness constraint that is in conflict with this *[-high,+ATR, +Nas] is IDENT-IO(nas). This constraint; IDENT-IO(nas), which follows Kager (1999: 29) is defined as (60): (60). IDENT-IO(nas): Correspondent segments in input and output have identical value for [nasal] The IDENT-IO(nas), constraint when dominated, will license the optimal candidate to drop the input nasal feature when adapting the source nasalised non- native vocalism. In other words, its violation will not have maximum effect on the offending candidate due to its significantly low ranking. In addition to these two constraints, I employ two constraints that were used in the previous chapters, namely *[+high,-ATR], which bans unadvanced high vowels for candidate that faithfully preserves the non-native segment, and MAX-IO, a constraint against deletion, for the present analysis. The discussion of the data in (58) is formalised as follows in Tableau 21. University of Ghana http://ugspace.ug.edu.gh 165 Tableau 22: The adaptation of the source nasality in [ɪ]̃ in Ga /àmaǹ̃ ɪ/́̃ *[+high, - IDENT-IO *[-high,+ATR, MAX-IO ‘trouble, adversity’ ATR] (nas) +nas] a. [àmaǹ̃ ɪ]́̃ *! b. [àmaǹ̃ é] * * c. [àmaǹ̃ e]́̃ *! d. [àmaǹ̃ ] **! Constraint ranking: *[+high, -ATR], *[-high,+ATR,+nas] » MAX-IO » IDENT-IO (nas) From Tableau 21, it can be observed that candidate (a) becomes suboptimal due to its fatal violation of the high-ranking *[+high, -ATR] for preserving the prohibited input vocalism. This constraint makes no reference to the nasal feature, otherwise the candidate would have satisfied it had it adapted the input segment as the closest high vowel, [i]. Candidate (c) also suffers the same fate for fatally violating the constraint against nasalising advanced mid vowels in Ga. The race for optimality is then decided by the crucially-ranked MAX-IO. Candidate (d), which employs deletion of the prohibited input segment as a repair strategy, incurs fatal violation marks for violating MAX-IO twice, namely for deleting the input segment, and for deleting the input nasality. Candidate (b), on the other hand, is penalised for deleting the input nasality, thus incurring only one violation mark. Though candidate (b) additionally violates IDENT-IO(nas), it emerges the optimal candidate due to its better violations portfolio of the crucial ranking of MAX-IO than candidate (d). University of Ghana http://ugspace.ug.edu.gh 166 5.2.2. Adaptations of Akan source nasalised [ʊ]̃ In the following discussions, we consider how the Akan source nasalised non- native back vowel is realised in Ga. In a sequel to the discussion in subsection 5.2.1, [ʊ̃] can be adapted as either a back vowel that can be nasalised or one that cannot be nasalised in Ga. I provide examples of how these adaptations can be done in Ga as (61). (61). Akan Ga Gloss i. àhʊ̀̃m̀ka ́̃ àhũŋ̀ k̀á joy, satisfaction ii. àkʊ̃̀ǹtá àkũ̀ǹtá brother-in-law iii. mań́̃ ꜝsʊ ̃́ mań́̃ sɔ ́̃ quarrel, strife, ill-feeling; litigation iv. ànʊ̃̀kʷá ànɔk̀̃ ʷá truth; indeed, truly, really, honestly v. ɔb̀ʊ̃̀ǹsaḿ́̃ àbòǹsaŋ́̃ ́ devil vi. dʊ̃́ḿ dómɔ ̀̃ trust, accept, believe in vii. aŋ́̃ ḱʊ́̃naḿ́̃ aŋ́̃ ḱònaḿ́̃ loner, unsociable person From the examples in (61), and following the discussion in chapter three, we can predict that whenever Akan source nasalised non-native [ʊ̃] is systematically adapted as the back high vowel [u], the source nasal feature is transferred onto the University of Ghana http://ugspace.ug.edu.gh 167 target vowel as can be seen in (61i) – (61ii). Similarly, unlike in the preceding discussions, whenever the same source vowel is adapted as the unadvanced back mid vowel [ɔ], the source nasality is transferred onto the target vowel to be realised as [ɔ̃], as in (61iii) – (61iv). However, whenever the source is adapted as the advanced back mid vowel [o], the nasal feature is completely dropped as in (61v) – (61vii). This adaptation strategy can be explained by the fact that as far as Ga phonology is concerned, aside from the high and the low vowels, it is only the unadvanced mid vowels [ɛ] and [ɔ] that can be nasalised in Ga. Therefore, Ga speakers will predictably drop the nasal feature whenever adapting the source vocalism with [o]. In Tableau 22, I present a formal account of this adaptation strategy. To formalise the adaptation of the nasality of the input nasalised non-native [ʊ]̃ , I employ the set of constraints and the ranking used for the analysis in subsection 5.2.1 for the present analysis. Tableau 23: The adaptation of the source nasality in [ʊ̃] in Ga /ɔb̀ʊ̃̀ǹsaḿ́̃ / *[+high, - IDENT-IO *[-high,+ATR, MAX-IO ‘devil’ ATR] (nas) +nas] a. [àbʊ̃̀ǹsaḿ́̃ ] *! b. [àbòǹsáŋ]́ * * c. [àbõ̀ǹsáŋ]́ *! Constraint ranking: *[+high, -ATR], *[-high,+ATR,+nas] » MAX-IO » IDENT-IO (nas) University of Ghana http://ugspace.ug.edu.gh 168 In Tableau 22, candidate (a) faithfully adapts the non-native nasalised vowel. This adaptation strategy fatally violates the highly-ranked *[+high, -ATR], the constraint that bans [ʊ] in the output in Ga. Candidate (c), on the other hand, avoids violating *[+high, -ATR] by adapting the source vowel as [o]. However, it also fatally violates the highly-ranked *[-high,+ATR,+Nas], a constraint that is ranked highly in the general phonology of Ga for nasalising the advanced back mid vowel, as in [õ]. Though each of these two candidates violates only one constraint, due to the high ranking of the violated constraints, candidate (b) emerges the optimal candidate for Tableau 22 for violating relatively lower ranked constraints, albeit one more than its competitors. 5.2.3. Adaptation of Akan source non-nasalised vowels as nasalised vowels In this subsection, the discussion focuses on two contexts of the adaptation, namely where Akan source oral vowels are nasalised in Ga, and where Akan source nasalised vowels are denasalised in Ga. Unlike in the preceding discussions, the source vocalic segments are not altered in the following discussions since they are all present in Ga. (62). Akan Ga Gloss i. àkǽ̃ńsí àkaŋ́̃ ʃ́iì̃ ̀̃ competition ii. kùmaá̃ ś̃ ɪ ́ kũm̀ aá̃ ś̃ é the city of Kumasi University of Ghana http://ugspace.ug.edu.gh 169 iii. àsɛm̀̀sùróẃ àsɛm̀̀̃ srò̀ fear of trouble iv. m̀mɔd́ɪŋ́̃ ́ mɔd́̃ ɛŋ́ ́ effort, esp. successful or laudable effort v. pàtá kp͡ at́̃ a ̀̃ make peace, settle a dispute In (62), it can be observed that the underlined Akan source oral vowels are realised as nasalised vowels in Ga. That is, in feature terms, a source [-nas] is realised with [+nas] in the target language. The nasalisation in the target language occurs in the neighbourhood of a [+nas] consonants in all the examples above with the exception of (62v). The questions then are, (1) does this nasalisation occur out of overapplication of the nasalisation rule? If yes, then (2) how do we explain such cases as (62v), where there is no nasal feature in the neighbourhood to serve as the bases of the (over)application of the nasalisation? (3) Could the application of the nasalisation rule also result from perception? By the end of this chapter, I will attempt to provide some answers to some of these questions. Comprehensive explanations to these questions, however, might require further investigation of the observed pattern. 5.2.4. Adaptation of Akan source nasalised vowels as non-nasalised vowels There are instances where the nasal feature of some Akan source segments is dropped in Ga in the adaptation process though the target vowels can be nasalised in Ga. I illustrate this adaptation strategy in the following examples in (63). University of Ghana http://ugspace.ug.edu.gh 170 (63). Akan Ga Gloss i. dàdɪs̀ɛŋ́̃ ́ dàdèsɛŋ́ ́ iron pot ii. ]̀ʨɪà̃ ḿ̃ ɪ ́̃ òʧàámɛ ́ linguist, spokesman for a chief iii. æ̀ʥìnam̀̃ ʊ̀ã ́̃ àʤrà̀mòá, an appellation for the domestic cat iv. æ̀diŋ̀̃ k̀rà́ àdìŋk̀rà́ cloth on which traditional symbols have been printed It is observable from the data in (63) that the Akan source nasality can be denasalised in Ga during the adaptation process. All the underlined Akan source vowels would have been expected to be nasalised in their nasalising environments, yet they fail to nasalise in the adaptation process in Ga. 5.3. Adaptations of Akan source nasality in Dangme In this section, I discuss how the Akan source nasality is realised in Dangme. The discussions focus on the realisations of nasality on source non-native vowels. The section also discusses the nasalisation of source non-nasalised vowels as well as the denasalisation of source nasalised vowels in the adaptation process in Ga. The strategies Dangme speakers employ when adapting the Akan source nasal feature on both native and non-native vowels will engage our attention in this section. Just as was done in section 5.2, the discussions first focus on how the speakers adapt the nasal feature on Akan source non-native vowels in Dangme. In chapter three, we realised that Akan source non-native vowels [ɪ] and [ʊ] are adapted in Dangme University of Ghana http://ugspace.ug.edu.gh 171 as [e, i, ɛ] and [o, u, ɔ] respectively. Some of these target vowels can be nasalised in the regular phonology of Dangme, while others cannot. The rest of this section discusses how the source nasality and non-nasality are realised in Dangme. But before we proceed with the discussions, we briefly discuss the nasalisation of vowels in Dangme. Aside from the peripheral vowels, namely [i, u, a], two out of the four mid-vowels can be nasalised in Dangme just as obtains in the sister language, Ga. These two mid-vowels are [ɛ]̃ and [ɔ̃]. We, therefore, begin the discussions with the assumptions that the Akan source nasalised non-native [ʊ̃] will be most likely adapted without the nasal feature as [o]. Similarly, we will expect the Akan source nasalised non-native [ɪ ̃] to be adapted without the nasal feature when adapted as [e]. Aside from these two vowels, the rest of the target vowels will preserve the nasal feature in the source. 5.3.1. Adaptations of Akan source nasalised [ɪ]̃ In the following examples, I show that Dangme speakers can faithfully maintain the source nasality on even source non-native vowels when adapting words from Akan. (64). Akan Dangme Gloss i. dànɪ ́̃ daí̃ ̀̃ change, turn into, metamorphose ii. m̀mɔd̀ɪń́̃ mɔd́í effort, esp. successful effort University of Ghana http://ugspace.ug.edu.gh 172 iii. àmaǹ̃ ɪɛ̀̃ ́̃ àmaǹ̃ íɛ ́ story, case, event, matter iv. ʥɥɪń́̃ ʤùɛ̃ ̀̃ reason, think v. ɔʨ̀ɪà̃ ̃́ꜝ mɪ ́̃ òʧìā!mɛ ́ spokesperson, linguist vi. àpɪḿ́̃ àkp͡ é thousand vii. bɪḿ́̃ béḿ innocence From (64), it is can be observed that just as was seen about Ga in section 5.2, when a nasalised source [ɪ ̃] is adapted as [i], the nasal feature tends to be maintained on the target vowel, as seen in (64i). However, this nasalisation rule fails to apply in (64ii) – (64iii), though the nasalising environment is there. In (64ii), for instance, we would have expected the source nasality to pass onto the target vowel prior to the deletion of the word-final nasal consonant as the trigger consonant does in (64i). Another case is (64iii) where, though the trigger nasal consonant is preserved in the target language, yet the source vocalism is denasalised. A comparison can be drawn for the adaptations of source [ɪ ̃] as either [i ̃] or [i] between (64i) and (64iii), and as [ɛ]̃ or [ɛ] between (64iv) and (64v) respectively. In (64vi) and (64vii), the lack of the nasality on the target vowel can be explained from the point of view of phonological markedness in Dangme, which does not permit [e] to be nasalised. To provide a formal account of how Dangme speakers adapt the nasal feature on Akan source vocalic segment, I employ the constraints that were used University of Ghana http://ugspace.ug.edu.gh 173 for the analysis in subsection 5.2.2 since it exhibits similar pattern to the Dangme adaptation strategy. Tableau 24: The adaptation of the source [ɪ]̃ in Dangme / bɪḿ́̃ / *[+high, - IDENT-IO *[-high,+ATR, MAX-IO ‘innocence’ ATR] (nas) +nas] a. [bɪḿ́̃ ] *! b. [béḿ] * * c. [beḿ́̃ ] *! Constraint ranking: *[+high, -ATR], *[-high,+ATR,+nas] » MAX-IO » IDENT-IO (nas) In Tableau 23, the illicit input [ɪ]̃ is realised in three different ways in the three candidates. In candidate (a) the illicit input has a perfect correspondent. The input segment is preserved with its nasal feature in fatal violation of the highly-ranked constraint that bans the input vowel, *[+high, -ATR]. Candidate (c), on the other hand, avoids violating this high-ranking constraint by replacing the illicit vowel with a permissible one [e]. However, by preserving the source nasality on [e], it also incurs fatal violation mark for the highly-ranked *[-high,+ATR,+nas] for nasalising an advanced mid vowel. Candidate (b), just as its main competitor, candidate (b), also adapts the input vowel as [e]. It, however, adapts the input illicit vocalism differently as it drops the input nasality while candidate (c) preserves it. By employing the deletion of the input nasality as a repair strategy, candidate (b) violates both MAX-IO and IDENT-IO(nas). Though it is the only University of Ghana http://ugspace.ug.edu.gh 174 candidate that violates the higher number of constraints, it still emerges the optimal candidate due to the low ranking of the offending constraints. 5.3.2. Adaptations of Akan source nasalised [ʊ̃] In this subsection, I discuss how the Akan source [ʊ̃] is realised in Dangme in the adaptation process. Consider the examples in (65) for illustration. (65). Akan Dangme Gloss i. ɔb̀ʊ̃̀ǹsaḿ́̃ àbùsaḿ́̃ ,àbùsiá̃ ḿ́̃ devil ii. æ̀ʥìnàmʊ̃̀á àʤlà̀mùá domestic cat iii. ànʊ̃̀kʷárɪ ́ ànɔk̀̃ úalé true, correct; truly, indeed iv. ɔb̀ʊ̀̃ǹsaḿ́̃ abòsiá̃ ḿ́̃ devil v. ɔp̀àpʊ́̃ àk͡pàk͡pó, k͡pàk͡pó male goat, billy goat, sex maniac vi. ànʊ̃̀kʷárɪ ́ ànòkwádé fact, truth From the examples in (65), it is observable that the illicit source vocalism [ʊ̃] is adapted differently. In (65i) – (65ii), the source is adapted as [u], however, against our expectation, the source nasality is dropped from [u]. That is, an expected [ũ] adaptation is realised rather as [u]. In (65i) the trigger nasal consonant is dropped, while in (65ii) it is present. It is worth re-emphasising here that Dangme has the University of Ghana http://ugspace.ug.edu.gh 175 nasalised [u]̃ in its vocalic system. In (65iii), however, as would be expected, the source nasality is preserved when [ʊ] is adapted as [ɔ]. Again, following the phonology of Dangme, the Akan source nasality is dropped whenever [ʊ]̃ is adapted as [o] as can be seen in (65v) – (65vi). Just as was done for Tableau 23, I formalise this pattern of adaptation in the following subsection. Since the adaptation pattern to be analysed is similar to the one in subsection 5.2.4, I employ the set of constraints and the subsequent ranking in the present analysis. Tableau 25: The adaptation of the source nasality in [ɪ]̃ in Dangme /ànʊ̃k̀ ʷárɪ/́ *[+high, - IDENT-IO *[-high,+ATR, MAX-IO ‘fact, truth’ ATR] (nas) +nas] a. [ànʊ̃̀kʷádé] *! b. [ànòkwádé] * * c. [ànõ̀kwádé] *! Constraint ranking: *[+high, -ATR], *[-high,+ATR,+nas] » MAX-IO » IDENT-IO (nas) From Tableau 24, candidate (a) fatally violates the high-ranking constraint against the unadvanced high vowel [ʊ], for preserving the input segment in the output. The two other competing candidates avoid this violation by adapting the banned input vowel as [o]. However, the input nasality is realised differently by the two candidates. While candidate (c) preserves the input nasal feature in fatal violation University of Ghana http://ugspace.ug.edu.gh 176 of another highly-ranked constraint against such nasality, candidate (b) drops the source nasality, thus violating both MAX-IO and IDENT-IO(nas). Though candidate (b) is the only output form that violates two constraints vis-à-vis one each by the other two competitors, it emerges the optimal candidate due to the relatively low ranking of the two constraints it violates. 5.3.3. Adaptation of Akan source non-nasalised vowels as nasalised vowels Aside from the adaptation patterns so far observed in Dangme in this section, there are instances where the Akan source vocalic segments that bear no nasal feature are adapted as a [+nasal] feature on them in Dangme. In the following examples I present a few of these cases. (66). Akan Dangme Gloss i. æ̀ʥɥàmań́̃ àʤùam̀̃ a ́̃ prostitute; fornication ii. æ̀ʥɥúma ́̃ àʤṹma ́̃ work, labour iii. kʷæ̀dú àkɔd̀ṹ, kɔd̀ṹ banana iv. ètú tṹ gun, shooting v. æ̀kùtú àkut̀̃ ú, àkũ̀tṹ sweet orange From the examples in (66), the Akan source [-nasal] vowels are realised as [+nasal] in Dangme in the adaptation process. From (66i) – (66ii), it can be tentatively explained that the nasal feature might have spread leftwards from the University of Ghana http://ugspace.ug.edu.gh 177 [+nasal] segments to the target vowels. There are exceptions to this explanation such as the adaptation of the Akan source words like æ̀tùpań́̃ ‘bottle’ which is adapted as àtũk͡pá with no [+nasal] feature in the adapted form. The explanation for this might be that the target vowel [u] got its nasality from an initially nasalised [ã] as the trigger, as in àtũkp͡ a,́̃ then the trigger vowel lost its nasality afterwards to realise the surface form àtũkp͡ á? However, the same explanation cannot be advanced for the examples in (66iii) – (66v), where we find no nasal segment in the neighbourhood of the source words to have triggered the nasalisation process in the adapted forms. It is hard to posit any phonological motivation for this nasal adaptation other than to assume that perhaps the speakers might have missed perceived the nasal properties of the target vowels in the adaptation process. This assumption would require perception test to confirm its validity or otherwise. 5.3.4. Adaptation of Akan source nasalised vowels as non-nasalised vowels There are cases, as in (67), where the Akan source [+nasal] vocalic segments do not have correspondent [+nasal] vowels in the target words in Dangme, though those vowels can be nasalised in Dangme. I illustrate this below. (67). Akan Dangme Gloss i. dɔŋ̀̃ k̀ɔ ́ dɔŋ̀k̀ɔ,́ òdɔk̀ɔ ́ slave ii. kɔŋ̀̃ k̀ɔǹ̀̃ sa ́̃ kɔk̀ɔs̀á slander; damaging report iii. ɔb̀ʊ̃̀ǹsaḿ́̃ àbùsáḿ, àbùsiá̃ ḿ́̃ devil University of Ghana http://ugspace.ug.edu.gh 178 iv. m̀mɔd̀ɪń́̃ mɔd́í effort, esp. successful effort v. àmaǹ̃ ɪɛ̀̃ ́̃ àmaǹ̃ íɛ ́ story, case, event, matter vi. àkʊ́kɔʥ́ań́̃ àkɔḱɔʤ́áí a kind of domestic fowl From (67), it can be seen that the nasal feature in the source segments are usually dropped in the adapted forms in Dangme. Could the lack of the nasal feature on the target vowels be due to the misperception of the prosodic feature on the part of Dangme speakers? A closer look at the examples above points to the fact that with the exception of dɔŋ̀k̀ɔ,́ Dangme speakers drop the contiguous nasal consonant, as seen in (67i) – (67iv), to license the denasalisation process. The dropped nasal consonant deletes with its nasal feature so that it has no effect on the potential target vowels. There are, however, few counterexamples where though the trigger nasal consonant is preserved, yet its effect is dropped on potential target vowels as seen in (67v). The example in (67vi) is very similar to the phonetic variant of the Asante [àkʊ́kɔʤ́aɪ́̃ ̃́] with the main variation being that while Asante has the nasal feature on the target vowels; it is dropped in the Dangme form. So far we have discussed the adaptations of Akan source nasality in the adaptation processes in Ga and Dangme. The discussions have focused on the adaptations of nasality on Akan source non-native vowels as well as the nasalisation and denasalisation of Akan source oral and nasalised vowels respectively in both Ga and Dangme. In the next sections, we look at how the University of Ghana http://ugspace.ug.edu.gh 179 Akan source syllabic nasals are realised in both Ga and Dangme. The focus is on the adaptations of word-initial, word-medial and word-final syllabic nasals. 5.4. Adaptations of Akan source syllabic nasals in Ga Akan has syllabic nasals either as part of stems or as affixes. These syllabic nasals can occur at every position of the word, namely initial, medial, and final positions. In this section, we consider how these nasals are realised in Ga in the adaptation process. A fact of Ga phonology, according to Dakubu (2002: 33), is that a sequence of double nasals in word-initial position is not permitted in Ga, but it is permitted in word-final position. And even that, it is limited to only velar nasals with the bilabial nasal [m] occurring only in loaned words from Akan (Twi). Dakubu continues that this bilabial nasal is even sometimes realised phonetically as velar nasal. In the subsequent subsections we look at how Akan source syllabic nasals are adapted in Ga. 5.4.1. Adaptations of Akan source word-initial syllabic nasals Akan allows a sequence of two nasal consonants in a word; either in the initial position or in the medial position, which Ga does not usually permit in its regular morphology. As a common phonological adaptation pattern for such illicit sequence, Ga speakers may either delete the offending nasal segment or insert a vowel to break up such sequence. In this subsection, we will observe that the former pattern is employed systematically in the adaptation process. I will use ‘N’ mostly to refer to syllabic nasal throughout the rest of this chapter. University of Ghana http://ugspace.ug.edu.gh 180 5.4.1.1. Where the source word-initial syllabic N is deleted In this sub-subsection, we discuss how Akan source word-initial N in the contexts of NC and N1N2, where N1 is the syllabic nasal, is realised in Ga. (68) Akan Ga Gloss i. ŋk̀ràń́̃ ꜝtɪ ́ klà́ńtè cutlass, machete ii. ŋk̀ɔŋ̀̃ k̀ɔǹ̀̃ sa ́̃ kɔŋ̀̃ k̀ɔŋ̀̃ s̀a ́̃ gossip, betrayal iii. ŋk̀ɔŋ́̃ ḱɔŋ́̃ ́ kɔŋ́̃ k̀ɔŋ́̃ ̀ a dry, chesty cough; whooping cough iv. m̀pɛṕɛẃ́ kp͡ ɛḱp͡ ɛó́ remains of the harvest v. m̀pókᶣíá k͡pòkùá young breasts From the examples in (68) above, it can be observed that the Akan source word- initial syllabic nasals have lost correspondents in the adapted forms in Ga. The deletion of this initial syllabic nasals in these examples cannot be attributed to phonotactic constraint since Ga permits the NC sequences in its regular phonological system, albeit most are loaned, in words such as ŋk̀áŋĺɛ ́ ‘Sciaena nigrita, marine fish’ (cf. Dakubu 2002, 2009b), and even in some loaned words of Akan origin such as ŋkp͡òkùá ‘young breasts’, as variant of (68v), for the source m̀pókᶣíá, and ŋk̀òɲaà̃ ́̃ ‘magic, magical rite; a kind of charm’ for ŋk̀ùɲaà̃ .́̃ University of Ghana http://ugspace.ug.edu.gh 181 It is not only in the NC contexts where the N is deleted. The N1 is also deleted in N1N2 contexts at the word-initial position, as can be seen in the examples in (69). (69). Akan Ga Gloss i. m̀mɔd́ɪŋ́̃ ́ mɔd́̃ ɛŋ́ ́ effort, esp. successful/laudable effort ii. ɲɲaà̃ ǹ̃ ɪ ́̃ ɲaǹ̃ é a type of bead iii. ǹnòm̀múḿ nòḿ slave, servant iv. m̀mɔb̀ɔm̀mɔb́ɔ ́ mɔb́ɔm̀ɔb́ɔ ̀ very pitiful v. m̀mʊá́ wá múáwá germ, virus In a similar pattern to the strategy adopted for (68), the word-initial syllabic nasals are systematically dropped in the adapted forms. Unlike in the examples in (69), there are no instances of the NN sequence at word-initial position in any native word in Ga. Therefore, there can be a rule in Ga that strictly bans a sequence of NN, but the rule may be relaxed in NC contexts as there are many exceptions to that in Ga. In Figure 4, a spectrogram showing the adaptation of the Akan source word-initial NN by deletion to support the analysis of the examples in (69). University of Ghana http://ugspace.ug.edu.gh 182 Figure 4: A spectrogram showing the syllable structure of the adapted Ga word mɔ̃́ dɛ̃́ ŋ̃́ ‘laudable effort’ It can be seen from Figure 4 above that the four-syllable source word in reduced to a three-syllable word in the adapted form by Ga speakers. The syllable structure reduction is done through the deletion of the source word-initial syllabic nasal. 5.4.1.2. Where source word-initial syllabic N is adapted as V We have so far discussed the application of deletion as a strategy to repair some word-initial syllabic nasal in NC and NN contexts. In this subsection, we look at the adaptations of source word-initial syllabic nasals through vocalic replacement. In the examples in (70), I show that Ga speaker can sometimes adapt Akan source word-initial syllabic nasals as a vowel, though Ga permits the syllabic nasals at the word-initial position in both the regular phonology and the loanword phonology. University of Ghana http://ugspace.ug.edu.gh 183 (70). Akan Ga Gloss i. ɲʨèʨèrèé àkèkèré, àkèklè́ headband, headgear, crown ii. ŋk̀òɲaà̃ ́̃ àkùɲaà̃ ́̃ magic, magical rites; a kind of charm iii. m̀púnim̀́̃ pú àpúnam̀̀̃ pú head-on, face-to-face, on the spot iv. ǹtɔm̀́mɔ ̀ àtómò sweet potato v. m̀pɪm̀̀̃ ma ́̃ am̀̀̃ pèma ́̃ a type of plantain From (70), we see that the source word-initial syllabic nasals are adapted as [a] in Ga. So far in all cases of such adaptation, the low vowel [a] is the only vowel Ga speakers employ as a replacement vowel. The example in (70v) rather shows the preservation of the word-initial nasal in addition to the prefix [a]. Following the adaptation pattern, we would have expected it to be realised as *ap̀̃ èma,́̃ where the syllabic nasal will be replaced with the low vowel. Is the source word-initial syllabic nasal maintained in addition to the prefixation of [a] meant to avoid a possible lexical clash in the Dangme? In other words, is it avoided by the speakers to ensure morpheme/word distinctiveness? From the Ga dictionaries, there does not seem to be like *pèma ́̃ and *ap̀̃ èma ́̃ do not seem to exist. Therefore, realising the source as either of the two would not have led to any lexical clash. Perhaps further investigations into this kind of adaptation might help explanation the phenomenon. University of Ghana http://ugspace.ug.edu.gh 184 5.4.2. Adaptations of source word-medial syllabic nasals It is not only at the initial position that a syllabic nasal is deleted, but also at the word-medial position as mentioned earlier. In this subsection, we will discuss the adaptations of word-medial syllabic nasals in NC and NN contexts. (71). Akan Ga Gloss i. bèŋk̀ũḿ àbɛk̀ú left hand, left arm, left side ii. àbèŋk̀ù̃ma ́̃ àbɛk̀úlɔ ̀ left-handed person; awkward, clumsy person iii. pɛs̀ɛm̀ɪŋ́̃ ḱʊ́m̃ ɪɲ́̃ a ́̃ pɛs̀ɛk̀úméɲá selfishness From the examples in (71), it can be observed that the word-medial NC is repaired by deleting the N, as exemplified above. However, there are many exceptions to this pattern in Ga phonology; both the native phonology and the loanword phonology do not ban such sequence at medial position. For example, kontommire ‘leaf of the cocoyam’ is adapted as kontomle/kontomre in Ga. The question that requires further investigation then is whether or not this adaptation is perceptually motivated. (72). Akan Ga Gloss i. kòǹtóḿmìré kòǹtómlè́, kòǹtómrè́ leaf of the cocoyam, ii. m̀mɔb̀ɔm̀mɔb́ɔ ́ mɔb́ɔm̀ɔb́ɔ ̀ very pitiful University of Ghana http://ugspace.ug.edu.gh 185 iii. ǹnòm̀múḿ nòḿ slave, servant iv. ǹtũḿmʊá́ ǹtúmúá a tiny biting fly, sun fly v. ŋk̀ań́̃ naŕ̃ ɪ ́ ŋk̀áꜝ̃ naĺ̃ é rust vi. ɲɲ̀iǹ̃ am̀̀̃ ɲiǹ̃ am̀̀̃ ɲràm̀̀̃ ɲràm̀̀̃ brightly, sparkling, sending out rays From (72) above, we observe that just as at the word-initial position, an NN sequence is not permitted in adapted forms in Ga. The only exception in the data is in (72vi) in which a sequence of two nasal consonants is found in an adapted form from Akan. And even that, a closer look shows that the nasals have different places of articulation, which should make it salient enough for Ga speakers to distinctly perceive them, as opposed to the double nasals of the same places of articulation in the rest of the examples. From the examples above, there seems to be a constraint that bans the sequence of nasals of the same place of articulation in Ga. However, this constraint can be relaxed to some extent on such sequence at the word-medial positions as the examples have clearly shown. Such OCP effect of avoiding nasal consonants in successive or adjacent syllables is evident and strong at word-medial positions as we have already seen. 5.4.3. Adaptations of Akan source word-final syllabic nasal The position in a word where Akan source syllabic nasal are usually allowed in the adaptation process is the word-final position. Though it is not all the nasal University of Ghana http://ugspace.ug.edu.gh 186 consonants that are permitted word-finally, we consider the realisations of these syllabic nasals of Akan source words. In the following examples, I discuss cases where source word-final syllabic nasal is deleted in some adapted forms. I show that deletion is motivated by the phonology of Ga, which does not usually permit the bilabial nasals at word-final position. (73). Akan Ga Gloss i. bèŋk̀ũḿ àbɛk̀ú left hand, left arm, left side ii. àdéŋḱũḿ àdéŋḱũ̀ a calabash rattle, a kind of music in which a rattle is used iii. àpɪḿ́̃ àk͡pé thousand, one thousand iv. pæ ̃̀m̀píḿ pam̀̀̃ pí trap, snare; obstruction to progress From (73), it is clear that the Akan source word-final syllabic nasal is dropped in the adapted forms in Ga. By deleting this final nasal consonant, there is a mismatch in terms of syllable quantity between the source and the adapted forms. There are exceptional cases to this pattern in which the source word-final nasal is clearly preserved in the adapted forms. For example, the source words, ʨɥɪm̀̀̃ ‘all together’ is adapted as ʧʷɛm̀̀ *ʧʷɛ;̀ while ɔb̀ʊ̀ǹsaḿ́̃ ‘evil’ is adapted as àbòǹsáḿ, àbòǹsáŋ,́ *àbòǹsá, and æ̀hũḿ ‘wind, storm, gale, blast of wind’ as àhũḿ, University of Ghana http://ugspace.ug.edu.gh 187 àhũŋ ́*àhṹ, etc. From these exceptional cases, it is clear that the decision to delete or not the source word-final syllabic nasal is without recourse to the environment, that is, the preceding vowel has no influence on the deletion. In the following examples, the Akan source word-final syllabic nasals are realised as vocalic segment either through compensatory lengthening of the immediately preceding vowel or they are adapted as a high front vowel. (74). Akan Ga Gloss i. à̃ǹsam̀́̃ aŋ́̃ śaà̃ ́̃ guinea fowl ii. àdam̀́̃ àdaà̃ ́̃ a type of music played by hunters’ assoc. iii. kiŋ̀̃ k̀aŋ̀̃ ŋ̀ ̀ kiŋ̀̃ k̀aà̃ ̀̃ fine, rich esp. of women’s clothing iv. æ̀ʥań́̃ àɡaì̃ ́̃ arrow v. dᶣìàdań́̃ dùáꜝdaí̃ ́̃ a building of the wattle and daub type vi. àdam̀́̃ àdɛì̃ ́̃ a type of music played by hunters’ assoc. From the examples in (74), it can be observed that the Akan source word-final syllabic nasal is either realised through deletion and subsequent compensatory lengthening, as seen in (74i) – (74iii), or with a high front vowel, as can be seen in (74iv) – (74vii). It is worth noting that the high tone of the source syllabic nasal is systematically preserved in the vowels with which they are adapted in Ga. Final alveolar nasal, being adapted as the front high vowel [i] as in (74iv) – (74v), University of Ghana http://ugspace.ug.edu.gh 188 follows the pattern well noted of the Asante dialect of Akan in the realisation of the word-final alveolar nasal at the phonetic level of representation. There is a unique process in the Asante dialect whereby an underlying word-final alveolar nasal is realised as the back high unadvanced vowel [ɪ] at the phonetic level of representation (see Dolphyne 1988, Abakah 2005a, among others for detailed discussion). It is a fact of Ga phonology that an underlying word-final /n/ is usually realised as [ŋ] at the phonetic level of representation (cf. Dakubu 2002), so one would have expected this pattern to systematically reflect in the adaptation of Akan source word-final [n] in Ga. The question then is, could the source loan words be the Asante variants, that is, with the high vowel, rather than the nasal at the word-final position? It is a possibility that those words might have historically entered Ga as the Asante dialect variants. Alternatively, the present adapted forms might have possibly entered Ga as Akuapem variants (i.e. with the final syllabic nasals). However, these words might have synchronically undergone influence by the Asante variant since the Asante dialect is perhaps the most influential dialect in terms of contact with Ga speakers. 5.4.4. Truncation of Akan source syllable structure In the final discussion of the adaptation of Akan source prosody in Ga, I look at some instances where the Akan source syllable structure is reduced in Ga in the adaptation process. Some of these reductions are instantiated by structural well- formedness requirements, while others are not. I exemplify this adaptation pattern in the examples in (75). University of Ghana http://ugspace.ug.edu.gh 189 (75). Akan Ga Gloss i. àpásá pásá tell lies; gossip ii. kũmaá̃ ́̃ kũma ́̃ junior iii. æ̀nim̀̃ ù̃òɲaḿ́̃ ànũ̀ɲaḿ́̃ glory iv. pɛs̀ɛm̀ɪŋ́̃ ḱʊ̃́mɪɲ́̃ a ́̃ pɛs̀ɛk̀úméɲá selfishness v. ǹnòm̀múḿ nòḿ slave, servant From the examples in (75), it can be seen that the syllable structure of the source words ais reduced or truncated in the adapted forms in Ga. This truncation here is not the result of the repairing of ill-formedness as we have observed so far in the discussions of the adaptations into Ga since the truncated syllables are not in any way ill-formed. For example, in (75i) and (75ii), the truncated monosyllable [a] is permitted at both word-initial and word-final positions in Ga. In examples (75iii) – (75v), on the other, two syllables in each word are truncated. Similarly, the truncation is not a strategy to repair ill-formedness in terms of structure. In examples (75iii) - (75v), two contiguous syllables of the source forms are truncated in the adapted forms. This truncation, as has been mentioned earlier, might not be phonologically-motivated, but perhaps just to facilitate faster speech, a phenomenon which is common across languages especially during first language acquisition by children. For example, a sequence of two vowels of the same University of Ghana http://ugspace.ug.edu.gh 190 quality as in (75ii) exists in Ga (cf. Dakubu 2002), therefore, reducing the sequence in the adapted forms cannot be phonologically-motivated. 5.5. Adaptations of Akan source syllabic nasals in Dangme In chapter one of this thesis, it was noted that the inventory of the syllabic nasals in Dangme is limited to the bilabial nasal [m], and even that Dangme speakers make efforts to completely avoid its occurring at the phonetic level of representation (cf. Dakubu 1987). In the following subsections we discuss the realisations of the Akan source syllabic nasals in Dangme. The subsection also discusses realisations of the syllable structures of some Akan words in Dangme. 5.5.1. Adaptations of source word-initial syllabic nasals Just as was mentioned in the preceding section, Akan permits the syllabic nasal at all positions of the word. By word-initial position, I refer to NC or NN contexts. In the subsequent subsections, we consider the strategies Dangme speakers employ in adapting the Akan source syllabic nasals in the adaptation process. In the following examples, the Akan source word-initial syllabic N is deleted in the context of NC in the adapted forms in Dangme. (76). Akan Dangme Gloss i. (tírí) ŋḱʷa ́̃ (tlí) kũ̄a ̄̃ expression of congratulations ii. ǹsékúró sìékú, sékú slander, gossip iii. ŋk̀ũ̀niḿ́̃ kùñ iḿ́̃ victory, success University of Ghana http://ugspace.ug.edu.gh 191 iv. ŋk̀áꜝtɪ ́ káꜝtíɛ ́ groundnut v. ɱ̀fòní fòní picture, photo From the examples in (76), it can be observed that the Akan word-initial syllabic nasals are systematically deleted in the adapted forms in Dangme. That is, unlike in Ga, Dangme speakers completely drop all syllabic nasals, including the velar nasal, which is permitted at the same position in Ga, as seen (76iii) – (76iv). To account for the complete prohibition of the Akan source word-initial syllabic nasal in the adapted forms in Dangme, there needs to be a position- sensitive constraint which will ensure this ban. This position-sensitive constraint is *σ1[+nas], and it is formulated, following Kenstowicz (1996) and de Lacy (2001), as follows: (77). *σ1[+nas]: A syllable specified for [+nasal] are prohibited at the initial syllable of a word The constraint, *σ1[+nas] prohibits, without any exception, any syllabic nasal at the word-initial position of Akan source words in Dangme. The constraint *σ1[+nas] has to dominate both the anti-deletion and the anti-insertion constraints; DEP-IO and MAX-IO to ensure that the output form that employs deletion of the prohibited initial syllabic nasal emerges the optimal candidate. University of Ghana http://ugspace.ug.edu.gh 192 Tableau 26: The deletion of the Akan source word-initial syllabic nasal in Dangme /ɱ̀fòní/ IDENT- DEP-IO MAX-IO ‘picture, photo’ IO(nas) *σ1[+nas] a. [ɱ̀fòní] *! b. [fòní] * c. [àɱ̀fòní] *! Constraint ranking: IDENT-IO(nas), *σ1[+nas] » DEP-IO» MAX-IO From Tableau 25, candidate (a) presents a perfect correspondence with the input word in terms of segmental match. By faithfully preserving the word-initial syllabic nasal, candidate (a) fatally violates the highly-ranked position-sensitive *σ1[+nas]. Candidate (c), on the other hand, repairs this illicitness in the input word- initial syllabic nasal by inserting the vocalic segment [a] leftwards to realise it in an accepted form [a ̀ɱ̀fòní], though this insertion is in fatal violation of the crucially-ranked DEP-IO. By employing deletion as the repair strategy for the illicit input word-initial syllabic nasal, candidate (b) violates MAX-IO. Though candidate (b) violates this anti-deletion constraint, yet it emerges the optimal output form due to the low ranking of MAX-IO. The deletion of the Akan source word-initial syllabic N is not limited to only NC contexts in Dangme. In the following examples, we consider the deletion of the same source syllabic nasals in N1N2 context, where the N1 is the syllabic nasal. University of Ghana http://ugspace.ug.edu.gh 193 (78). Akan Dangme Gloss i. m̀mɔd́ɪń́̃ mɔd́é effort, esp. successful effort ii. m̀mùsú mòsú, mùsú abomination, taboo, ill luck iii. m̀mɔb́ɔ ́ mɔb́ɔ,̀ mɔḿɔ ̀ pity, sorrow iv. m̀mààwá maẁ̃ á servant v. m̀màrá mlà́, mlàá law From the examples in (78), it is observable that the source word-initial syllabic nasals are systematically deleted in the adapted forms in Dangme. What we have observed so far is the prohibition of N1 in an N1N2 sequence at the word-initial position in Dangme. In Figure 5 below, I provide a spectrogram showing the deletion of the Akan source word-initial syllabic nasal in the adapted form in Dangme. Figure 5: A spectrogram showing the syllable structure of the adapted Dangme word ‘pity, sorrow’ mɔ̃́ bɔ̀. University of Ghana http://ugspace.ug.edu.gh 194 From Figure 5, it can be seen that an Akan trisyllabic word is realised as disyllable in Dangme. The initial syllabic nasal has been deleted. This 5.5.1.1. Where word-initial N is realised as V In this sub-subsection, I look at the vocalic replacement of the illicit Akan source word-initial syllabic nasal with vowels in Dangme as a phonological adaptation pattern. (79). Akan Dangme Gloss i. ŋk̀áꜝtɪ ́ ákáté groundnut ii. m̀màá àmàá clappers used for musical time keeping iii. ɱ̀fòní òfòní picture, photo In the examples in (79), rather than deleting the phonotactically restricted syllabic nasal at the word-initial position, Dangme speakers replace them with vowels in the adapted forms. In (79i) – (79ii), the source word-initial [ŋ] and [m] are realised as [a], while [ɱ] is realised as [o] in (83iii). A closer look at the replacement process seems to point to a pattern of first stem-internal vowel copying to the place of the deleted word-initial syllabic nasal in the adapted forms. For example, the first stem-internal vowel [a] is copied in place of the deleted syllabic nasal in (79i) – (79ii), while the first stem-internal [o] University of Ghana http://ugspace.ug.edu.gh 195 replaces the source word-initial syllabic nasal. I exercise caution here with this line of explanation because such examples are too few in the data gathered for the present study to be able to make a well-grounded generalisation. As it stands now, the explanation for this replacement process seems to reside in Dangme speakers’ strategy to preserve the source syllable quantity in the adapted forms. 5.5.2. Adaptations of Akan source word-medial syllabic nasals In this subsection, we will know if this prohibition is extended to word-medial and word-final positions in Dangme before we can generalise this prohibition. Aside from the word-initial position where the Akan source syllabic nasal is banned, such syllabic nasals are banned in the word-medial position. In this subsection, the adaptations of the Akan source word-medial syllabic nasals are discussed, where it will be shown that in both NC and NN contexts, the source syllabic nasal is usually deleted in the adapted forms in Dangme just as was observed with Ga. I begin the discussions with the adaptation of the syllabic nasals in the NC context. The first context of the deletion of the source word-medial syllabic nasal is the NC context. In the following examples, we observe that the medial syllabic nasals are deleted in the adapted forms. We will soon find out that there are some few exceptions to this deletion pattern. (80). Akan Dangme Gloss i. àm̀pèsí àpèsí boiled green plantain ii. æ̀tùm̀pań́̃ àtùpaà̃ ḿ̃ , àtup̃̀ àmí, a large talking drum University of Ghana http://ugspace.ug.edu.gh 196 iii. maḿ́̃ pam̀́̃ maṕ̃ ām̄ monitor lizard iv. bòǹsù bòsò whale v. kòǹtrò̀ɱ̀fì kòtròfì chimpanzee vi. ŋk̀óńtóḿpó kútúpɔ ́ undermining, intrigue From (80), it can be seen that Akan source word-medial syllabic nasal as the ante- penultimate syllable is dropped in the adapted forms as seen in (80i) – (80iii). Similarly, the target syllabic nasal in the penultimate position is also dropped as in (80iv) – (80vi). Again, in both (84v) and (80vi), both stem-internal syllabic nasals are dropped in the adapted forms. There are very few exceptions to this word- medial syllabic nasal deletion pattern in the adaptations of Akan source words such as ŋk̀óńtóḿpó ‘undermining, intrigue’, which can also be adapted as kútúḿpɔ.́ In this example, the penultimate syllable is preserved in Dangme. There is also the preservation of the medial syllabic nasal in adapting the Akan source word ʥɥòŋk̀ú ‘thigh’, which is realised as ɡ͡bòŋɡ̀ú, and not *ɡ͡bòɡú in Dangme. In the following discussions, we look at the adaptation of the same Akan source word-medial syllabic nasals, but in the NN context. In the examples it will be shown that just as obtains in Ga, the NN sequence is not permitted word- medially in Dangme words of Akan source. University of Ghana http://ugspace.ug.edu.gh 197 (81). Akan Dangme Gloss i. àkɔń́nɔ ́ àkɔńɔ ́ lust, strong desire ii. m̀mɔb̀ɔm̀̀mɔb́ɔ ́ mɔb́ɔm̀ɔb́ɔ ̀ pitifully, sadly iii. dòǹnó òdònó hour-glass drum iv. ɔk̀úḿbɔm̀̀mɔf̀ʊ́ɔ ́ òkúḿ-mɔm̀ɔf̀ūɔ the poisonous twig or vine snake The examples in (81) also show that the sequence of two nasal consonants is not permitted in Dangme. And as a repair strategy, the syllabic nasal, which is usually the first of the two nasals, is dropped. The example in (81iv) seems to be an exception to this pattern. However, the adapted form in Dangme seems to be a two-word compound than a singleton word. This might explain its exceptionality. From the discussions in (78) and (81), it seems appropriate to say that Dangme speakers do not permit the NN sequence in the adaptation of the Akan source words. This ban can be expressed in the structural well-formedness constraint formulated and defined below: (82). *NN A sequence of two nasal consonants is prohibited. Though the constraint *NN does not make any particular reference to the syllabic nasal, it is assumed that since neither the Akan source nor the target Dangme permits consonant clusters, one of the Ns, consistently the N1 is syllabic. Again, University of Ghana http://ugspace.ug.edu.gh 198 the same constraint can account for the ban on such nasal sequence at word-initial position as well as word-final position, as I shall discuss later on in this chapter. This constraint has to dominate the constraint against deletion, MAX-IO since as we have seen so far that it is deletion, and not insertion, is employed as the repair strategy. Tableau 27: The deletion of the Akan source syllabic nasal in NN context in Dangme /àkɔń́nɔ/́ IDENT- DEP-IO MAX-IO ‘lust, strong desire’ IO(nas) *NN a. [àkɔń́nɔ]́ *! b. [àkɔńɔ]́ * c. [àkɔń́dɔ]́ *! Constraint ranking: IDENT-IO(nas), *NN» DEP-IO» MAX-IO In Tableau 26, candidate (c) avoids violating the highly-ranked *NN through the denasalisation of the N2 to realise the word as àkɔń́dɔ,́ a form equivalent to the Fante variant of the word. By so doing, it rather fatally violates IDENT-IO(nas), an equally highly-ranked constraint. Candidate (a), the most faithful output form to the input, is also ruled out of the competition due to its fatal violation of *NN for preserving the NN sequence in the input. By deleting the N1 of the N1N2 sequence, candidate (b) avoids violating the high-ranking constraint against such sequence, *NN. However, this satisfaction is at the expense of violating the anti-deletion constraint, MAX-IO. Candidate (b) emerges the optimal candidate despite this violation because of the low ranking of MAX-IO in the present Tableau. University of Ghana http://ugspace.ug.edu.gh 199 5.5.3. Adaptations of source word-final syllabic nasals As noted above, Dangme does not generally permit the syllabic nasal at the word- final position with the exception of [m], which even occurs only in loaned words from Akan (cf. Dakubu 1987). As a result of this, several repair strategies are usually employed when adapting the Akan source word-final syllabic nasals. In this subsection, we discuss the adaptations of the Akan source word-final syllabic nasals by Dangme speakers. The discussions focus on two main strategies employed by the speakers in the adaptation process, namely deletion and vocalic replacement. It will be shown that usually the speakers either delete or replace the same word-final syllabic nasals when adapting words from Akan. It will also be shown that the speakers may avoid the syllabic nasals in word-final position through epenthesis. I begin the discussions with the use of deletion as a strategy for adapting Akan source word-final syllabic nasals. It is not only at the non-final positions that Dangme speakers adapt Akan source syllabic nasals. In the examples in (83), I provide instances of the deletion of the target nasals at word-final position. (83). Akan Dangme Gloss i. æ̀tùpáń àtuk̃̀ ͡pá bottle, esp. beer bottle ii. m̀mɔd́ɪń́̃ mɔd́é effort, esp. successful effort iii. æ ̀ʥɥàmań́̃ àʤùam̀̃ a ́̃ prostitute; fornication University of Ghana http://ugspace.ug.edu.gh 200 iv. àpɪḿ́ àk͡pé thousand As can be seen in (83), the source word-final syllabic nasals [n] and [m], as in (83i) – (83iii) and (83iv) respectively do not surface in Dangme. There are, however, few exceptions to this generalisation where the final nasal either surface with the same place of articulation as the source or with a change in the place of articulation as in the adaptation of Akan source æ̀tùm̀pań́̃ ‘a large talking drum’, with an alveolar nasal in the final position, as àtùpaà̃ ḿ̃ , where the final alveolar nasal [n] is realised as the bilabial nasal [m]. There are also examples of the complete preservation of the source word-final nasal as in ɔb̀ʊ̃ǹ̀saḿ́̃ ‘devil’ being adapted as àbòsáḿ, àbùsáḿ. 5.5.3.1. Where word-final syllabic N is adapted as V The Akan source word-final syllabic nasals, aside from they being deleted in the adapted forms in Dangme, can also be adapted as a vocalic segment. In the following examples, unlike what obtained in our discussion of similar syllabic nasals adaptations at the word-initial position, the high front vowel is systematically the vowel with which these syllabic nasals are adapted. University of Ghana http://ugspace.ug.edu.gh 201 (84). Akan Dangme Gloss i. diń́̃ ń dií̃ ̀̃ reserved, habitually quiet; quietly ii. æ̀ʥań́̃ ɡaí̃̃ bow iii. sùbań́̃ súbaí̃ ́̃ character iv. æ ̀ʥɥúmáʥań́̃ àʤúmaʤ́̃ aí̃ ́̃ unprofitable labour v. tań́̃ taí̃ í̃ ́̃ unpleasant, ugly It can be observed from the examples in (84) that the Akan source word-final alveolar nasal is adapted as [i]̃ . This particular realisation very closely resembles the Asante Twi phonetic realisation [dĩ́ĩ ]̃́ ̃ of the word. An exception to the high vowel as the replacement vowel can be found in the adaptation of the source ʥɥɪŋ́̃ ́ ‘reason, think’, in which the final velar nasal [ŋ] is realised as [ɛ], but not [i] as in ʤʷũɛ̀.̀̃ The word-final alveolar nasal being adapted as a high front vowel is very consistent with what obtains in the Asante dialect of Akan. In the Asante dialect, the word-final alveolar nasal in the underlying level of representation is systematically realised as [ɪ ̃]. However, the explanation for the process in the Asante dialect is not that the word-final [n] changes to [ɪ ̃], but that the nasal is systematically followed by the high front vowel in the underlying level of representation in a CVNV structure. Therefore, the intervocalic N rather deletes University of Ghana http://ugspace.ug.edu.gh 202 passing its nasal feature to the final V, and not that it is replaced with the high vowel (cf. Dolphyne 1988; Abakah 2004, 2005a; among others). Just as the process affects only the alveolar nasal in the Asante dialect; it is only the alveolar nasal that is adapted as the high front vowel in adaptation process in Dangme. It is not yet established in Dangme phonology that this intervocalic nasal elision applies; therefore, it can be assumed for the purpose of the present discussion that the source that Dangme speakers adapt might be specifically the Asante dialect (Asante variants) of Akan. 5.5.3.2. Where V is epenthesised to the word-final syllabic N A third strategy by which Akan source word-final syllabic nasal can be adapted is epenthesisation. In this process, the source final nasal is preserved, however, a vocalic segment is epenthesised to its right to reduce its sonority. I exemplify this strategy in the examples in (85). (85). Akan Dangme Gloss i. æ̀hũḿ àhũmì storm, gale ii. ʨɪńaḿ́̃ kénaḿ̃ i ́̃ fried fish iii. ʥɥáḿ ʤūāmì inside the market iv. æ̀tùm̀pań́̃ àtù̃pàní, àtũ̀pàmí a large talking drum v. súńsꜝúḿ sūsūmá spirit, image University of Ghana http://ugspace.ug.edu.gh 203 In the examples in (85), the Akan source word-final syllabic nasals lose their syllabicity in the adaptation process; they assume the onset status after the epenthesisation of the vowels. This loss of the syllabicity is indicated by the shift of the tone bearing units from the source word-final nasals to the epenthetic vowels in the adapted forms. It is worth pointing out that, as would be expected, one Akan source word-final syllabic nasal can be adapted differently perhaps due to the dialectal differences among Dangme. For example, the Akan source word ɔb̀ʊ̀ǹsaḿ́̃ ‘devil’ can be adapted as àbòsáḿ, àbùsáḿ, abosiá̃ ḿ́̃ , àbùsiá̃ ḿ́̃ , àbùsiá̃ ḿ̃ i,́̃ according to Dakubu (2013). 5.5.4. Truncation of Akan source syllable structure Just as was observed in the discussions of the adaptations of the Akan source syllable structures in Ga, there are instances also in Dangme where the syllable structure of the Akan source word can be truncated in the adaptation process, though limited. These syllable structures that are targeted for the truncation are the monosyllable CV, and the disyllable VV and CVN, where the N is the syllabic nasal. This truncation can occur either word-medially or word-finally in Dangme. I provide examples of this truncation process below in (86). (86). Akan Dangme Gloss i. àsàbrà̀nɪ ́ àsàblà́ appellation, title ii. æ̀ɥísiá̃ á̃ ́̃ àhũsa ́̃ orphan, orphanage University of Ghana http://ugspace.ug.edu.gh 204 iii. pɛs̀ɛm̀ɪŋ́̃ ḱʊ́m̃ ɪɲ́̃ a ́̃ pɛs̀ɛk̀úmíɲa ́̃ egotism, self-centredness In (86i), the source word-final CV is truncated in the adapted form, whereas disyllables VV and CVN are truncated in (86ii) and (86iii) respectively. Just as was explained for a similar process in Ga, the motivation for this process in not clear since the sizes of the source syllables do not seem to violate those of Dangme. Again, it is worth pointing out the interesting targets for the truncation process in (86ii). Following the regular pattern in such instances, we would have expected the final two syllables targeted, instead of the preceding [i], for the truncation process. This adaptation may be due to misperception on the part of Dangme speakers especially when the nasal front high [i]̃ , which is specified for [cor], glides into the following nasal central low [ã]. Even though the distance between the two vowels are wide apart in the vowel space, [i]̃ , being immediately preceded by [s], which also a coronal segment, can cause some difficulty in its perception. For similar studies on the cross-linguistic differences in perception of coronals in CV combinations by children, see van der Feest et al 2016). 5.6. Summary of the chapter This chapter has discussed the strategies Ga and Dangme speakers employ in adapting Akan source nasality and source syllable structures. The adaptations of the nasal feature focused on the discussions of how the speakers adapt the [+nasal] feature on Akan source vowels, in particular, non-native vowels. It was concluded University of Ghana http://ugspace.ug.edu.gh 205 that with the exception of instances where the vowels with which the source non- native vowels are adapted can be nasalised, the source nasality is dropped in the adapted forms, a pattern which is in consonance with what generally obtains in the general phonology of the two target languages (cf. Dakubu 1987, Dakubu 2002, among others). It was also observed that there are instances where, though a source nasalised vowel, which is also nasaliseable in the target languages, is denasalised in both target languages in the adaptation process. To license the denasalisation process in Dangme, speakers were observed to drop the neighbouring nasal consonant. However, in Ga, it was observed that the denasalisation can take place without dropping the neighbouring nasal consonant. Similarly, a source oral vowel is sometimes nasalised in the adapted forms, even in non-nasalising environment, especially in Ga. With regard to the adaptations of the Akan syllabic nasals in both Ga and Dangme, it was shown that at the word-initial position, the syllabic nasal may or may not be preserved in the adapted forms in Ga, whereas they are completely dropped in Dangme. These same initial syllabic nasals can also be realised as vowels, particularly the low vowel in both target languages. At the word-medial position, on the other hand, both languages do not usually permit such NN sequences, thereby reducing such sequence through deletion of the syllabic nasals in such contexts. For the adaptations of the Akan source word-final syllabic nasals, the two target languages differ to some extent. While some of the nasals, particularly [m], can sometimes be preserved in the adapted in both languages, the other nasals are either deleted or are realised as front high vowels. In Ga, there is University of Ghana http://ugspace.ug.edu.gh 206 also an additional process of compensatory lengthening that sometimes accompanies the deletion of those word-final syllabic nasals. Dangme speakers also exclusively apply epenthesis of a vowel, usually a front high vowel, as a phonological pattern for the adaptation of word-final syllabic nasals. In OT, these adaptation patterns were systematically observed to have resulted from the highly- ranking of the markedness constraints *σ1[+nas] for initial syllabic nasals, and *NN for a sequence of syllabic nasals elsewhere, over the faithfulness constraints DEP- IO and MAX-IO. By this ranking, either deletion alone or deletion and subsequently insertion (by replacing the deleted syllable) is resorted to as the adaptation pattern. The chapter concluded with the discussions of the adaptations of the syllable structures of some of the adapted forms from Akan in Ga and Dangme. It was observed that between the two target languages, CV monosyllables and CVN or VV disyllables could be targeted for the truncation process. It was realised that the truncation might not be phonologically-motivated, but perhaps it is only meant to facilitate speech among the speakers of the two target languages. In the table below, I present the similarities as well as the differences in how the prosodic features discussed in this chapter are realised in both Ga and Dangme. University of Ghana http://ugspace.ug.edu.gh 207 Table 13: Adaptations of source prosodic features in Ga and Dangme Akan prosodic features Ga Dangme Nasalised vowel: [ɪ ̃] Oral when realised as Oral when realised as [e] or [ɛ] [e], but nasal/oral when realised [i] or [ɛ] [ʊ̃] Oral when realised as Oral when realised as [o], but nasalised [o], but nasalised when realised as [ɔ] when realised as [u] Can oral source be nasalised Yes Yes in non-nasalising context? Can source nasalised be oral Yes No denasalised in nasalising context? Denasalisation through No Yes deleting the trigger nasal consonant? Syllabic nasals: Initial NC N can delete N always deletes Initial NC N can also become V N can also be realised as V Initial N1N2 N1 always deletes N1 always deletes Medial N1N2 N1 always deletes N1 always deletes Final N N can delete N always deletes when non-labial University of Ghana http://ugspace.ug.edu.gh 208 CHAPTER SIX AKAN TONES ADAPTATIONS IN GA AND DANGME 6.1. Introduction This chapter is dedicated to the discussions of how the tonal melodies of some Akan source words are realised in both Ga and Dangme. The discussions are limited to the adaptations of some Akan source lexical tone on the noun/nominal class to the exclusion of the discussions of other lexical classes because the issues of tone are generally complex and broad to be comprehensively dealt with in a chapter. Even with the noun class, the discussions are narrowed down to only the noun/nominal stems in isolative construction, without mentioning nouns in possessive/associative construction. The simply reason for this is that in such constructions, even the three major dialects of Akan, which serve as the sources of adaptation for Ga and Dangme speakers, manifest them differently. In chapter one it was noted that though Akan, on the one hand, and Ga and Dangme, on the other hand, are tone languages, tone phenomena manifest differently among the three languages. It was mentioned that while both Akan and Ga, on the one hand, operate a two-level tone system, Dangme, on the other hand, operates a three-level tone system. This chapter, therefore, discusses how the lexical tonal melodies of the nominal stems that are adapted in both Ga and Dangme are realised. It also looks at the realisations of Akan source tone rules, including lexical downstep in some nominal stems in the two target languages. In this chapter, the tones are marked ( ́ ) for High, ( ̀ ) for Low, and ( ̄ ) for Mid. The use of H stands for High University of Ghana http://ugspace.ug.edu.gh 209 tone, M for Mid tone, and L for Low tone will also be observed. The symbol ꜝH represents downstepped High. I will use the symbol ꜝH interchangeably with the expression, downstepped H. Dolphyne (1988/2006) has identified two main types of Akan nominal stems, namely CV monosyllabic nominal stems and disyllabic nominal stems, which can be of any syllable size including CVV, CVCV, and CVN. According to Dolphyne (op cit), with the exception of a few cases, the CV monosyllabic nominal stems are realised with H tone. The disyllabic nominal stems, on the other hand, are subdivided into five distinctive types based on their tonal melodies. These disyllabic nominal stems are of the following types: HH, LH, HꜝH, LL, and HL (cf. Dolphyne 2006: 76-78). The HL, according to Dolphyne (op cit) and Abakah (2005b), is mostly found in English loans in Akan (cf. Dolphyne 2006: 78). It is worth noting that though Dolphyne identified only two types of Akan nominal stems based on their tonal melodies, namely monosyllabic and disyllabic, there are nominal stems of trisyllables and even polysyllables also in Akan (cf. Appah 2003: 6-7) that are adapted in both Ga and Dangme. Again, the disyllabic nominal stems can also be of CVS structure, where the ‘S’ stands for any syllabic consonant (not only nasals) in Akan. By the use of the term nominal stems, I redundantly factor out the nominal affixes (both prefixes and suffixes) in the discussions since we know, from chapter three, that these affixes may not be adapted in the target languages. It is worth pointing out that the nouns in Akan have also been classified by other criteria by different authors. For example, Bodomo & Marfo (2006) have University of Ghana http://ugspace.ug.edu.gh 210 classified the nouns purely according to their morphological structure. Osam (2005), on the other hand, used their semantic properties to categorise nouns in Akan. Other authors too have discussed the classifications of nouns in Akan. However, my decision to follow Dolphyne’s (1988/2006) classification stems from the fact that she, in her classification, considers both the syllable structure and tonal patterns of the nominal stems which is consistent with the discussions in this chapter. In this chapter, the discussions focus on the adaptations of tonal patterns of nominal stems of different syllable sizes, namely monosyllables, disyllables, and polysyllables as they occur in both Ga and Dangme. I also briefly discuss the adaptations of Akan source tones in compound stems in both Ga and Dangme in this chapter. A tone rule that will engage our attention in this chapter is the downstep rule. The downstep rule in tone languages refers to the process of lowering the pitch range of H after a L-toned syllable so that a targeted H tone is lower in pitch than the H tones that precede it (cf. Yip 2002: 11). In Akan, the phenomenon has received a considerable attention (cf. Dolphyne 1988/2006, Abakah 2005b, Kügler & Genzel 2011b, Kügler 2016, among others). In the subsequent sections, I discuss the realisations of the downstep phenomenon in both Ga and Dangme in the adaptation process. In this chapter, I provide spectrograms, where necessary, to show the pitch realisations of each syllable in Akan source words and as comparisons, I show how the source pitches are realised in both target languages. In the spectrograms, the horizontal direction represents time, which is also same as waveforms, and the University of Ghana http://ugspace.ug.edu.gh 211 vertical direction represents frequency, which is measured in Hertz (Hz). The greyish area has broadband shaded areas which is called the formants. The darker the formant is, the higher the energy densities, and vice versa. The blue dotted line across the formant represents the pitch levels (cf. Boersma 2011). Each spectrogram has been divided into five tiers (both interval and point tiers), namely Segment, Syllable, Tone (point tier), Word (phonetic), and Gloss. The relative pitch of the syllable is indicated at the end of each syllable. In measuring the pitch level, I took the mean value for each syllable. For the Dangme pitch spectrum, an average pitch range between the values of 85Hz and 100Hz is considered L. The average pitch for M is between 101Hz and 115Hz. The average pitch for H, on the other hand, ranges from 120Hz and above. Therefore, a pitch value that falls in between average values for M and H can be considered as a downstepped H in Dangme. For the Ga pitch spectrum, the average pitch for L is observed to range between 85Hz and 115Hz. The average pitch range for the H, on the other hand, is from 125Hz and above. Therefore, a pitch value which falls in between the values for the H and the L in context can be considered as that of a downstepped H in Ga. These pitch ranges used in this thesis are the averages of the pitches of three male speakers for each of the languages. The spectrograms analysed in this thesis are those of male Akan, Ga, and Dangme speakers. The rest of the chapter is organised as follows: section 6.2 discusses the adaptations of Akan source lexical tone on nominal stems in both Ga and Dangme. The discussions focus on how some lexical tonal melodies in Akan source monosyllabic, disyllabic, and polysyllabic nominal stems are realised in Ga and in University of Ghana http://ugspace.ug.edu.gh 212 Dangme. The discussions consider tonal adaptations in both monomorphemic and bimorphemic (compound) stems. The issue of how Akan source downstepped H is realised in both Ga, a language that also operates the lexical downstep rule, and in a language that is not noted for operating the lexical downstep phenomenon in its native tonology, is considered in section 6.3. The chapter focuses on how Akan source downstepped H can be realised as either H or L in Ga. I also look at how the source downstepped H, which is sometimes preserved in Dangme, can also be realised as H or M, and in some exceptional cases, as L in the same chapter. The discussions in this chapter are summarised in section 6.4. 6.2. Adaptations of Akan source lexical tone system in Ga and Dangme In terms of tone system, though both Akan and Ga operate a two-level tone system, tone does not manifest in the same ways in the two languages. Again, while Akan has had the tone patterns of its noun class clearly defined (cf. Schachter & Fromkin 1968, Osam 2005, Dolphyne 1988/2006, Abakah 2000, 2005b, Bodomo & Marfo 2006, among others), not much has been done in Ga on Ga nouns with the exception of Wentum (1997), who studied the tones of Ga noun classes. She categorised Ga nouns based on their tone patterns. The study of tone in Dangme, in general, has similarly received less attention, and even the most recent important study of the phenomenon in Dangme dates back to the 20th century by Apronti (1967) and Dakubu (1974, 1987). According to Dakubu (1987: 19-20), there are three tone levels in Dangme. They are H, M, and L and they can be combined to derive several tone patterns, but the University of Ghana http://ugspace.ug.edu.gh 213 L and M tone levels never pattern together. There can be HH, LH, MH, MM, LL in two-syllable verbs, but not ML or LM. It is worth pointing out that the main dialectal subgroups that exist in Dangme realise the tone patterns quite differently. While the Inland dialects including Klo and Sɛ have LL different from MM verbs, the Coastal dialects such as Ada and Ningo do not have LL pattern at all (cf. Dakubu 1987). Therefore, the LL pattern in the inland dialects is realised as MM in the coastal dialects. Again, in the coastal dialects, a non-H adjacent to H is realised as L. So in terms of constraints, the Coastal dialects have the constraint *LM and *LL, which means the sequences of LM and LL are prohibited. These constraints do not apply in the Inland dialects, therefore, they are demoted in those dialects of Dangme. The foregoing, notwithstanding, Dakubu (1987:22) observes that within intonation phrases, only the Klo dialect that allows M to follow L in formal or semi-formal pronunciation. The other dialects do not permit this sequence at all in all contexts. However, a sequence of MML can be found in lexical items such as pɔp̄ɔl̄ì ‘earthworm’, though very limited cases (F. T. Owulah, personal communication, 18th July, 2018). Therefore, a potential LM sequence in the underlying level of representation can be repaired through raising the pitch of the L to M to derive MM in the Ada dialect, which can also occur in fast speech in the Klo dialect. Therefore, in effect, it is possible to realise a sequence of LMH in the Klo dialect in, for example, intonation phrases. The process of raising the pitch levels of a source L to M before M, has the objective of avoiding an ill-formed LM sequence in the Ada and the Klo dialects. Therefore, actual cases of the University of Ghana http://ugspace.ug.edu.gh 214 realisations of the downstepped H syllables in lexical items in Dangme seem to be limited to loanwords, probably from Akan (Dakubu 1987:24). This discussion seeks to shed light on noun classification in Ga and in Dangme based on their tone patterns. The discussions in this section focus on the adaptations of the tones of Akan source monosyllabic, disyllabic, and polysyllabic nominal stems in Ga and Dangme. It specifically looks at how the Akan source tonal melodies including LH (which is very common with noun classes, cf. Abakah 2005b), HH, LLH, and LHH are realised in Ga and Dangme. 6.2.1. Adaptations of tone of Akan source nominal stems I begin the discussions in this subsection with the adaptations of the Akan source tones including H, HH, LH, and LHH in monomorphemic stems in Ga and Dangme. The subsection ends with discussions of the realisations of the Akan source tones in compound (bimorphemic) stems in Ga and Dangme. This subsection discusses level tone adaptation. The discussions of tone rule of downstep are deferred to the latter part of this chapter. 6.2.1.1. Adaptations of Akan source H In the adaptation of the Akan source H-toned CV nominal stem, the pitch level of the source is almost always maintained in the adapted forms. I exemplify this in (87) and (88). University of Ghana http://ugspace.ug.edu.gh 215 (87). Akan Ga Gloss i. fɔ ́ fɔ ́ guilt ii. kʊ ́ kó wandering iii. ɛs̀ʊ̃́ sɔ ́̃ blame; responsibility iv. ɛb̀ɛ ́ àbɛ ́ proverb, adage; indirect speech act v. bɛ ́ àbɛ ̀ pincers From the examples in (87) above, it is observable that the Akan source H is systematically adapted as the same H in Ga, with the exception of (87v) which is the only exception throughout the data for the present study. The motivation for the change in tone in example (87v) by Ga speakers is not clear. We may suggest that perhaps it is to distinguish the adapted stem in (87v) from that in (87iv), since both have the same orthographic form. This postulation does not seem to hold for the adaptations of the Akan source word baŋ́̃ ́though, since the same form with the same tonal marking is realised as one same adapted form baŋ́̃ ́both for ‘character, nature’ and for ‘boundary, limit, demarcation’. In the adaptations of the Akan source H in CV syllables in Dangme, as was seen in (87), the source H is maintained in the adapted forms. I illustrate this with the examples in (88). University of Ghana http://ugspace.ug.edu.gh 216 (88). Akan Dangme Gloss i. fɔ ́ fɔ ́ guilt ii. èsú sú character, behaviour, nature iii. ɛb̀ɛ ́ àbɛ ́ proverb iv. ɔf̀a,́̃ àfa ́̃ àfa ́̃ half/soothsaying It can be observed from the examples in (88) above that the Akan source CV nominal stem with H is adapted as H in Dangme. A symmetry pattern of realising a source H as H has been observed in Ga, which also almost invariably maintains the source tone in the CV stems in the adapted forms. 6.2.1.2. Adaptations of Akan source HH(H) The adaptations of the Akan source HH, unlike in the preceding cases, are different in that the source is adapted to three different tonal melodies, namely HH, LH, and HL in CVCV, CVN, and CVV nominal stems. I provide examples for further illustrations in (89). (89). Akan Ga Gloss i. æ̀pítí àkp͡ ítí a cake, made of corn ii. àpáʨɪ ́ àk͡pákí a large calabash iii. æ̀ʥáń àgaì̃ ́̃ arrow University of Ghana http://ugspace.ug.edu.gh 217 iv. æ̀kŕó/èkŕó àkrò́/àklò́ scar, sore; ulcer v. òkᶣíá òkʷàá farming, agriculture vi. àhʊ́rá àhórà a very serious insult, taunt vii. m̀mɔb́ɔ ́ mɔb́ɔ ̀ pity, sorrow From (89), it can be observed that the source HH tone pattern is adapted as either LH as in (89iii) – (89v), where the pitch level of the stem-initial syllable is lowered or with HL, where the stem-final H syllable is realised as L as seen in (89vi) – (89vii) in Ga. In all these three cases of the realisations, the structure of the source stem can be the disyllabic C(V)CV, CVN, or CVV stems. From the existing literature (cf. Wentum 1997, Dakubu 2002, among others) the HH pattern in nominal stems is attested in Ga. Therefore, to realise a source HH as LH or HL could not be as a result of restrictions imposed by the native tonology. Again, source HH should be perceptually distinct enough for it to be misperceived as LH or as HL. It is worth pointing out the source words are nouns and not nominals (i.e. derived nouns). Therefore, their tonal alterations in Ga cannot be attributed to the result of any juncture, such as morpheme or word boundary, phenomenon. The same observation holds for the majority of the data examined in this chapter. Dangme speakers adapt the Akan source HH tone pattern of CVCV, CVV, and CVN nominal stems in a very similar pattern as was observed in Ga. See the data in (90). University of Ghana http://ugspace.ug.edu.gh 218 (90). Akan Dangme Gloss i. ɔẁáwá wáwá a large tree ii. pápá k͡pák͡pá goodness, hospitality, correct thing iii. m̀mɔb́ɔ ́ mɔb́ɔ,̀ mɔḿɔ ̀ pity, sorrow iv. àtɪá̃ ́̃ àtiá̃ ,̀̃ àʧiá̃ ̀̃ cashew nut tree; cashew nut v. æ̀hṹḿ àhṹmì storm, gale vi. æ̀ʥań́̃ ɡaí̃ ̀̃ bow From the examples in (90), we observe that Dangme speakers can adapt the Akan source HH as HH as seen in (90i) – (90ii), but in addition to maintaining the source HH pattern, the same source pattern can be adapted as the HL tone melody as seen in (90iii) – (90vi). That is, the pitch of the final syllable is lowered to that of L in the adapted forms just as was also observed in Ga, even though the two target sister languages permit the HH pattern in their tonology (cf. Wentum 1997). It is worth pointing out that (90v) is one of the variants of the realisations of the source word in Dangme. I provide pictorial evidence to exemplify the discussion in (89) and (90) above in a form of spectrogram showing the pitches of the Akan source HH stem word in Figure 6. I show in Figures 7 & 8, spectrograms showing a different realisation of this source HH in Ga and in Dangme. University of Ghana http://ugspace.ug.edu.gh 219 Figure 6: A spectrogram showing the pitch levels of m̀mɔb́ɔ ́‘pity, sorrow’ in Akan From Figure 6, the first H is realised at the pitch level of 111.37Hz almost the same as the second H, which is said at 118.45Hz. The nominal prefix is said at relatively lower pitch level of 88.91Hz. I compare this spectrogram with one each by Ga speaker and Dangme speaker for the same word adapted in Ga and in Dangme in Figures 7 & 8 below. Figure 7: A spectrogram showing the pitch levels of mɔb́ɔ ̀ ‘pity, sorrow’ in Ga University of Ghana http://ugspace.ug.edu.gh 220 From the spectrogram in Figure 7, the first syllable (i.e. H) is realised at the pitch level of 147.68Hz while the last syllable is relatively said at a significantly lower pitch level of 106.94Hz. It is evident that the Akan source HH pattern is realised as HL in the Ga pitch spectrum. In Figure 8, I present a spectral analysis of a similar realisation of the same source word in Dangme. Figure 8: A spectrogram showing the pitch levels of mɔb́ɔ ̀ ‘pity, sorrow’ in Dangme From Figure 8, in a very similar fashion to the spectrogram in Figure 7, contrast in pitch levels of the stem-initial and stem-final syllables is clearly observable in the spectrogram for the realisation of Akan source word for ‘pity, sorrow’ in Dangme. While the stem-initial syllable is said at the pitch of 125.49Hz, relatively higher than that of the stem-final syllable is at 96.84Hz. From the Dangme pitch University of Ghana http://ugspace.ug.edu.gh 221 spectrum, the stem-initial syllable can be considered as a H toned-syllable and the stem-final syllable as L. The same Akan source HH(H) can additionally be adapted as a LL tone melody in Ga, which is expatiated in the examples in (91). (91). Akan Ga Gloss i. ɔm̀aŋ́̃ ́ maŋ̀̃ ̀ town, country, state ii. ɔk̀ɔ́ꜝ rɪ ́ òkɔl̀è eagle iii. òkṹꜝna ́̃ òkul̃̀ a ̀̃ widowhood iv. æ̀bíꜝsá àbìsà fortune, destiny, lot, situation in life v. bǽꜝɡúá bɛɡ̀ʷà assembly, council of elders The examples in (91) show the fourth melody with which an Akan source HH can be adapted. These adaptations occur in syllable structures such as CVN and CVCV in the adapted forms. The examples in (91) present a challenge for perception; a source HH being perceived as LL by Ga speakers is quite baffling when these source words have the same structure as those discussed earlier in this sub- subsection, which were realised differently. We, therefore, should have expected the source HH to be faithfully adapted as HH in Ga, and not as the contrastive LL melody. From the examples in this sub-subsection, it will be difficult, if not untenable, to attribute these instances of unfaithful adaptation of source tone melody to a requirement imposed by the tonology of the target language. University of Ghana http://ugspace.ug.edu.gh 222 Similarly, an input HH is distinctive enough to be misperceived as LL in a target language especially when the source tone melody is present in the target language. I, therefore, recommend further investigation into the phenomenon to establish what might motivate the unfaithful adaptation of the tone melodies of those target words. 6.2.1.3. Adaptations of Akan source LH Just as was observed in the adaptations of the source H or HH-toned stems, the Akan source LH melody can be adapted as HH melody aside from the LH levels. This is exemplified in (92) and (93) below. (92). Akan Ga Gloss i. àfɔr̀ɪ ́ àfɔl̀é sacrifice, offering ii. ɱ̀fʊ̀á ɱ̀fóá pimple iii. àjɪr̀á àjálá the supernatural power to vanish iv. àpàkaŋ́̃ ́ àk͡pàkaì̃ ́̃ palanquin, travelling basket From the examples in (92ii) – (92iii), the source LH pattern of the nominal stem is realised as HH in the adapted forms in addition to the expected faithful preservation of LH in (92i). Our expectation of the source melody to be maintained in the adapted forms stems from the fact the the LH pattern of nouns is well attested in Ga just as it is in Akan. The syllable structures of the source stems, University of Ghana http://ugspace.ug.edu.gh 223 that is, CVV and CVCV in (92iii) and (92iv) respectively, do not play any role in determining the pattern of tone in the adapted forms just as it has not so far in the adaptation process. Finally, the source LHH pattern in (92iv) does not follow the regular pattern of simply raising the pitch level of the stem-initial syllable as in (92) to realise HHH, but instead the pitch level of the first H rather drops to L as the stem is realised as LLH. The Akan source LH can also be realised differently aside from this melody being maintained in the adapted forms by Dangme speakers. Among the different realisations of the source LH melody are HH, MH, and LL. I exemplify these different patterns of realisations in (93). (93). Akan Dangme Gloss i. pæ̀tú pàtú owl ii. òhià̃ ́̃ òhià̃ ́̃ poverty iii. ɔp̀àpʊ́ k͡pāk͡pó male goat, billy goat iv. æ̀tᶣìá tóá attack v. æ̀bùda ́̃ àbùda ̀̃ foolish man; a hen-pecked man It can be observed from the examples in (93) that, in addition to maintaining the source LH in Dangme, the pitch level of the source stem-initial L can rise to H to realise HH in the adapted forms, as seen in (93iii), or the pitch of the same source stem-initial L can rise to the M level to realise MH. Alternatively, the pitch level University of Ghana http://ugspace.ug.edu.gh 224 of the stem-final H-toned syllable can rather drop to L to realise a LL melody in the adapted forms in Dangme, as can be seen in (93v). One would have been expected that, for the (93v), the pitch level of the stem-final H could drop to M from H. However, it should be noted that as noted earlier in this section, the output of that realisation, which is LM, is generally prohibited in Dangme. Therefore, it can be concluded that a source LH melody can be realised as LH, MH, HH, and LL, but not LM. In the examples in (94), we consider the adaptations of an extension of the basic LH pattern in LLH(H). Again, we observe different adaptation strategies of the source tone melody including the raising of the pitch levels of the first, the second, and also the third or final syllables respectively of the nominal stems in the adapted forms. (94). Akan Ga Gloss i. æ̀dìbaŋ̀̃ ́ àdìbaŋ̀̃ ́ food, heavy or staple food ii. àkàsàá àkásàá/àkáʃàá cymbal iii. tɛk̀ɹɛ̀m̀a ́̃ tɛḱrɛ̀m̀á tongue iv. àpààwá àpàáwá a flat round wooden tray v. kaŋ̀̃ k̀aǹ̃ ɪ ́̃ kàŋk̀áŋ ́ civet cat vi. æ̀dìæ̀ʨírí àdíáʧírí store-keeper, steward vii. ǹtù̃ŋk̀ṹḿ tuŋ̃̀ ɡ̀ũ̀ḿ/tùŋ̃ ɡ̀ũ̀ṹ new palm wine University of Ghana http://ugspace.ug.edu.gh 225 In the examples in (94), it is seen that a source stem-initial L can be realised as H tone in the adapted forms as seen in (94ii) – (94iii). Similarly, a stem L in the penultimate position can be realised as H in the adapted forms as in (94iv) – (94v). The pitch level of an H-toned syllable of the stem, on the other hand, can be lowered at the penultimate position in the adapted forms, as seen in (94vii). It is worth noting that the same source LLHH can also be realised in two different ways in Ga, namely either as HHHH, as seen in (94vi) or as LLLH, as in (94vii). So far no explanations have been offered on either phonological or perceptual grounds for the variations we observed in the adaptation of source tone melodies. Perhaps the explanation for this and other similar unfaithful adaptation might be that although speakers correctly perceive the source tone melodies, yet they adapt them to the variant melodies present in their language taken into an account the source syllable structure. Aside from the basic LH melody, other extensions such as LHH can also be adapted by Dangme speakers. I provide some examples for this pattern in (95) below. (95). Akan Dangme Gloss i. ŋk̀ũ̀níḿ kù̃niḿ́̃ victory, success ii. æ̀tùpań́̃ àtùk̃ ͡pá bottle, esp. beer bottle iii. sùbań́̃ súbaí̃ ́̃ character University of Ghana http://ugspace.ug.edu.gh 226 From the examples in (95), it can be observed that the source LHH can be realised with different patterns in Dangme. Though there are a limited number of such cases of tonal changes, yet the source LHH can be additionally realised as simple LH as in (95ii), which can as well be realised as LHH supposing the source stem- final were not deleted as a repair strategy. Alternatively, it can be realised through the raising of the pitch level of the stem-initial syllable from L to H in the realisation of HHH, as seen in (95iii). Other possible realisations such as MHH and LLL are not attested in the data examined. Another extension of the Akan source LH melody with which Dangme speakers can adapt the source tonology is the LLH pattern. Here the adaptation patterns are similar to the source LH. I provide examples to illustrate this adaptation pattern in (96). (96). Akan Dangme Gloss i. àbɪr̀ɪẁá àblè̀wá old lady ii. kʷàsɪà́ kùàsìá a fool, stupid person iii. pàtæ̀kú pɛt́ɛḱú hyena iv. jààwá àjáwá a large metal pan, tray or box v. bʊɹ̀ ɔf̀ʊ́ blɔ̀f̄ónò European, foreign From the examples in (96), a major tone melody with which the Akan source LLH can be adapted is HHH as can be seen in (96iii) – (96iv), where the pitch level of University of Ghana http://ugspace.ug.edu.gh 227 the LL is raised to that of HH to realise HH(H) in the adapted forms. Another strategy is to raise the pitch level of the second syllable from L to M to realise a melody of LMH as seen in (96v). Again, LLL and LLM, where the final H is lowered to L or M are not attested in the data examined for this kind of adaptation, although the source’s first two syllables, LL can rise to HH in the adapted forms in Dangme as in (96iii). 6.2.1.4. Adaptations of Akan source (L/H)HLH So far we have looked at cases of the realisations of either H or those of two distinct level tones in adjacency in the Akan source. This sub-subsection looks at the realisations of the Akan source stems that terminate in the HLH melody in Dangme. Just as have been consistently observed throughout this chapter, a source tone pattern, with the exception of the tones in some CV nominal stems, can be realised in different ways in the adapted forms. In this sub-subsection, we observe similar different realisations of same source melody, which I exemplify in (97). (97). Akan Dangme Gloss i. áḿpaǹ̃ ɪ ́̃ ápāní edible species of bat, esp. rosette bat ii. má̃ḿpam̀́̃ maṕ̃ ām̄ monitor lizard iii. æ ̀dúòkú àdùòkú a nickname for the rat iv. ɔs̀ɹà́àfʊ́ áslàáfó soldier University of Ghana http://ugspace.ug.edu.gh 228 From the examples in (97) above, it is observable that the source terminating HLH melody, unlike in most of the discussions thus far, is never maintained in the adapted forms in Dangme; it is consistently realised differently. Similarly, the tone of the ante-penultimate syllable can be maintained as in (97i) – (97ii), or it can be lowered to L, as in (97iii) – (97iv). The intervening L can be raised to M, as in (97i) – (97ii), or even to H, as in (97iv). Again, the LH of both the penultimate and the final syllables can be realised as MM, as seen in (97ii). Similarly, the ante- penultimate H syllable can be lowered so that it is realised as L, as seen in (97iii) – (97iv). The final syllable H is usually maintained, though it can also be lowered to a M, as seen in (97ii). However, a drop in the pitch level to that of L will result in ML sequence in (97i) – (97ii), which we already know is not generally permitted in Dangme. In a similar manner, when the same syllable drops in pitch to the level of M, it will yield an LM sequence in (97iii), which is again not permitted in all dialects of Dangme. 6.2.1.5. Adaptations of Akan source HL In the examples in (98), I show that the Akan source HL melody can be realised differently in Ga, namely as LH and LL in addition to maintaining the source pattern. (98). Akan Ga Gloss i. ɔd̀áẁ dáà mouth, mouth cavity ii. àdárɪ ̀ àdídà, àdédà billhook, pruning shears University of Ghana http://ugspace.ug.edu.gh 229 iii. òbóǹtó bòǹtò a type of fishing boat, trawler iv. bʊ́àpɪá̀ bòápìà a kind of monkey From the examples in (98), it is observable that the source HL melody can also be realised as LL, thereby causing the lowering of the pitch level of the H (in the ante-penultimate syllable) in (98iii), and also lowering of the pitch level of the H (in the penultimate syllable) as in (98iv). In the case of first two syllable of example (98iv), on the other hand, the source HL is realised with a contrastive LH tone pattern in the adapted form in Ga. We have already seen the source HH being adapted as HL in (89), which means this pattern is attested in Ga. Therefore, adapting a source HL as other tonal melodies cannot again be as a result of restriction on the pattern by the native tonology. 6.2.1.6. Adaptations of tones in Akan source compound stems In the following examples the adaptations of the Akan source nominal stems that are underlyingly compounds or bimorphemic are discussed. The second morpheme (i.e. morpheme2) of the compound stem seems to be pre-specified for [+Low] tone whenever it occurs in CVV stems. (99). Akan Stem1+Stem2 Ga Gloss i. fáꜝdáŋ́ (ɛf̀á+dáŋ)́ àfáꜝdáŋ ́ swish building ii. sùbaŋ́̃ ́ (èsú+báŋ)́ súbaŋ́̃ ́ character University of Ghana http://ugspace.ug.edu.gh 230 iii. àkǽ̃ńsí (kǽ̃ń+sí) àkaŋ́̃ ʃ́iì̃ ̀̃ competition iv. àdɪ ́ꜝbɔ ́ (àdɪ+́bɔ)́ àdébɔɔ̀ ̀ nature, creation v. ɡʷáꜝbɔ ́ (èɡʷá+bɔ)́ ɡʷábɔɔ̀ ̀ assembly, conference vi. àɥɪr̀ɛt̀ɔ ́ (ɔɥ̀ɪŕɛ+̀tɔ)́ òwélètɔɔ̀ ̀ avenging vii. bɪḿ́̃ bú (bɪḿ́̃ +bú) béḿbùù acquittal, declared innocent The adaptation of the compound stems exhibits an interesting pattern. That is, whenever the second stem of the source word is CV, the final V is lengthened and the derived CVV structure automatically receives an LL tone pattern as against the original H in the source. This pattern of tone realisation can be observed in (99iii) – (99vii). It could not be verified in this study whether or not Ga speakers are aware that these nominals which are targeted for this process are underlyingly compounds. From the discussion above, there emerges a pattern that lengthens the final V of morpheme2, and thus alters the source H to LL. Again, perceiving a source stem-final H as LL as in (99iii) – (99vii) might not be due to any known tone rule in Ga. I illustrate the realisation of examples (99ii) in the spectrogram below in Figure 9 and 10. Figure 9 is the spectrogram of the Akan source word sùbaŋ́̃ ́‘character’ and its realisation in Ga is shown in Figure 10. University of Ghana http://ugspace.ug.edu.gh 231 Figure 9: A spectrogram showing the pitch levels of sùbaŋ́̃ ́‘character’ in Akan In Figure 9 above, the initial syllable is realised at the pitch level of 88.72Hz, while the second and the last syllables are said at 116.12Hz and 119.95.71Hz respectively. The pitches of the initial syllable, on the one hand, and those of the last two syllables, on the other hand, fall within the averages for L and H respectively in the Akan pitch spectrum. In Figure 10, I present another spectrogram showing the realisation of the same Akan source word in Ga. It will be seen that the initial syllable in Ga is realised rather with H. University of Ghana http://ugspace.ug.edu.gh 232 Figure 10: A spectrogram showing the pitch levels of súbaŋ́̃ ́‘character’ in Ga From the spectrogram in Figure 10 above, it can be seen that the LHH melody of the Akan source word for ‘character’ is realised as HHH for all the three syllables in Ga. That is, the source stem-initial syllable, was realised as L in Akan, is said at the pitch level of 139.61Hz, which falls within the pitch range for H in Ga. This pitch is even slightly higher than the two subsequent syllables which are said at 131.41Hz and 125.48Hz respectively. As has been done for Ga, I briefly look at the adaptations of the Akan source words which have compound stems at the underlying level of representation in Dangme. In Ga, it was observed that, for most of such stems, the stem-final vowel is lengthened and this tone changes from H in the source to L in the adapted forms. A similar pattern is also observed in Dangme. However, unlike in Ga, the lengthening of the final vowel does not apply in Dangme. Again, the source stem-final H can be realised as M in Dangme. I provide examples to illustrate this in (100). University of Ghana http://ugspace.ug.edu.gh 233 (100). Akan Stem1+Stem2 Dangme Gloss i. àmaǹ̃ ɪh̀̃ ṹ (àmaǹ̃ ɪ+́̃ hṹ) àmaǹ̃ èhlṹ, àmaǹ̃ èhṹ affliction, suffering ii. ɡʷúáḿ (ɡʷúá+mṹ) ɡʷúám̄ a gathering iii. àkʊ́kɔb́ɛ́ꜝ sa ́̃ (àkʊ́kɔ+́bɛśa)́̃ àkɔḱɔb́ɛśa ̀̃ a climbing/straggling thorny shrub iv. àɥɪr̀ɛh̀ʊ́ (ɔɥ̀ɪrɛ+̀hʊ)́ àwɛr̀ɛh̀ò, àwìlɛh̀ò grief, sorrow, sadness v. àmaǹ̃ ɪb̀̃ ɔ ́ (àmaǹ̃ ɪ+́̃ bɔ)́ āman̄̃ iɛ́ b́ɔ ̀ mission, report, news From (100), it can be seen that although the final syllable of the compound nominal stem is H in the source language, it can also be systematically realised as non-high, i.e. either M, as in (100ii) or as L, as in (100iii) – (100v) in the adapted forms in Dangme. This adaptation follows a similar pattern to the one observed about Ga. In the following spectrograms, I show how Akan source compound stem-final H, as shown in Figure 11, is realised as L in Dangme, as seen in Figure 12. University of Ghana http://ugspace.ug.edu.gh 234 Figure 11: A spectrogram showing the pitches in àmaǹ̃ ɪb̀̃ ɔ ́‘mission, report, news’ in Akan From the spectrogram in Figure 11, the pitch of the word-initial L syllable is 93.34Hz, while the second and the penultimate L syllables are realised at 87.30Hz and 88.72Hz respectively. The final syllable, on the other hand, is realised at a relatively higher pitch of 116Hz. The significant variation in pitch between non- final L syllables and the final H syllable is observable from the spectrogram, which results in LLH melody in the stem. In Figure 12, I show that this melody can change to LHL in adapted forms in Dangme. University of Ghana http://ugspace.ug.edu.gh 235 Figure 12: A spectrogram showing the pitches in āman̄̃ íɛb́ɔ ̀‘mission, report, news’ in Dangme From the spectrogram in Figure 12, the first syllable, which is L according to Dangme pitch spectrum, is realised at the pitch of 107.05Hz virtually the same as the second syllable, which is at 106.27Hz. The ante-penultimate and the penultimate H syllable are both realised at the pitches of 124.61Hz and 123.08Hz respectively. The stem-final syllable, which is L, on the other hand, is realised at the pitch of 99.74Hz. With our focus on the penultimate and the stem-final syllables, it can be seen that the Akan source tonal pattern of -LH is realised as opposite melody of -HL in Dangme. 6.3. Adaptations of Akan source downstepped H in Ga and Dangme Although the downstep rule is active in Ga, sometimes the downstepped H in some Akan source nominal stems are adapted as non-downstepped H tone. Similarly, the Akan source downstepped H can also be adapted as non-downstepped H in University of Ghana http://ugspace.ug.edu.gh 236 Dangme, a language in which lexical downstep is rare (cf. Dakubu 1987). Unlike Ga, in Dangme, which is a three-level tone language, the downstep phenomenon is caused by either the L or the M tone, whereas it is only L tone that triggers the phenomenon in Akan, the source language. The target for the downstep rule is usually the H-toned syllable, though both the H and the M tones can be targeted in the downdrift process in Dangme (cf. Dakubu 1987: 21). In this subsection, I discuss how the Akan source downstepped H syllable is realised in both Ga and Dangme in the adaptation process. It will be observed in the following discussions that the downstepped H can sometimes be realised as L, M, or H. Again, it will be seen that some Akan source H-toned syllables can even be realised as a downstepped H in Ga. 6.3.1. Adaptation of Akan source ꜝH as ꜝH In the following examples, I show that the downstep rule in Akan is also active in Ga even in its loanword phonology. It is, therefore, expected that an Akan source downstepped H syllable is maintained as such in Ga in the adaptation process. (101). Akan Ga Gloss i. ɔǹʊ̀kʷáꜝfʊ́ ànɔk̀̃ ʷaꜝ́fó a truthful person ii. àpásáꜝfʊ́ àpásáꜝfó liar, gossip, idle chatterer iii. áḿꜝpé áḿꜝpé a rhythmic jumping game iv. ŋk̀áꜝtɪɛ́ ́ ŋk̀áꜝtíɛ,́ ŋk̀ɛ́ꜝ tíɛ ́ groundnut, peanut University of Ghana http://ugspace.ug.edu.gh 237 In the examples in (101) above, the tonal melodies and the pitch levels of the Akan source syllables are faithfully maintained in the adapted forms in Ga, as seen in the spectrogram in Figure 14. This faithful adaptation is expected in that both the source and the target languages operate the downstep rule, which is the focus of the present discussion. This faithful adaptation is accounted for in terms of perception, that is, Ga speakers correctly perceive the pitch levels of each of the source syllables, hence they map them onto the syllables of the adapted forms. In Figure 13 below, I show a spectrogram of how the downstepped H is realised in the Akan source word for ŋk̀áꜝtɪɛ́,́ and show how the downstep is maintained in the adapted form in Ga in Figure 14. Figure 13: A spectrogram showing downstep in ŋk̀áꜝtɪɛ́ ́ ‘groundnut, peanut’ in Akan University of Ghana http://ugspace.ug.edu.gh 238 In the spectrogram in Figure 13, while the stem-initial H is realised at the pitch level of 121.56Hz, the ante-penultimate and the final downstepped H syllables are said at virtually the same pitch levels of 110.38Hz and 106.46Hz respectively. In the following spectrogram in Figure 14, I show how Ga speakers maintained the source downstep in adapted forms. Figure 14: A spectrogram showing downstep in ŋk̀áꜝtíɛ ́‘groundnut, peanut’ in Ga From the spectrogram in Figure 14, the stem-initial H syllable is said at the pitch of 143.03Hz as compared to the following penultimate and final Hs, which are realised at 123.78Hz and 113.36Hz respectively. This pattern is similar to what we observed in Figure 13 where the stem-initial H syllable was significant higher in pitch than the subsequent H syllables. Just as was observed in Ga, the Akan source downstepped H can be maintained in the adapted forms in Dangme. I provide illustrations of this adaptation in the examples in (102). University of Ghana http://ugspace.ug.edu.gh 239 (102). Akan Dangme Gloss i. ɔk̀ɔ́ꜝ tɔ ́ òkɔ́ꜝ tɔ ́ a small greenish river crab ii. ɔʨ̀ɪà̃ ̃́ꜝ mɪ ́̃ òʧìāꜝmɛ ́ spokesperson, linguist iii. ŋk̀áꜝtɪɛ́ ́ káꜝtíɛ ́ groundnut From the examples in (102), it is seen that the pitch levels of the stem-final syllables in the source words are maintained in the adapted forms though the target language does not seem to operate the downstep rule. On the exceptional case above, it has been explained in the existing literature that occurrence of the downstepped H seems to be limited to loanwords (Dakubu 1987: 24). Is it because of the fact that it is a language with seven dialects? Though there is a native Dangme word for the ‘blue plantain eater (Corythaeola cristata)’ which is oʧ̀ iŕ ã ́ꜝ ko,́ which has a downstepped H in it. The complete explanations for this exception might not reside in the phonology of Dangme, but perhaps in perception. As we will see in the next cases of this exception, the same source downstepped H can be adapted as H. So the question then is, how does the phonology again explain the adaptation of the same pitch differently in the same language? 6.3.2. Adaptation of Akan source ꜝH as L It is not always the case that the pitch levels of the source syllables are maintained in the adapted forms in Ga. Unlike the cases of the faithful adaptations in (103) University of Ghana http://ugspace.ug.edu.gh 240 above, in the examples in (103), it can be seen that the Akan source downstepped H is rather realised as L. (103). Akan Ga Gloss i. ɔk̀ʊ̀bɔ́ꜝ fʊ́ kòbɔf́ò vagrant ii. òbúbúáꜝfʊ́ òbúbúáfò cripple, paralysed person iii. mań́̃ ꜝsʊ ̃́ mań́̃ sɔ ̀̃ quarrel, strife, litigation iv. ɔd̀áꜝsɪ ́ òdásè testimony v. ɔd̀ɪ ́ꜝ ɕɪ ́ òdéhè, òdéè noble, free born person vi. ɔd̀ɛɲ́́ꜝʨɛḿ́ òdɛń́ʧɛ ̀ a small type of crocodile From the examples in (103) above, we see that the Akan source downstepped H syllable, being the last syllable, is realised as L in Ga. In other words, Ga speakers produce the source downstepped H with a relatively lower pitch, hence its realisation as L in the adapted forms. It is generally expected that the pitch level with which a downstepped H syllable is expected to fall between the levels of a H pitch and a L pitch for languages that contrast between these two levels. Therefore, for Ga, which falls within this category of tone languages, it can be expected that if Ga speakers do not correctly realise the level of pitch for a downstepped H syllable, they are likely to produce it as L-toned syllable. In Figure 15 below, I show the spectrogram of the word for ‘testimony’ in Akan and also show the downstepped H in the source word is realised in Ga in Figure 16. University of Ghana http://ugspace.ug.edu.gh 241 Figure 15: A spectrogram showing the pitch level of ɔd̀áꜝsɪ ́‘testimony’ in Akan From Figure 15, the stem-initial H syllable is realised at the pitch of 124.46Hz as against the stem-final H syllable, which is said at a relatively lower pitch of 112.77Hz. The pitch of the stem-final syllable compares with the word-initial (nominal prefix) L syllable which is realised at 90.32Hz. This means the stem-final syllable is realised as downstepped H relative to the preceding H syllable. In Figure 16, we observe that the source stem-final H syllable is rather realised with a lower pitch in Ga. University of Ghana http://ugspace.ug.edu.gh 242 Figure 16: A spectrogram showing the pitch level of òdásè ‘testimony’ in Ga In Figure 16, the stem-initial H syllable is realised at the pitch of 151.77Hz, while the stem-final syllable, also H, is realised at 130.92Hz almost the same as the word-initial L syllable with the pitch of 128.76Hz. This clearly shows that unlike in Akan, the stem-final syllable is rather realised as L. It is not only in Ga that Akan source downstepped H can be realised as L, but also in Dangme as well. In the following examples I consider cases where the pitch of the source downstepped H is reduced to the level of L. Though there are fewer such examples in the data, I exemplify those cases in (104). (104). Akan Dangme Gloss i. àdɪ ́ꜝbɔ ́ àdébɔ ̀ creation, nature ii. æh̀ ṹḿ àhṹmì storm, gale iii. àkʊ́kɔb́ɛ́ꜝ sa ́̃ àkɔḱɔb́ɛśa ̀̃ a climbing or straggling thorny shrub University of Ghana http://ugspace.ug.edu.gh 243 In the examples in (104), it can be seen that the source stem-final, which was realised as downstepped H, is realised as L by Ga speakers. The adaptation is quite unexpected since the pitch level of the source downstepped H (at 126.22Hz, as seen in the spectrogram in Figure 17) should be high too to be realised as L, especially when the pitch range for L in Dangme falls below even 120Hz. I show in the following spectrograms how these pitches are realised in both Akan and Dangme. Figure 17: A spectrogram showing the pitch level of àdɪ ́ꜝbɔ ́ ‘creation, nature’ in Akan In the spectrogram in 17, the stem-initial H syllable is realised at the pitch of 131.29Hz, while the following downstepped H syllable is said at a slightly lower pitch of 126.22Hz. The word-initial syllable, however, is realised at a significantly lower pitch of 97.22Hz making it a L-toned syllable. I compare this spectrogram University of Ghana http://ugspace.ug.edu.gh 244 with another one showing how the source pitches are realised in the same word adapted in Dangme in Figure 18. Figure 18: A spectrogram showing the pitch level of ādébɔ ̀ ‘creation, nature’ in Dangme In the spectrogram in Figure 18, the stem-initial H is realised at the pitch of 124.67Hz while the stem-final syllable, which is L, is said at the pitch of 98.50Hz. The gap between the H and the L is wide enough for the stem-final syllable to be regarded as either downstepped H, as was the case in Akan, or even M, since it falls outside the average pitch range for non-L tones. It is interesting to note that the word-initial syllable, which is M, is even realised at the pitch of 112.00Hz. University of Ghana http://ugspace.ug.edu.gh 245 6.3.3. Akan source ꜝH as M This sub-subsection discusses how ꜝH is realised as M in some adapted forms. In the examples in (105), it will be seen that the same Akan source downstepped H is not always maintained, but can also be realised as M in adapted forms in Dangme. (105). Akan Dangme Gloss i. ɔd̀ɪ ́ꜝ ɕɪ ́ ōdéhē noble, free born person ii. òhià̃ ̃́ꜝ fʊ́ òhià̃ f́̃ ō poor person iii. ɔ̀dáꜝsé ōdásē testimony, witness iv. áꜝkó ákō a bird, parrot v. ɔʨ̀ɪà̃ ̃́ꜝ mɪ ́̃ òʧaà̃ ḿ̃ i ̄̃ spokesperson, linguist From (105), we observe that the Akan source downstepped H is adapted as M in Dangme. This adaptation pattern is expected in a three-level tone language. Again, from the examples in (105) the adjacent source H-toned syllable maintains its pitch level in the adapted forms. In these examples, Dangme speakers do not seem to be faithful to the source tonology, but instead they impose Dangme native tonal pattern on the adapted forms. The pitch level of the stem-final downstepped H is further reduced to M, a tone level which is present in Dangme instead of faithfully maintaining the highly marked ꜝH. It can, therefore, be explained that perhaps Dangme speakers might have perceived the downstepped H rather as more M, University of Ghana http://ugspace.ug.edu.gh 246 which is close to a downstepped H in the adaptation process. It is worth pointing out that a downstepped H being perceived as M has also been observed in Bimoba, a Gur language spoken in Ghana by Snider (1999). In Figure 19, I provide a spectrogram to show that the source downstepped H in the word for ‘testimony, witness’, as shown in Figure 15 above, is realised as M in Dangme. Figure 19: A spectrogram showing the pitch level of ōdásē ‘testimony, witness’ in Dangme From the spectrogram in Figure 19, the stem-initial H syllable is said at the pitch of 126.39Hz while the following stem-final syllable, which is M, is realised at 103.07Hz. The pitch of the stem-final syllable is quite close to that of the word- initial L syllable, which is said at 112.07Hz. If the stem-final syllable were realised as a downstepped H as was shown in Figure 14 to be the case in Akan, we would have expected the pitch to be in the range of between 116Hz and 119Hz. University of Ghana http://ugspace.ug.edu.gh 247 6.3.4. Adaptation of Akan source ꜝH as H It is not in all cases that the pitch level of the Akan source downstepped H is reduced when adapted in Dangme. There are a few instances where source downstepped H is perceived with even a raised pitch level to H. I provide examples of these limited cases in the examples in (106). (106). Akan Dangme Gloss i. ŋk̀áꜝtɪ ́ ākáté groundnut ii. ɔp̀ɛ́ꜝ sɛ ́ òpɛs̀ɛ ́ brush-tailed porcupine; hedgehog iii. kɔt̀ɔ́ꜝ kɔ ́ kɔt̀ɔḱɔ ́ crested porcupine From the examples in (106), it is seen that the source downstepped H is realised rather as H with the pitch level of the adapted syllable higher than that of the source syllable. Unlike in the preceding cases in (105), adapting the source downstepped H as a H syllable is not unexpected since the two are phonetically close enough in terms of pitch. In Figure 20 below, I provide a spectrogram for the realisation of the adapted word ākáté ‘groundnut’ showing that the source stem- final downstepped H is realised as H in Dangme. University of Ghana http://ugspace.ug.edu.gh 248 Figure 20: A spectrogram showing the pitch levels of ākáté ‘groundnut, peanut’ in Dangme The stem-initial H in the spectrogram in Figure 20 is said at the pitch of 139.06Hz, almost at the same level as the following H, which is the stem-final, is realised at 135.41Hz, a value that obviously falls outside of the range for downstepped H. The word-initial M syllable, on the other hand, is realised at the pitch of 112.44Hz. These figures, compared with those we saw in Akan in Figure 13, lend credence to the claim that I have already made that the Akan source downstepped H syllable can also be realised as rather H in Dangme. 6.3.5. Adaptation of Akan source H as ꜝH Sometimes the Akan source H-toned syllable can be adapted as a downstepped H syllable in Ga. To do this, Ga speakers employ strategies to create the context for the realisation of lexical downstep. In the examples in (103), I discuss how the University of Ghana http://ugspace.ug.edu.gh 249 speakers create the downstep environment in order to realise the source H as downstepped H. (107). Akan Ga Gloss i. àpààfʊ́ òpàáꜝfó day-labourer ii. ɔk̀úkúbàŋ ́ òkúkúꜝbá name of a small animal iii. àkɔt̀ʊá̀ àkóꜝtówá a small barrel, keg From the examples in (107), it is clear that the source non-downstepped H is realised in Ga as downstepped H. The two main strategies Ga speakers employ in achieving this are as follows: first, is to raise the pitch of a penultimate L, which precedes a stem-final H to the level of H as seen in (107i). Second, is to raise the pitch level of the targeted syllable from L to ꜝH when it is preceded by H as seen in (107ii). The last strategy, which involves two syllables, is to raise the pitch levels of both the target syllable and the preceding syllable from L to ꜝH and H respectively as in (107iii). 6.4. Summary of the chapter This chapter has discussed the adaptations of Akan source tone for the nominal class forms in both Ga and Dangme. The discussions have focused on the adaptations of some H, HH, LH, LLH, and LHH tonal melodies of Akan source words in the two target languages where it was concluded that both Ga and University of Ghana http://ugspace.ug.edu.gh 250 Dangme speakers usually preserve the Akan source tonal melodies when adapting them. That is, the source melodies are adapted in the two language target languages. However, additionally, all the source melodies can be realised with different melodies in both target languages. This lack of consistency presents a challenge for the formalisation of the adaptation patterns within the OT, our working theoretical framework. The adaptations of the Akan source compound stems were also discussed where it was concluded that both Ga and Dangme realise the Akan source compound stem-final H-toned CV syllable with a L tone. However, while Ga speakers lengthen the final vowel of the stem-final CV syllable, Dangme speakers do not lengthen this vowel. The downstep phenomenon was also considered in the discussions where it was observed that while Ga either faithfully preserves the Akan source downstepped H or realises it as L, Dangme can adapt this same downstepped H as H, M, or L in addition to faithfully preserving the pitch level of the source syllable, though the downstep rule is not active in Dangme. Again, it was observed that Ga exclusively employs some strategies for realising even a source L-toned syllable as downstepped H through creating the necessary environment for the downstep rule to apply. So, for example, a source stem melody of LLH can be realised as HꜝHH in Ga in the adaptation of àkɔt̀ʊ̀wá‘a small barrel, keg’ as àkóꜝtówá. In Table 14 below, I summarise the commonalities and differences between Ga and Dangme in terms of how they realise Akan source tonal melodies. University of Ghana http://ugspace.ug.edu.gh 251 Table 14: Adaptations of Akan source tonal melodies in Ga and Dangme Akan source tonal melody Ga Dangme Monosyllabic: H H H Disyllabic: HH HH,HL,LL,LH HH,HL LH LH,HH LH,HH,LL HL HL,LL N/A Trisyllabic: LHH N/A LHH,HHH LLH LLH,HLH,LHH, LHH,LHL,MH,HHH HHH Compound stem-final H H, LL H, M, L Stem ending in HLH N/A HMH,HMM,LLH LHH Downstepped H ꜝH,L ꜝH,H,M,L Stem-final H Can be ꜝH N/A University of Ghana http://ugspace.ug.edu.gh 252 CHAPTER SEVEN SUMMARY, RECOMMENDATION, AND CONCLUSION 7.1. Introduction This thesis has discussed the adaptations of Akan (Kwa, Tano Central language) source words in two sister languages, namely Ga and Dangme (both Kwa, Ga- Dangme languages). The study has specifically considered the various strategies that both Ga and Dangme speakers employ in adapting segmental (both vowels and consonants) and prosodic properties, which focused on nasality, syllable structure and tone, in the two target languages. In this chapter, I summarise and draw conclusions to all the discussions made in this thesis. I also show the significance of this study as well as make recommendations based on the findings in this thesis for future studies. 7.2. Summary of preceding chapters In chapter one, I provided a general introduction to the entire thesis. In this chapter, I also provided some relevant information about the three languages that have been discussed in this thesis. In this same chapter, I mentioned the problem statement that informed and served as the basis for this research study as well asset out the main objectives for the study. Based on the objectives, I asked the research questions that have been addressed within the study. The chapter was concluded with the discussion of ethical considerations that guided this study. University of Ghana http://ugspace.ug.edu.gh 253 In chapter two, I reviewed relevant literature on the subject matter of the present thesis where it was concluded that there has been considerably less attention paid to the study of the loanword phenomenon both within and across language families in Ghana. The ongoing debate over the theoretical approach or model to the phonological study of loanword adaptation was also discussed in this chapter. Finally, I briefly discussed two phonological theories, namely DF and OT. It was briefly explained why OT was chosen for the formalisations of the discussions in the thesis. The phonological patterns both Ga and Dangme speakers employ in the adaptation of some vowels of Akan source, namely [ɪ], [ʊ], and [æ], have been discussed in chatper three. It was concluded that both [ɪ] and [ʊ] can be adapted to three vowels each that are present in Ga and Dangme. That is, the source [ɪ] can be adapted as [o], [i], or [ɛ]. For [æ], on the other hand, it is systematically adapted as [a] in Ga, while it can be additionally realised as [ɛ] in Dangme. These adaptation strategies, it was concluded, could be generally motivated by both perceptual and phonological factors. The pattern of adaptation where it was observed that in frequency count a source non-native vowel is most likely to be adapted as its perceptually closest counterpart in the target language supported the outcome of perception experiment carried out in this study. The other adaptation strategies were also attributed to phonological factors. Both Ga and Dangme speakers considered phonological factors such the height or the ATR value of the source vowels when adapting it. The conclusion, therefore, both phonological and University of Ghana http://ugspace.ug.edu.gh 254 perceptual factors are important in explaining the adaptations of Akan source non- native vowels in both Ga and Dangme. The phonological adaptations were formalised within the OT. I did a brief comparison of both Ga and Dangme in terms of how each adapts Akan source non-native vocalic segments. It was concluded that there are more resemblances than differences between the two target languages. In chapter four, I discussed the phonological patterns employed in the adaptations of some Akan source consonants, which are not present in both Ga and Dangme. The discussion mainly focused on the phonological patterns which speakers employed in realising the alveo-palatals, labial-palatals, and labial- palatalised consonants in Ga and Dangme. It was observed that both Ga and Dangme speakers correctly anticipate and perceive the Akan source non-native labial-palatality, hence, they systematically replace the feature with the labial feature, which is present in their phonology systems. Another issue that engaged our attention in this chapter was the adaptations of the Akan source illicit word- final labial-velar glide [w] in both Ga and Dangme. It was observed that two main strategies, namely vocalisation and deletion are employed in repairing such illicitness in the two target languages. With regard to the vocalisation, it was observed that source illicit word-final [w] consistently vocalises into [o] in both Ga and Dangme. In addition to the [o], the word-final [w] can also be realised as [u] in Dangme. It was also shown that when the deletion strategy is employed, it could result in compensatory lengthening in Ga, usually when the rest of the final word will beless than disyllable after the application of the deletion rule. OT was used to formalise the phonological accounts for the patterns of adaptations of the University of Ghana http://ugspace.ug.edu.gh 255 Akan source non-native and positionally ill-formed consonants in both Ga and Dangme. The chapter five has discussed the strategies Ga and Dangme speakers employ in adapting Akan source nasality and source syllable structures. The discussions of the adaptations of the nasal feature focused on how the speakers adapt the [+nasal] feature of Akan source vowels, in particular the three non-native vowels. It was concluded that with the exception of instances where the vowels which the source non-native vowels are adapted as can be nasalised, the source nasality is dropped in the adapted forms, which pattern is consistent with what generally obtains in the general phonology of the two target languages. It was also observed that there are instances where, although a source vowel can be nasalised in the target languages, it is not nasalised, yet the same becomes nasal in the adapted forms, and vice versa. With regard to the adaptations of the Akan syllabic nasals in both Ga and Dangme, it was shown that at the word-initial position, the syllabic nasal may or may not be preserved in the adapted forms in Ga, whereas they are completely dropped in Dangme. The initial syllabic nasals can also be realised as vowels, particularly as the low vowel, in both target languages. At the word-medial position, on the other hand, both languages do not usually permit such NN sequence, thereby reducing such sequence through deletion of the syllabic nasals in such contexts. These restrictions in sequences were formalised within the OT. For the adaptations of the Akan source word-final syllabic nasals, the two target languages differ to some extent. While some of the syllabic nasals, particularly [m] can sometimes be preserved in the adapted in both languages, University of Ghana http://ugspace.ug.edu.gh 256 other nasals, including [n], and [ŋ], are either deleted or are realised as a front high vowel. In Ga, there is also an additional process of compensatory lengthening that sometimes accompanies the deletion of those word-final syllabic nasals. Dangme speakers also exclusively apply epenthesis of a vowel, usually a front high, as a repair strategy for the final nasals. The chapter concluded with the discussion of the adaptations of the syllable structures of some of the adapted forms from Akan in Ga and Dangme. It was observed that between the two target languages, CV monosyllables and CVN or VV disyllables could be targeted for the truncation process. It was realised that the truncation might not be phonologically-motivated, but seems to, perhaps, only facilitate speech among the speakers of the two target languages. The chapter six has discussed the adaptations of Akan source tone for the nominal class forms in both Ga and Dangme. The discussion focused on the adaptations of some H, HH, LH, LLH, and LHH tonal melodies of Akan source words in the two target languages where it was concluded that, aside from the source H in CV stem which is virtually faithfully adapted I same pitch height, the tonal melodies of all the other nominal stems of different syllable sizes can be realised differently in both target languages. The adaptations of the Akan source compound stems were also discussed where it was concluded that both Ga and Dangme realise the Akan source compound stem-final H-toned CV syllable with a L tone. However, while the Ga speakers lengthen the final vowel of the stem-final CV syllable, the Dangme speakers, on the other hand, do not lengthen this vowel. The downstep University of Ghana http://ugspace.ug.edu.gh 257 phenomenon was also considered in the discussion. It was observed that while Ga either faithfully preserve the Akan source downstepped H or it is realised as L, Dangme, on the other hand, can adapt this same downstepped H as H, M, or L in addition to faithfully preserving the pitch level of the source syllable though the downstep rule is not active in Dangme. Again, it was observed that Ga exclusively employs some strategies for realising even a source L-toned syllable as downstepped H. This can be done by creating the necessary environment for the downstep rule to apply. So, for example, a source stem melody of LLH can be realised as HꜝHH in Ga in the adaptation of àkɔt̀ʊẁ á ‘a small barrel, keg’ as àkóꜝtówá. In chapter seven, I summarise all the discussions in each chapter. I also make some conclusions to the discussions made. The significance of the study and the recommendations made in this thesis are presented in this chapter. 7.3. Concluding remarks The findings from the analysis of the various phonological patterns observed in the adaptations of words of Akan source by both Ga and Dangme speakers clearly lends credence to the postulation by Peperkamp & Dupoux (2003) that: “Loanword adaptations generally receive a phonological analysis, according to which they constitute phonologically minimal repairs that render illegal foreign forms in conformity with the native phonology. However, more than one such phonologically minimal University of Ghana http://ugspace.ug.edu.gh 258 repair is often available. … there can be more than one possibility to turn a non-native segment into a native one by a single feature change.” (Peperkamp & Dupoux 2003: 368). Indeed, from the analyses done in this thesis, it was concluded that often there is more than one possibility to nativise Akan source non-native vowels. It was observed that a non-native vowel can be systematically adapted to three native vowels in both Ga and Dangme by just a change in a feature or two in the source segment. This was accounted for in OT by the crucial ranking of the markedness constraints *[+low,+ATR] and *[+high,-ATR] in addition to the faithfulness constraint IDENT-IO(Rd) over MAX-IO(F). This ranking ensures that the optimal candidate systematically avoids fatally violating the highly-ranked *[+low,+ATR] and *[+high,-ATR] while at the same time preserving the source roundness at the expense of altering the height and the ATR values of the input vowel. On the adaptations of Akan source non-native consonants, specifically the labial-palatals and the labial-palatalised consonants, their realisations as labialised consonants in both Ga and Dangme are clear examples of adaptations due to segmental ‘deafness’ (following Peperkamp & Dupoux 2002, arguing from psycholinguistic point of view). This is due to the fact that these types of consonants are not present in both target languages. Therefore, the speakers of both languages might not be able to accurately perceive these source consonants in the adaptation process, hence their replacement with the closest ones in the target languages. However, for Akan source palatal fricative [ɕ] to be sometimes adapted University of Ghana http://ugspace.ug.edu.gh 259 asglottal fricative [h] in Ga, when it has [ʃ] in its sound system, cannot be attributed to misperception neither can it be due to the phonology of Ga. The explanation for this adaptation, as has already been offered, was due to historical facts. That is, perhaps at the time of the incorporation of those words from Akan, the underlying segment /h/ had not yet completely undergone palatalization in Akan, presumably as late as in the 19th Century (cf. Stewart 1966 following Koelle 1854, also Dolphyne 1988/2006, Abakah 2004). Within the OT, it was concluded that the phonological adaptations of the non-native consonants can be accounted for by positing a higher-ranked IDENT-IO(lab) and also markedness constraint *[+cor,-cont,+lab] over MAX-IO(F). Per this ranking, the preservation of the labiality of the Akan source non-native consonants together with the avoidance of fatally violating a constraint on the prohibited source consonants, *[+cor,- cont,+lab], is crucially by both Ga and Dangme speakers. Dakubu (2002) had opined about the distribution of the syllabic nasal in Ga that “a series [sequence] of two syllabic nasal onset syllables does not occur” (Dakubu 2002: 33). In the present study, it has been found out that in the process of the adaptation of Akan source words in Ga, it is not only the sequence of two syllabic nasals that is prohibited at the word-initial position, but also all N1N2 sequences, where the N2 is non-syllabic. Again, this sequence is not prohibited only at the word-initial position, but also virtually word-medially as well. While this prohibition may be repaired at the word-initial position through either deletion or vocalic replacement, it is only the former strategy that applies when adapting such sequence at the word-medial position in Ga. University of Ghana http://ugspace.ug.edu.gh 260 With regard to the adaptations of the syllable types in both Ga and Dangme, it has already been established in the existing literature that “…nonnative … syllable types are mapped onto the closest native ones” (cf. Peperkamp & Dupoux 2002: 368-9) by the speakers of the target languages in the adaptation processes. It was, therefore, expected that both Ga and Dangme speakers would repair the illicit N1N2 sequences at word-medial and word-final positions to N2. It is not only NN sequence; the NC sequence is even prohibited at word-initial position in Dangme, hence it is realised as C. This is what has been described in the existing literature as deafness due to phonotactic constraints (cf. Peperkamp & Dupoux 2002; see also Uffmann 2013: 650). The adaptations of the Akan source tone patterns, however, did not seem to following the so-called suprasegmental ‘deafness’ postulations by Peperkamp & Dupoux (2002). This is because it was observed that all the source tone patterns are present in the two target languages (with the exceptions of the tone rule, downstep, which has already been explained is not active in Dangme). Therefore, there is no need for any sort of ‘repair’ of the source pattern in the adaptation process. For example, how could a source HH(H) melody be misperceived as LL(L), and which phonotactic constraint will motivate this adaptation when the source pattern is present in the target language? It has been observed in this study that virtually all the source patterns can be adapted faithfully and at the same time unfaithfully in the target languages. I have not been able to offer explanations for the instances of the unfaithful adaptations, and therefore would suggest further investigations to find answers to such questions. These unfaithful adaptations of University of Ghana http://ugspace.ug.edu.gh 261 the source tone melodies, I have assumed, do not seem to be on either phonological or perceptual grounds. My conclusion, therefore, is that both Ga and Dangme speakers usually preserve the Akan source tonal melodies when adapting them. That is, where available, the source melodies are adapted in the two language target languages as is. However, in addition to faithfully preserving the source melodies, all the source melodies can be realised with different melodies in both target languages. The explanation for this additional melody adaptations, I conclude, is not due to prosodic ‘deafness’, but rather that both Ga and Dangme speakers map a particular source melody to the variety of melodies available in the language that correspond with the structure of the source word. So, for example, a source HH can primarily map onto HH, and additionally HL, LL, and LH in Ga. This might also explain why a source HLM melody can be realised as HMH, HMM, LLH, and LHH in Dangme. In comparing how Ga speakers on one hand, and Dangme speakers, on the other hand, adapt words of Akan source, it is concluded that the adaptation patterns of the two target languages have a lot in common than they differ in in terms of both segmental and prosodic adaptations. The differences in the patterns in the adaptation process have been observed to have resulted from the peculiarity that each of the two target languages possesses. Finally, from the findings made in this thesis, I conclude that both perceptual and phonological factors are equally crucial in accounting for the strategies that speakers employ in the adaptations of Akan source segmental and prosodic features in both Ga and Dangme. Though not discussed in this thesis, University of Ghana http://ugspace.ug.edu.gh 262 non-grammatical factors such as orthography, period of entry of the words, among others could also equally play significant roles in the adaptation process, which potentially supports Uffmann’s (2013) call for the inclusion of such factors in determining the model of the loanword adaptation. 7.4. Significance of the study/contribution to knowledge This study is significant in the following three ways: As far as I know, there is no such study devoted to the loanword phenomenon between two or more Ghanaian languages with the exception of Owusu-Adjei (2007). Others, including Heine (1968), Koffi (1998) and Dorvlo (2008), have usually mentioned the phenomenon in passing in their studies. A few existing literature on loanword phonology in Ghanaian languages have mostly focused on English loans in one particular Ghanaian language or the other with some examples from the Kwa languages as English loanword in Adanme (Dangme) by Nartey (1976); English loanword in Akan by Sey (1990); Addo (2002); Abakah & Tenteh (2005); Adomako (2008, 2013); Apenteng (2013); Apenteng & Amfo (2014), among others; English loanword in Ewe Dzameshie (1996), Agbedor (2006), among others, just to mention a few. Although it is not uncommon to identify some words of Akan origin in some other Ghanaian languages and some words from the other Ghanaian languages, yet no detailed linguistic study has as yet been carried out in this phenomenon. It is, therefore, expedient to establish how the adaptations occur among the Ghanaian languages to facilitate better understanding of the phenomenon cross-linguistically and University of Ghana http://ugspace.ug.edu.gh 263 subsequent phonological theorisations in terms of segmental and suprasegmental adaptations, hence the significance of the present study. The study may also serve as a starting point for similar academic enterprises in case someone wants to replicate it for other Ghanaian languages, which may eventually contribute to our knowledge of the loanword phenomenon in general, and in the Ghanaian languages (specifically among the Kwa sub- family) and African languages in particular. This will go a long way to also contribute to complementing the existing literature on the loanword phenomenon. From a theoretical point of view, although the loanword phenomenon is pervasive among the languages spoken in Ghana, yet not much has been contributed to the current phonological debate over the theoretical approach to loanword adaptation in particular, and in linguistic theory by any of the Ghanaian languages as far as I know. This study has attempted to fill this gap by providing OT analyses and also statements on the approaches to the adaptations of Akan source non-native vowels. 7.5. Recommendations for future studies This thesis has focused only on the study of segmental (both vowels and consonants) as well as prosodic (nasality, syllable structure and tone) adaptations of Akan source words in Ga and in Dangme without discussing the roles that socio-cultural factors such as the generational differences, sex, level of education (literacy) among the speakers that can influence the loanword adaptation process. These factors were not considered at all in this study due to its focus, that is, purely on the phonology of the loanword phenomenon. It is, therefore, University of Ghana http://ugspace.ug.edu.gh 264 recommended that these factors be investigated in future studies so that the account for the inter-languages adaptation between Akan, which belongs to the Tano Central group, on the one hand, and Ga and Dangme, belonging to the Ga- Dangme group, on the other hand, of the Kwa sub-family languages will be complete. Therefore, a comprehensive understanding of the phenomenon needs to also take into account such factors as well as the historical (diachronic) information concerning the adaptation process. There could also be other factors. I, therefore, recommend that these non-grammatical factors also be investigated in the future to complement the findings made in the present study. Some questions concerning literacy and orthography that can be investigated in the future can include whether or not orthography could better explain the vocalic adaptations of the non-native vowels in both Ga and Dangme. For instance, the source non-native vocalic segments [ɪ] and [ʊ] being adapted as [e] and [o] respectively has been attributed to perceptual motivations in this thesis. It is worth noting that incidentally, the source vowels are orthographically represented as the adapted vowels in the following. Source orthography Ga Dangme i). /ʊ/ → [o] [o] ii). /ɪ/ → [e] [e] iii). /æ/ → [a] [a], &[ɛ] It will be interesting to ascertain whether or not orthography can influence the adaptations of Akan source segment in Ga and Dangme. For example, to know University of Ghana http://ugspace.ug.edu.gh 265 whether the source non-native vowels which are adapted as perceptually closer vowels, which match the orthographic sounds, are as a result of sheer coincidence or the orthography rather than perception has motivated the adaptations. Another question that may be of interest to investigate follows from the orthography issue is, how formally literate were Ga and Dangme speakers of the source orthography to have been influenced by it when the Akan source words were entering the two target languages? Though Ga and Dangme speech communities, just as can be argued about most of the speech communities in Ghana, have not been noted to have been formally literate communities earlier in history as compared to what have been established in other parts of the world, yet the fact that it does not necessarily have to take the involvement of the entire speech community to contemporaneously initiate the adaptation or borrowing process makes the orthography argument a possibility. The process usually begins with a few bilinguals in a particular speech community (cf. Uffmann 2013). Therefore, if there is the slightest possibility of orthography influencing the adaptations of such non-native vowels, then it calls for investigation in the future since as Peperkamp (2004) opines about the possible role orthography can play in loanword adaptations that “... in the case of integrated loanwords, the influence of orthography is not always easy to establish, in on-line adaptations that are gathered experimentally, orthography is a factor that can be controlled for” (Peperkamp 2004: 350). See also Vendelin & Peperkamp (2006) for further discussions on influence of orthography on loanword adaptation. One question concerning the different realisations of some of the source consonants in the adaptation process is, at what point in time did the adaptation of University of Ghana http://ugspace.ug.edu.gh 266 the glottal fricative [h] instead of the palatalised variant [ʃ] take place in Ga? When did the Akan source words that contain the glottal fricative, instead of its palatal allophonic variant, find their way into the Ga lexicon? Was it at the time this consonant had not yet undergone the palatalisation rule? Perhaps, future study may assist in finding answers to the above questions. Finally, regarding the issue of tone, as has already been indicated in this thesis, for the analysis of tone adaptations of the verbs, which is complex. I would have needed to set up elaborate and complex experimentations to be able to determine the patterns of adaptation taking into account grammatical factors such as the tense, aspect and mood of the source verbs at the point of the adaptations. This could not have been done in this study looking at how broad the scope of verb tones is. I, therefore, recommend that the tone adaptations of the Akan source verbs be studied in the future to complement a comprehensive study of the loanword adaptation process. University of Ghana http://ugspace.ug.edu.gh 267 APPENDICES Appendix 1: Elicitation of words (for Ga) INSTRUMENT #1. To the respondent/participant: 1. Please I am going to give you some words in English and I would want you to provide me with their Ga equivalents. For each stimulus (word given), you are to repeat your response thrice (3 times). I will do both audio and manual recordings of all your responses. English words Ga equivalent 1. to insult 2. to carry a heavy load 3. to gush out 4. to feel well, well-built 5. sandals 6. main road 7. skinny 8. a medium-sized drum 9. bottle 10. character 11. to measure, calculate 12. spirit, soul 13. Tongue 14. wide, broad, flat 15. to gather, help 16. turn over, capsize 17. to change, turn into 18. box, trunk 19. Witness 20. Groundnut 21. proverb, adage 22. left hand, left arm 23. long-legged goat 24. devil 25. chameleon 26. matter, thing 27. food 28. dwarf 29. fruit 30. flower, blossom University of Ghana http://ugspace.ug.edu.gh 268 31. large canoe 32. meeting, private consultation 33. prostitution, fornication 34. a young girl, woman 35. scissors 36. one thousand 37. large calabash 38. trouble, adversity 39. event, news, report 40. food containing boiled yam, plantain, cocoyam, etc. 41. greed, covetousness 42. truth 43. somebody-or-other, so-and-so 44. mirror 45. scar 46. rebellion, defiance 47. fist, blow with fist 48. glass (bottle) of liquor 49. bride, newly married woman 50. innocence 51. think, thought 52. remember, remind 53. fried fish 54. vagrant 55. fan 56. a clothe measurement, 12 yards 57. liar 58. quarrel, strife, litigation 59. waterfall 60. rainbow 61. testimony 62. pity, sorrow 63. effort, esp. successful effort 64. law 65. taboo, abomination 66. mushroom 67. germ, virus 68. congratulation 69. misconduct University of Ghana http://ugspace.ug.edu.gh 269 Appendix 2: Elicitation of words (for Dangme) INSTRUMENT #2. To the respondent/participant: 1. Please I am going to give you some words in English and I would want you to provide me with their Dangme equivalents. For each stimulus (word given), you are to repeat your response thrice (3 times). I will do both audio and manual recordings of all your responses. English words Dangme equivalent 1. Good character 2. Remind, remember 3. Fried fish 4. Foolish person 5. Law 6. Taboo, abomination 7. Pity, sorrow 8. Effort, successful effort 9. Crippled person 10. Bottle 11. Testimony, witness 12. Noble, royal 13. Poor person 14. Soothsayer, traditional priest 15. Squirrel 16. Turn over, capsize 17. to change, turn into 18. box, trunk 19. cannons, heavy artillery, very big guns 20. Groundnut 21. proverb, adage 22. Christian priest 23. Spokesperson, linguist 24. devil 25. egotism, self-centredness 26. hyena 27. hand-piano 28. spirit, image 29. expression of congratulations 30. a kind of drum & music for chief University of Ghana http://ugspace.ug.edu.gh 270 31. child of a man’s sister 32. servant/maid 33. prostitution, fornication 34. old lady 35. creation 36. royal antelope 37. mirror 38. affliction, suffering 39. event, news 40. boiled green plantain 41. sugar cane 42. domestic cat 43. goat 44. truth, fact 45. cashew nut 46. rebellion, defiance 47. grief, sorrow, sadness 48. bride, newly married woman 49. large talking drum 50. innocence 51. think, reason 52. palm tree 53. whale 54. messenger, angel, helper in need 55. European, foreigner 56. To gather, collect, assemble 57. Exclamation no! 58. Iron pot 59. Buy, sell on credit 60. Picture 61. Bow (with arrow) 62. Assembly University of Ghana http://ugspace.ug.edu.gh 271 Appendix 3: Recordings of social events INSTRUMENT #3. 1. Please I would want you (the group) to perform for me two (2) Ga war songs to be audio-recorded. For Ga respondents. 2. Please I would want you (the group) to perform for me two (2) Dangme war songs to be audio-recorded. For Dangme respondents. University of Ghana http://ugspace.ug.edu.gh 272 Appendix 4: List of Akan loanwords in Ga (Approximately 959 suspected Akan source entries in Ga) NO. AKAN GA GLOSS 1. àbààwá àbààwá,àblà̀àwá Maidservant, female house help 2. àbàkéré àbàkĺé A wristlet or bracelet worn by parents of twin. Also used to dress corpses. 3. ɔb̀àsá àbàsá A bet, wager. 4. àbàsáŋ́ àbàsáŋ̀ Fathom, length of double arm spread, including the width of the chest 5. bɛ ́ àbɛ ̀ Pincers 6. ɛb̀ɛ ́ àbɛ ́ Proverb, adage; indirect speech of any kind 7. abɛb̀re àbɛ́ꜝ bŕé A variety of palm fruit, very dark when unripe and deep red when ripe 8. ?abebuo àbɛb̀ùá Parable, illustration, proverb, proverbial expression 9. abɛb̀re àbɛ́ꜝ dáŋ ́ A variety of palm fruit, deep green when unripe. 10. abɛde àbɛd̀é Commission on a sale, esp. of fish 11. benkum àbɛk̀ú Left hand, left arm, left side 12. abenkuma àbɛk̀úlɔ Left-handed person; awkward, clumsy person 13. abɛt̀inta àbɛt́íǹtá A palm tree with a branching stem 14. abɛt̀iŋ, abɛt̀ɪnɪ àbɛt̀íŋ,́ àbɛŋ̀t̀íŋ ́ A very old, tall palm tree 15. ?æ̀bísá àbìsà Fortune, destiny, lot, situation in life 16. abǒraá àblà̀á European woman (obsolesce.) 17. abereʨi àbĺéʧí Long-legged goat 18. abʊrɔɡua àblò̀ɡʷá, àblò̀ŋʷá Deck chair, arm chair, straight chair. 19. ?abobonua àbòbònùà A kind of bird, woodpecker. 20. abɔdɪammɔ àbòdìàmɔ ́ Any round green glass bottle. 21. àbòḿ àbòḿ Simple stew 22. ɔb̀ʊ̃̀ǹsaḿ́̃ àbòǹsáḿ, àbòǹsáŋ ́ Devil 23. ɔbʊsʊm-akɪtɪw àbósúmáŋḱɛ́ꜝ tɛó́, Chameleon àbósúmákéꜝtéó, abusumakutre 24. aboʥekutu, àbɔʤ̀èkútú Whiskers, sideburns abɔʥɥe-futu 25. abradɪ àbrà̀dè, àblà̀dè Dignitary, high-born person, person of character, person of royal lineage University of Ghana http://ugspace.ug.edu.gh 273 26. aberekuri ábŕékùlí, àbĺéꜝkúí, Moray eel, painted eel. ábĺékù 27. æbusᶣia àbùsuã̀ ́̃ Family; the nuclear family. 28. æbusᶣiafʊ àbùsuà̃ f̀̃ óì Relatives 29. ɔd̀àbáŋ ́ àdàbáŋ́ Door bar (iron bar) 30. àdàfi ́̃ (to betray, àdàfí News, information, announcement. disclose, forewarn) 31. àdáḿ àdàŋ,́ àdaŋ̀̃ ,̀ àdaà̃ ,́̃ A type of music played by hunters’ àdɛì̃ ́̃ association 32. ɔd̀áẃ (dàẃ-bɔ ́ dàwùʧò, àdàòʧò Jawbone ‘conversation) 33. àdáwʊ̀rʊ̀ma ̀̃ ádáróòma ̀̃ Grace, assistance, intervention 34. àdɪ ́ àdé Matter, phenomenon, thing, existence 35. àdɪb́ɔ ́ àdébɔɔ̀ ̀ Nature, creation, the natural scheme of things 36. àdéŋḱũḿ àdéŋḱũ̀ A calabash rattle, a kind of music in which a rattle is used 37. àdɪs̀a ́̃ àdèsa ́̃ Story, tale 38. ádɪs̀a,̀̃ ádàsa ̀̃ ádèsa ̀̃ Human being 39. *àdɪʨ̀ɪà̃ ǹ̃ i ́̃ àdèʧaà̃ ǹ̃ i,́̃ àdìʧaà̃ ǹ̃ i ́̃ Early morning tasks, the custom of visiting the grave a day after burial… 40. ádɛ ̀ àdɛ ́ A bead 41. ɛd̀ɪŋ́̃ ́(difficulty) àdɛŋ́ ́ Danger (semantic shift) 42. æ̀dì-æ̀ʨírí àdíáʧírí Store-keeper, steward (subordinate officer) 43. æ̀dìbaŋ̀̃ ́ àdìbaŋ̀̃ ́ Food, heavy or staple food 44. àdárɪ ̀ àdídà, àdédà Billhook, pruning shears 45. æ̀dìŋk̀rà́ àdìŋk̀rà́ Cloth on which traditional symbols have been printed 46. àdʊ̀bɛ ́ àdòbɛ ́ Raffia palm 47. æ̀dᶣìàwá àdùàwá Fruit 48. æ̀dìfùdɪ ́ àdùfùdé Greed, greediness, avarice 49. ɔf̀a,́̃ àfa ́̃ àfa ́̃ Part, half, piece, portion 50. àfaà̃ f̀̃ a ́̃ àfaá̀̃ fa,́̃ àfaf́̃ aá̃ f́̃ a ́̃ Sideways, one eighth 51. fá-baŋ́̃ ́ àfábaŋ̀̃ ́ Wall, fence, protection; both spiritual University of Ghana http://ugspace.ug.edu.gh 274 and physical 52. fá-dáŋ ́ àfáꜝdáŋ ́ Swish building 53. àfànìwàá àfànìwàá Name of a variety of cassava 54. àfàsɪẃ́ (Ewe?) àfàséó, àfàsó Winged yam, water yam, violet, purple, purple blue 55. àfɪ ́ afí (àfí=partridge, Year bush-fowl) 56. fʊ́fʊŕ ɔ ́(new) àfòfòrò,àfòfrò̀, Flower, blossom, bud àfòflò̀ (semantic shift ) 57. àfɔr̀ɪ ́ àfɔl̀é Sacrifice, offering 58. æ̀fṹ àfṹ Bulge, hump, hunch, bubble, lather, foam 59. æ̀fùá àfùá Mist, haze 60. æ̀fṹpɔŋ́̃ ḱɔ ́ àfṹkpɔŋ́ɔ,́ àfĺúꜝkpɔŋ́ɔ ́ Camel 61. æ̀fùɲaŋ́̃ ́ àfuɲ̀̃ aŋ́̃ ́ A hairy worm 62. æ̀ʥań́̃ àɡaì̃ ́̃ Arrow 63. ʥáẁ àɡàwú Gap, space between the teeth 64. àɡùàséḿ (no Ak. àɡʷàséìàŋ,̀ àɡʷàséàŋ̀ Respectable standing in society, as of word) an elder 65. àɡùà-àsɪ ́ àɡʷàséŋ ̀ Assembly of male elders representing the community 66. ɔh̀àǹtáŋ ̀ àháńtà A large tree 67. àhɛm̀̀bá (Fn) àhim̀̃ a ́̃ A large type of canoe (with its crew) 68. ahɪnɛsɪ (no Ak. àhiŋ́̃ ɛs̀é A large boat, lighter formerly used to word) unload ships berthed beyond the surf 69. àhɪŋ̀̃ k̀ʷàá àhiŋ̀̃ k̀ʷrà́ King’s servant, attendant 70. àhʊ́rá àhórà A very serious insult, taunt 71. ?àhʊ́rɔ ́ àhɔĺé Sputum, phlegm 72. hũ (see) àhṹ Chance discovery 73. æ̀hũḿ àhũḿ, àhũŋ,́ àhuꜝ́̃ mɛ ́ Wind, storm, gale, blast of wind 74. æwuɲadɪ àhṹɲad́̃ é Auction, sale by auction (semantic shift) 75. àhʊ̀̃m̀ka ́̃ àhũŋ̀ k̀á Joy 76. ahurutoa ahurutia àhùrútíà A poisonous striking snake, a viper 77. æ̀ʥàm̀má áꜝʤámà A kind of plant, a small tree or shrub 78. æʥina àʤìna ́̃ Meeting, private consultation, arbitration 79. æʥinamʊa àʤrà̀mòá, àʤàŋm̀òá An appellation for the domestic cat University of Ghana http://ugspace.ug.edu.gh 275 80. æʥɥumaʥaŋ àʤúmaʤ́̃ aŋ́̃ ́ In vain, for nothing, to no purpose 81. æʥɥumataŋ (no àʤúmat́̃ aŋ́̃ ́ One’s working clothes, which are put Ak. word) into coffin when he dies. 82. æ̀ɡùàmáŋ ́ àʤʷàmaŋ́̃ ́(Is it from Prostitution, fornication Asante?) 83. àʥɥɪŋ́̃ ́ àʤʷɛŋ́ ́ A bird, any wagtail, mud-fish, freshwater catfish 84. akaʥinam àkàʤìnam̀́̃ [no Surety transcription] 85. àkań́̃ àkaŋ́̃ ́ Competition, challenge, contest 86. àkǽ̃ńsí(é) àkaŋ́̃ ʃ́iì̃ ̀̃ Competition 87. àŋk̀áẃ àkáò The fresh-water tiger-fish 88. àkàsàá àkásàá, àkáʃàá cymbal 89. àkásé àkásè Blue, bluing, washing blue 90. àkàtàmań́̃ sʊ́ àkàtàmań́̃ sò Chief’s umbrella 91. àkàtàɥìá àkàtàŋẁìá Umbrella, parasol 92. ʨéʨé àkéké [no tone Spindle (obsolesce.) marking] 93. ɲʨèʨèrèé àkèkèré, àkèkrè́, Headband, headgear, crown àkèklè́ 94. àkʊ̀rʊ̀tɪá̃ à̃ ̀̃ àklò̀tiá̃ ,̀̃ àklò̀ǹtiá̃ ,̀̃ Pin, straight pin, dressmaker’s pin akontia, akɔntia 95. kófi ́̃ áꜝkófí Nursing bed for seedlings of e.g. yam 96. àkʷæ̀tìá àkòtìà Dwarf, a drum used in music for particular deity 97. akotròko ákòtrò́kò, àkóꜝtŕókò A long-stemmed, edible mushroom 98. àkɔt̀ʊá̀ àkóꜝtówá A small barrel, keg. 99. akɔdi, ɛkɔm [no àkɔd̀ì Hunger Ak. word] 100. àkʊ̀̃fà na ́̃ àkɔf̀̃ là ́̃ A horn, originally made from an elephant tusk 101. akʊkɔ bɛsa àkɔḱɔb́ɛśa ́̃ A thorny shrub 102. akʊkɔ wa àkɔḱɔẁá A young girl or woman [semantic ‘chicken’ shift] 103. akɔnɪa ba àkɔǹìàbá, ŋʃ̀rà̀hṹ Missing, failure to make contact 104. ãkɔń́n ɔ ́ àkɔńɔ ́ Quality of being delicious, appetizing, desirable; desirability, lust, University of Ghana http://ugspace.ug.edu.gh 276 infatuation 105. àkóḿ , àkɔḿ́ àkɔŋ́ ́ A type of music associated with traditional worship 106. ?àkɔk̀ ʊ́̃nʊ̃́ àkɔŋ́k̀rɔ̀ ́ A kind of maggot found in rotten palm trees 107. ɔk̀ɔt́ɔḱ úrɔ ́ àkɔt̀ɔk̀ĺò A hooked drumstick, a wooden fork used in thatching 108. àpàka ŋ́̃ ́ àk͡pàkaì̃ ́̃ Palanquin, travelling basket 109. àpàʨɪ ́ àkp͡ àké Crippled person 110. àpáʨɪ ́ àkp͡ ákí A large calabash 111. pàdɪɛ́,́ pàrɪ ́ àkp͡ àlé Rafter, ceiling support 112. m̀pæ̀ǹ tú, àk͡paǹ̀̃ tú A kind of plantain, used to prepare apænturɛŋ, fufu. àpæ̀ǹtú 113. àpàsʊ ɔ́ ,̀ àpàsʊ̀ɔ ́ àk͡pàsó Scissors; a kind of witch who cuts the victim’s blood vessels 114. apatɪr am àk͡pàtrà̀mɔ ̀̃ Butter bean; lima bean 115. àpɪḿ́̃ àkp͡ é Thousand, one thousand 116. àpɪśɔẃ́ àkp͡ ɛśɛó́, àpɛśɛó́ A small animal, the potto 117. pétèpè ré? àkp͡ ɛt̀ɛk̀͡plè̀bí A small shrub-like savannah tree, and its fruit 118. æ̀pítì, æ̀pítí àk͡pítí A cake, made of corn 119. æ̀kŕó, èkúró àkrò́/àklò́ Scar, growth over a sore 120. àkʊ̀rɔ m̀a ́̃ àkrò̀ma,́̃ àklò̀ma ́̃ Hawk, lizard-buzzard; an animal impersonated by cult dancers 121. akron noi? àkrò̀nòí, àklò̀nòí, Rupture, hernia, of men àklò̀nùé 122. æ̀kùrò wá àkrò̀wá, àklò̀wá Village, farm settlement 123. akrub ɛʨɛ àkŕúbɛʧ́ɛ ́ Deformity of the arm due to injury or birth defect 124. èkúẁ àkú Heap 125. àkʊ̃̀ǹt á àkù̃ǹtá Brother-in-law 126. ŋk̀òɲa à̃ ́̃ àkùɲaà̃ ,́̃ ŋkuɲaa, Magic, magical rites; a kind of charm akuɲaa 127. æ̀kùtú àkùtú Orange, grapefruit 128. àkʷáà bà àkʷáàbà Welcome! 129. kʷæ̀d ú àkʷàdú Banana 130. ?ànaǹ̀̃ tá àlaǹ̀̃ tá Condition of being knock-kneed 131. àma,̀̃ ?àmań́̃ àma-̀̃ obsolesce. Gum, glue; tar, pitch University of Ghana http://ugspace.ug.edu.gh 277 132. amakw anʨɥa? ámaḱ̃ ʷaǹ̀̃ ʧṹa ́̃ A small animal, the bush genet. No Ak. word [semantic shift from fem. name-not sure if personal] 133. àmaǹ̃ ɪ ́̃ àmaǹ̃ é Trouble, adversity 134. amanɪ hunu àmaǹ̃ èhṹlṹ Affliction, suffering 135. àmaǹ̃ ɪ ɛ̀̃ ́ àmànìɛ ́ Events, news, report 136. àmaɱ̀̀̃ fʊ́̃ àmaŋ̀̃ f̀ó, àmaŋ̀̃ f̀ɔ ́ Old town, original or home town; desolated, ruined place, ghost town 137. amaŋk ani àmaŋ̀̃ k̀aǹ̃ i ́̃ A root crop, cocoyam, taro 138. am̀pàŋ ̀ áḿpaŋ̀̃ ́ áḿpaŋ̀̃ ,́ áḿpaŋ̀̃ ̀ Fruit bat 139. áḿꜝpé áḿꜝpé A rhythmic jumping game, mainly played by girls 140. m̀pɪm̀̀̃ ma ́̃ am̀̀̃ pèma ́̃ A type of plantain 141. àm̀pès í am̀̀̃ pèʃí Food consisting of plain boiled yam, cassava, plantain, or cocoyam 142. æ̀níbɪ ŕɪ ́ àníbŕé, àníbĺé Greed, covetousness 143. ?æ̀nìd ań̃ ɪ ́̃ ànìdaí̀̃ Women’s disease 144. æniha w àníꜝháó Laziness 145. ànʊ̀̃m aà̃ ́̃ + kʷàʨɪ ́ ànòmákʷàʧé A kind of bird 146. ànʊ̃̀kʷ á ànɔk̀̃ ʷá Truth; indeed, truly, really, honestly, actually, in fact, as a matter of fact 147. ɔǹʊ̀k̃ ʷ áꜝ́fʊ́ ànɔk̀̃ ʷaꜝ́fó A truthful person 148. nʊkʷa rɪ, anʊkʷadɪ ànɔk̀̃ ʷálé Truth 149. tɛw, a ntɛw (Fn), ań́̃ ꜝtɛó́ Ambush, waylay ntɛ 150. ?æ̀nim̀̃ ũ̀òɲaḿ́̃ ànũ̀ɲaḿ́̃ Glory 151. áŋḱʊ̃ń aḿ́̃ aŋ̃́ ḱònaḿ́̃ Loner, unsociable person; a kind of beads, worn esp. by widows 152. ansam aŋ́̃ śaà̃ ,́̃ ań́̃ saŋ̀̃ ,́ ań́̃ sam̀́̃ Guinea fowl 153. àǹʨɥɛ r̀ɪ ́ àŋʧ̀ʷèlé Ladder, steps 154. pàá àpàá Hire, job 155. àpààw á àpàáwá A flat round wooden tray, carried on the head or placed on a table for displaying and carrying items 156. àpaḿ́̃ àpa,́̃ àpaŋ́̃ ́ Bonds, shackles, fetters, anything used to secure one who is to be beaten or executed… 157. apásá àpásá Falsehood, lie 158. àpásá ꜝfʊ́ àpásáꜝfó Liar, gossip, idle chatterer 159. apɛsɛɪ̀ ́ àpɛs̀ɛɛ̀,́ opɛnsɛ Brush-tailed porcupine University of Ghana http://ugspace.ug.edu.gh 278 160. àpɪśɔw àpɛs̀ɛò́, àk͡pɛs̀ɛò́ A small animal, Bosman’s potto 161. aprap ransa àprà̀prà̀ǹsa ́̃ A food, a corn and palm dish 162. mpun impu àpúnam̀̀̃ pú Head-on, face-to-face, on the spot 163. àsá àsá Central room, hall, sitting room 164. nsaasa (wa) àsààsáꜝá Patchwork, made with cloths of many designs sewn together 165. ɔsajɪrɪ àsààjéré, àsrà̀jéré A type of war music and dance, performed by women 166. àsábɪr̀ á àsáblà́ (obscele) A children’s disease 167. àsàfʊ̀ àsàfò Military company; group, congregation, assembly 168. àsàfʊ̀k ań́̃ àsàfòkaŋ́̃ ́ Leader, head of an army, head of a vanguard 169. sam̀̃ aŋ̀̃ ̀ àsam̀̃ aǹ̃ úkpá Small supernatural person, dwarf; an ape, the chimpanzee 170. àsrà̀m̀ àsrà̀ Fever 171. asŕà àsŕà Snuff; brown 172. ɔs̀ésé àsésé, àséséù Hut, temporary shelter; tabernacles 173. àsɛm̀̀s ùróẃ àsɛm̀̀̃ srò̀ Fear of trouble 174. ǹsam̀̀̃ d ʊ́ (Fn) àsɛǹ̀̃ dó Space between roof and ceiling; access hole in ceiling; ceiling 175. àsìáńn àá aʃana (No trans. A food Obsolesce.) 176. ?æsie du kɛtɛkɪrɛ àʃíèdúkɛt̀ɛk̀lɛ̀ ̀ Initial words in a drummed appellation for Ga Mashie; an electoral district, Central Accra 177. æsikr e-sìam̀́̃ àʃìkìǹʃaŋ́̃ ,́ ʃìkìǹʃaŋ́̃ ́ Wheat flour 178. æsìkò (tò) àʃíkò A game (Obsolesc) 179. æ̀símæ̀ sí àʃímàʃí, òbɛǹ̀tɛŋ́ ́ Somebody-or-other, so-and-so 180. asitíw (deafness) àʃìtíú Making an error due to inattention, carelessness (Semantic shift) 181. æsiwa àʃìwá A type of music 182. aɕɥɪɕɥ ɛ àʃʷìʃʷɛ ́ Mirror 183. asʊpr ɔpɔ àsòk͡plò̀ White mangrove; button wood tree 184. asrafʊ kuw àsrà́àfóàkú Regiment 185. asraaf ʊ àsrà́àfóɲò Soldier 186. atadɪ àtàdé, àtàlé Dress, garment, shirt any price of dress esp. of European style; clothing, apparel 187. àtɛŋ́̃ ́ àtɛŋ́̃ ́ Judgment University of Ghana http://ugspace.ug.edu.gh 279 188. ?àtiá̃ ́̃ àtiá̃ ̀̃ Cashew nut 189. ætírìd ìí àtírìdìí, àtŕìdìí Fever, malaria 190. àtʊ̀fʊ̀ àtòfò A pillow worn by women at the back, as a bustle 191. ǹtɔm̀́m ɔ ̀ àtómò Sweet potato 192. àtɔr̀òp ó àtrò̀k͡pò, àtrò̀k͡pó, Aubergine, eggplant àtrò́k͡pó 193. atrɔm a àtrò̀ma,́̃ àtɔr̀ɔm̀a,́̃ Morsel, bit of food, mouthful of food; àtrɔ̀m̀a ́̃ confusion 194. aʨɛ ́ àʧɛ ́ A basket-work trap, used in river- fishing 195. àʨɪḿ́̃ àʧɛḿ́ The Akyem country 196. aʨɪŋ́̃ ʨ́ ɛŋ́̃ ́ àʧɛŋ́̃ ʧ́ɛŋ́̃ ,́ àʧéŋʧ́éŋ,́ Unripe palm nut àʧɛń́̃ ʧŕɛ ́̃ 197. æ̀ʨiá̃ ḿ́̃ pɔǹ̀̃ (Also àʧiá̃ m̀̀̃ pɔŋ̀̃ ̀ A herb, Acheampong, Busia, Ak. personal Krawuna or Siam weed. name) 198. èʨɥa ́̃ àʧʷa ́̃ Scar, cicatrice, cuts made on the skin; quarrel, difference, misunderstanding 199. ?aʨɥɛ rɪ àʧʷèré Ladder, staircase 200. àʨɥɛr̀ ɪ ́ àʧʷɛr̀ɛ,́ àʧʷɛl̀ɛ ́ Fist, blow with the fist 201. æ̀tᶣià̃ ́̃ àtũ̀a ́̃ Rebellion, defiance 202. æ̀tᶣià̃ ́̃ àtù̃aʧ́̃ élɔ ̀ Rebel 203. æ̀tùm̀ paŋ́̃ ́ àtũm̀̀ pań̃ i,́̃ àtũ̀m̀paŋ́̃ ,́ A large kind of drum; a type of court tù̃m̀pań̃ i ́̃ music associated with this drum 204. tumpa ŋ́̃ àtũ̀ŋk̀pa ́̃ Glass bottle of liquor, glass bottle 205. túrúẃ ‘spatter’ àtúꜝtŕú Chaff, fragments, remnant of shot expelled when a rifle is fired 206. àwàrɪ ́ àwàlé Spoon 207. ɔẁàrɪ ̀ àwàrè, àwàlè A game played by two people at a time, using four counters per hole on a board of twelve holes, six for each contestant 208. aɥɪrɛf iri àwèrɛf̀í, wèrɛf̀í Forgetfulness 209. àɥɪr̀ɛh̀ ʊ́ẃ àwèrɛh̀ó Grief, sadness 210. æɥí (t heft) àwí Wickedness (Semantic shift) 211. awʊsu àwòsó, àwùsú Birth waters 212. æ̀wùr àá àwùlá, àwùrá Lady, lady-like person; title for a young lady University of Ghana http://ugspace.ug.edu.gh 280 213. æ̀wùlá -æ̀kᶣíꜝá àwùláꜝákúꜝá A kind of cassava 214. æwule , ewure àwùlé Surgeon-fish (Fn) 215. æ̀ɥísía á̃ ́̃ àwúsa,́̃ àwísa ́̃ Orphan 216. jɪr̀á (l ose, be lost) àjálá An absent person; the supernatural power enabling one to become invisible, vanish 217. àjʊ́wá àjáwá Copper, brass; a metal container; tray, salver, metal box, brass pan 218. àjɛf̀ʊ́r ɔ ́ àjèm̀fŕó, àjɛm̀̀fŕó, Bride, newly married girl àjɛm̀̀fĺó 219. ojibir ibi (a tree) àjíblì̀bì A kind of beads, made of black seeds (Semantic shift) 220. ajɛfar ɪ àjífálé, àjéfáré, A fine, claimed by husband from àjífáré another man for adultery leading to divorce 221. ?báhà báhà Fainting; extreme tiredness 222. ɔbaɲʨ ɪ baǹ̀̃ ʧí Cassava 223. baŋ́̃ ́ baŋ́̃ ́ Boundary, limit, demarcation 224. baŋ baŋ́̃ ́ Character, nature 225. bàtàfʊ ́ bàtàfó Bush pig, wild pig 226. bàjɪŕɛ ́ bàjálé, bàjérɛ,́ bàjɛĺé A kind of yam 227. ?baas ʊnfʊ beesonfo (no tone Chief fisherman’s council marking) 228. ɛb̀ɪḿ́̃ béḿ, béŋ ́ Innocence 229. ?bɪḿ́̃ b ú béḿbùù Innocence, declared innocent 230. bɛń́̃ tʊ ́á béńꜝtóá A plant; a calabash made from the fruit of the plant; a calabash or other container used as a syringe for giving enemas 231. béw béó Armadillo 232. àbébé ẃ bɛ́ꜝ bɛó́ A tiny river snail 233. ɔb̀ɛd́ɛ ẃ́ bɛd́ɛó́ A makeshift basket, made of palm leaves, used on the farm 234. bɛdiak ɔ (come eat bɛd́íákɔ ̀ A variety of cassava and go) No Ak. word 235. bǽɡúá bɛɡ̀ʷà Any meeting of elders 236. ʥina (stay) bɛʤ̀ìná, òbɛʤ̀ìná An exclamation made while shaking hands after an outdooring University of Ghana http://ugspace.ug.edu.gh 281 237. bɛŋ̀̃ ʨ̀ɛ ́ bɛŋ́̃ ḱɛ ̀ Approach, draw near, be close to something 238. bɛt́ɪb́ɛ t́ɪ ́ bɛt́ɛb́ɛt́ɛ,́ batabata Loquacious, talkative 239. bisiba saa bísíbàsàà Confused 240. bitibit i bítíbítí Close, crowded 241. bɪŕɛẃ́ bĺɛó́ Palm branch, palm frond 242. bɪr̀ɛẁ̀ blɛ̀ò̀ò Slow, soft, slight 243. buroɲ a bĺóɲa,̀̃ bĺòɲa ̀̃ Christmas 244. buros uo blò̀ʃóó Large and long, of pawpaw (obsolesce.) 245. brɔb́rɔ brɔ blɔ̀b̀lɔ̀ɔ̀b̀lɔ̀,̀ Delicious bĺɔb́ĺɔɔ́b́ĺɔ ́ 246. bʊ̀rɔf̀ʊ ́ blɔ̀f̀ó Pertaining to or derived from European or white culture; anything perceived as foreign 247. burob uro bĺúbĺú All 248. bʊ́àpɪ á̀ bòápìà A kind of monkey 249. bòdᶣíá bòdúá, ahuja Horse-tail fly whisk 250. ?bòka ́̃ (Fn) bòka ́̃ East 251. bòǹsù bòǹsò Whale 252. bóńsú wá bóńsúá A type of small pot 253. òbóǹt ó bòǹtò A type of fishing boat, trawler 254. ɔb̀ósá ẁ bóꜝsáó, bóꜝsá Plantain fibre, used as sponge or toilet paper 255. bosro bo bòsrò̀bò An appellation for the sea 256. ?bɔ ̀ bɔ ̀ Create, set in motion-numerous expressions 257. ɛbɔ bɔ A kind of flying ant 258. ɔb̀ɔ́ꜝ fʊ ́ bɔf́ò Messenger; ambassador, envoy, high commissioner; angel 259. ɔb̀rà́ brà́, blà́ Menstruation 260. ɔb̀rà́ꜝf ʊ́ brà́ꜝfó, blà́ꜝfó Executioner; an appellation for the odum tree 261. bírìsìí bĺìʃìí, bŕìʃìí A black-dyed cloth 262. bʊ̀á búà Assemble, gather 263. ?bùá bùá Judge, arbitrate 264. æ̀bùàd á búádá, bùádá Fasting 265. bùtú bútù Capsize, overturn be overset; overthrow, annihilate University of Ghana http://ugspace.ug.edu.gh 282 266. dà (as ɪ) dà Thank 267. da da ́̃ Stand, pose 268. dáá; d a (day) dáá Always, perceptually, every day, every time 269. ɔd̀áẁ dáà Mouth, mouth cavity 270. ?dæ̀æ̀ bí, dæ̀bí dààbí, dàbí Excl. no 271. ?dæ̀bí dà dàbídà Excl. no, emphatic denial 272. dàdàà dà dàdààdà Long time ago 273. dàdɪ ́ dàdé Iron, metal; equipment; cutlass; cash 274. dàdɪb̀ aŋ́̃ ́ dàdèbaŋ́̃ ̀ Iron bar, cross-bar 275. dàdɪs̀ɛ ŋ́̃ ́ dàdèsɛŋ́̃ ́ Iron pot 276. damfʊ ̀ʥan dam̀̀̃ fòʤàì A double-edged knife or dagger, with a sheath 277. dam̀̀̃ p aŋ́̃ ́ dam̀̀̃ paŋ́̃ ,̀ àsá Hall, sitting room 278. daŋkʷ ansɪrɪ dàŋk̀ʷàǹsèré An edible bat 279. dédé dédé Make a noise 280. ɔd̀ɛẃ̀ dɛɛ́ŋ́ ̀ Nicely, attractively 281. dɛɛʨi riaɱfa dɛɛ̀ʧ̀ŕíàɱ̀fa ̀̃ Name of an ‘abɛbua’ emblem of the Otublohum division of Accra, a bird looking back 282. deŋde ŋ dɛŋ́d̀ɛŋ́ ̀ Carelessly, roughly; very hard, assiduously 283. dì (ke ep in mind) dì Be mindful of; follow; be habitual, a bad habit 284. didim u dìdìmú Glutton 285. diŋ́̃ ŋ́ ́ díŋŋ́ ́ Quiet 286. dòḿ dòḿ, dòŋ ́ Travel, go abroad, emigrate, set off 287. dʊ̃́ḿ ( favour) dómɔ ̀̃ Trust, accept, believe in (Semantic shift) 288. dòẃ ( Fn) dòò Be browned by roasting 289. dɔ ́(lo ve) dɔ ̀ Be helpful, kind, considerate; cherish (Semantic shift) 290. dᶣia-a dɪ (stick- dùàdé Cassava, manioc yam) 291. dᶣiada n dùáꜝdaí̃ ́̃ A building of the wattle and daub type 292. dᶣìásʊ d̀ ᶣìá dùásòdùá, dùásèdùá Name of several semi-parasitic plants 293. dᶣìé dùé Excl. an expression of sympathy, condolence 294. ɛf̀iń́̃ , ò fié́̃ ń (Fn) fɛí̃ ́̃ A fish, marine spinehead or scorpaenid University of Ghana http://ugspace.ug.edu.gh 283 295. ?fɛẃ́ fɛó́ Beauty; goodness, pleasantness 296. ?fɛw fɛó́ Teasing 297. fitaa fítáá, fútáá White, pure white, clean 298. ?fìrí flì́ Buy or sell on credit 299. fʊrɔ flɔ̀ɔ̀,́ flɔ̀,́ flɔ̀ò́ Stew; meat or fish and vegetables cooked in oil 300. ?fɔ ́ fɔ ̀ Be wet; soak, wet, drench 301. ?fɔ ́ fɔ ́ Guilt 302. fùfú fùfùí A food of pounded cassava, plantain, yam, etc. 303. ?fùtú fútù Mix, join 304. fúꜝtúr ó fúꜝtúrú Gold dust 305. æ̀ʥaŋ́̃ ́(Mande ɡaì̃ ,́̃ ɡaǹ̃ i,́̃ ɡɛì̃ ́̃ Bow; a spiny structure on the back of source) a catfish; porcupine quill 306. ɡʷá ɡʷá Meeting, council 307. ɡʷáꜝb ɔ ́ ɡʷábɔɔ̀ ̀ Assembly, conference 308. òɡʷàǹ tɪŋ́̃ ́ ɡʷaǹ̀̃ tɛŋ́̃ ́ Sheep; fool 309. ɡùàẁ ɡʷàò, ɡʷàà Whip, beat, lash, flog 310. hamah ama ham̀̃ ah̀̃ am̀̃ a ̀̃ Wavy, rough, of water 311. haŋŋ háŋŋ́,́ heŋŋ Bright, well lit. 312. hàẁ hàò Worry, bother, annoy 313. haw hàòhàò Worried, upset 314. háráŋ ́ háráŋŋ́́ Elegantly, beautifully, of dressing 315. ɕɛŋ̀̃ ŋ̀ ̀ hɛŋ́̃ ŋ́ ́ Flame, colour, bright 316. hià̃ ́̃ hià̃ ́̃ Be necessary, be important 317. àhɪḿ̃ a ́̃ hiḿ̃ a ́̃ Very early 318. wòsóẃ hósò Shake, quake 319. hɔ-́nɪ-̀ hɔ ́ hɔńìhɔ,́ hɔńɔh̀ɔó́, Half and half, in half measures, more hɔh́ɔńɔh̀ɔó́ or less, so-so 320. hòrón óà hŕɔń̃ ɔ,̀̃ hĺɔń̃ ɔ ̀̃ Blister; swelling around a wound 321. hùrúw hrù́, hlù́ Jump, jump about 322. húẁ hùù Blow, blow on; fan; fight, wage war; beat; boasting 323. hùàŋ ̀ hʷaŋ̀̃ ̀ Dislocate, dislodge, a joint; limp, drag foot along the ground when walking; reduce, cut 324. ɕɥɪá̃ ɕ́̃ ɥ ɪá̃ á̃ ́̃ hʷiá̃ h́̃ ʷiá̃ ,́̃ hʷiɛhʷiɛ Thin, slender, slim University of Ghana http://ugspace.ug.edu.gh 284 325. ʥàw ʤá Worship, revere 326. ʥáásɪ ́(royal ʤáásé Council, state council (Semantic shift) household) 327. ?ʥàbá ʥàbá ʤàbéʤàbé Haphazardly 328. ʥɥàrɪ ,́ guàrɪ ́ ʤálé Rinse, wash without soap (bathe) 329. ʥáŋ ́ ʤáŋ ́ Common, simple, ordinary , mere; empty 330. ʥata (Mande ʤàtá Lion. The emblem of the Gbese ʥara) division of Accra 331. òʥátɛ ŋ́ ́ ʤátɛŋ́́ Torch, made of dried palm fronds 332. ? ʥàw ú ʤàwú A variety of yam 333. ʥìʥá ʤíꜝʤáá Lose control, panic; get out of control 334. ʥìʥé ʤìʤé Resound, echo; provoke, tease 335. ʤìmí ʤìmí Go mad, be silly, be stupid 336. ?æ̀ʤì na ́̃ ʤìná Consideration, thought, discussion 337. ʥɥòŋ k̀ú ʤòŋk̀ú Hip joint, thigh bone, loins 338. ʥɥó ʤú Monday, “female” day in Ga tradition 339. ʥɥófɔ ??? ʤúfɔ ̀ Tuesday, “male” day in Ga tradition 340. ʥɥá ʤʷrà̀, ʤʷlà̀ Break into pieces; sabotage 341. ʥɥeti ri ʤʷétŕí Treasure, money, capital for trading; silver 342. ʥɥè ʤʷɛɛ̀ ̀ Briefly and sharply 343. ʥɥɪŋ́ ̀ ʤʷɛŋ̀ ̀ Think, thought 344. ʥɥɪt́ɛ ̀ ʤʷìɛt́ɛì́, ʤʷùɛt́ɛɛ̀ ́ Silver, ornaments 345. æʥɥi ni, æʥɥine ʤʷíné, ʤʷúné A collection of trinkets 346. ʥɥìrá ʤʷlà́ Bless, cleanse, purify 347. ?ka ́̃ ka ́̃ Turn fufu in a mortar 348. ? ka ́̃ ka ́̃ Swear; decide, determine; reprove, reprimand 349. kàbìì kàbìtìì Thick darkness 350. káɱ́fʊ ́ káfò, kaŋ́̃ f́ò, káɱ́fò Praise 351. kàɪ ́ káì Remember, remind 352. ?kǽꜝʥ á káʤà The root of kanya, used as a chewing stick 353. kakad uro kákáóʧòfa,̀̃ kákáʧòfa ̀̃ Ginger 354. krà ́̃ klà́ (no nasal) Guess, Divine, foretell University of Ghana http://ugspace.ug.edu.gh 285 355. kaɱfr atoto (Fn) kaŋ̀̃ f̀làt́̃ ótó The scales of the long-finned herring; a small silver coin 356. kàǹtó (crooked kàǹtó Shin, shin bone, tibia legs) 357. kań́̃ ʨɪ ḿ̃ a ́̃ (Fn) kań́̃ ʧiḿ̃ a ́̃ Catering, the practice of a woman cooking for an unmarried man; female caterer 358. kæ̀ɲaŋ́̃ ̀ kaɲ́̃ a ̀̃ Urge, encourage, prompt, stimulate, inspire 359. kàɲaŋ́̃ k̀àɲaŋ́̃ ̀ kaɲ́̃ aḱ̃ aɲ́̃ a,́̃ kaɲ́̃ aá̃ ́̃ Thin, skinny 360. kàŋk̀à nɪ ́̃ kàŋk̀áŋ ́ Civet cat 361. kàẃ káó Cake, biscuit 362. ?kàsà ʨɥí kàsàŋʧ̀ʷì Insinuation; indirect, allusive speech 363. ʨɪńaḿ́̃ kéꜝnáŋ́ Fried fish 364. kente kèǹté Kente cloth, hand-woven cloth; a length of this cloth 365. kete kètè A type of music 366. ?kɛǹ̀̃ tɛ ŋ́̃ ́ kɛǹ̀̃ tɛŋ́̃ ́ Basket 367. kiŋka ŋŋ kìŋk̀àŋŋ̀,̀ kiŋ̀̃ k̀aà̃ ̀̃ Fine, rich esp. of women’s clothing, gorgeously aparrelled 368. kɪt́ɪḱɪt́ ɪ ́ kítíkítí Small 369. ŋk̀rà́ń ꜝtɪ ́ klà́ńtè Cutlass, machete 370. krobó w klɔ̀b̀ɔó́, klɔ̀b̀ɔɔ́,́ klɔ̀b̀ɔ ́ White clay, smeared on the body by the priests, priestesses, and by ordinary people during certain rites; a perfumed ointment, made with white clay 371. kʊ́ kó Wandering 372. ɔk̀ʊb̀ ɔ ́ꜝ fʊ́ kòbɔf́ò Vagrant 373. kòòha á̃ ̀̃ kòháá, kòháŋ ́(no A large bird nasals) 374. koʥo besia kòʤóbɛs̀íá An effeminate man 375. kóꜝkŕó kóꜝkĺó An old palm wine 376. kokon te kòkòǹté Dried cassava flour; thick porridge made from cassava flour 377. kóniì̃ ́ ̃ kóniŋ̀̃ ŋ̀,̀ kóniì̃ ̀̃ Very quiet 378. koŋko hɪḿ̃ a ́̃ koŋkohema (no Chief fish-monger transcription) 379. kòŋk̀ò hɪńɪ ́ koŋkohene Chief fish-monger 380. konto mmere kòǹtómlè́, kòǹtómrè́ Leaf of the cocoyam, a spinach-like University of Ghana http://ugspace.ug.edu.gh 286 vegetable 381. kontro mfi kòǹtrò̀ɱ̀fì (no A large monkey or ape, chimpanzee transcription) 382. kóm̀m̀ , kóḿḿ kóŋŋ́ ́ Quiet 383. akɔfra /akɔfla kòòfɛŋ́́ A chief’s horn 384. kòóꜝk ó kòóꜝkóó Haemorrhoids, piles 385. kòóꜝk ó kòóꜝkóó A type of cocoyam 386. kòsɛ ̀ kòsɛ ̀ Excl. showing disgust, surprise, refusal to act as told 387. kɔk̀ɔ ̀ kɔk̀ɔ ̀ Warning, admonition 388. kɔk̀ɔb̀ ʊ ́ kɔk̀ɔb̀ó Dwarf mongoose 389. kɔkɔtɛ (Fn) kɔk̀ɔt̀ɛ ̀ Sea catfish 390. kɔntɔŋ kɔǹ̀tɔ,́ kɔŋ̀t̀ɔ ́ Curve, semi-circle 391. ŋk̀ɔŋ́̃ ḱ ɔŋ́̃ ́ kɔŋ́̃ k̀ɔŋ́̃ ̀ A dry, chesty cough; whooping cough 392. ŋk̀ɔŋ̀̃ k̀ ɔǹ̀̃ sa ́̃ kɔŋ̀k̀ɔŋ̀s̀a ́̃ Gossip, betrayal 393. kɔɔ̀ ̀ kɔɔ̀ ̀ Bright red colour 394. kòòʥ ań́̃ kɔɔ̀ʤ̀áń (no nasal) Red mourning cloth 395. ?kɔns ɔŋkɔnsɔn kɔs̀ɔŋ̀k̀ɔśɔ ́ Chain, neck chain; necklace 396. kɔt̀ɪ ́ kɔt̀í Penis 397. ?kɔtob onsam kɔt̀òbòǹsàḿ Unconsciousness 398. kɔt̀ɔ́ꜝ k ɔ ́ kɔt̀ɔḱɔ ́ The crested porcupine 399. ɛpaɪ kp͡ àí A flexible rush basket 400. pàpà k͡pàkp͡ à Fan 401. pápá kp͡ ák͡pá Good; genuine, right, correct; common, normal 402. ?pa (s elect) k͡páꜝk͡pĺá Acquire, buy a large amount (Semantic shift) 403. paḿ́̃ kp͡ aŋ̀̃ ,̀ paŋ̀̃ ̀ Discuss, deliberate, plan, think about 404. ɔp̀àpʊ ́ kp͡ àk͡pó Male goat, he-goat 405. pra k͡pĺa Herd, bring (sheep, cattle); clean off, clear off, be clean; be cleared from 406. pràm (quick) k͡plà̀k͡plà̀ Unexpectedly; urgent, immediate 407. pàtá k͡pat́̃ a ̀̃ Make peace, settle a dispute, propitiate, pacify; destroy, annihilate; be destroyed 408. apata kp͡ àtá Shed, shelter 409. pátaa kp͡ átáá Open, flat, bare expanse 410. ?pæ̀tú k͡pátù Act deliberately University of Ghana http://ugspace.ug.edu.gh 287 411. péŋ kp͡ éŋ ́ Brittle, break easily 412. pesú kp͡ èsú A basket, for keeping fowls 413. mpɛṕ ɛw kp͡ ɛḱ͡pɛó́ Remains of the harvest 414. pɪǹɪ ́ k͡plɛ̀ ́̃ Admit, acknowledge, agree to, approve; get along, concede, cooperate, behave in a social, non- individualistic manner 415. pɛm̀̀ k͡pɛŋ́ŋ́ ́ Hard, firmly 416. ɛp̀rà́ kp͡ ĺá An animal, giant ground pangolin 417. ɛp̀ʊ́ kp͡ ó A cloth measure, 12 yards; volume, tome 418. ?popo roku kp͡ òk͡plò̀kú New corn 419. pʊ̀pʊ ́ k͡pók͡pò Shake; shiver 420. ?m̀pó pò (a corn k͡pòk͡pòǹsù Corn porridge, made from unstrained dish)+ nsu dough 421. m̀pók ᶣíá kp͡ òkùá, ŋk͡pòkùá Young breasts 422. ?pòkú pòkú k͡pòkúk͡pòkú A large black and white bird (egret) 423. pʊ́ k͡pòô, k͡pòó Reject, refuse to accept; disown 424. ɛpɔẃ kp͡ ɔ ́ Lump, bump; knot, knot in a mesh; joint; misunderstanding, disagreement; unclear point, as in a discourse or discussion; locality; a plant 425. ?pɔʨ̀ɪ ́ k͡pɔḱè Break off a piece, take a piece of, snatch 426. pɔtɔ,́ pɔtɔẃ k͡pɔt́ɔ Rot, pollute; mash, knead, soften; mill flour 427. pɔt̀ɔɔ̀ ̀ k͡pɔt́ɔḱ͡pɔt̀ɔ,̀ k͡pɔt̀ɔɔ̀ ̀ Dirty, esp. of water, untidy, unclear, muddy 428. mpɔtɔ mpɔtɔ k͡pɔt̀ɔm̀̀k͡pɔt́ɔ ́ Porridge of mashed cooked root vegetable, mixed with oil and condiments 429. ɔkŕá kŕa, kĺa Soul, aspect of the human person that is believed to join God after death; page, young who carries the stool of a chief and sits in front of him at gatherings. 430. kramo (Mande- krà̀mò, klà̀mò Muslim karamɔgɔ) 431. krana a kŕań̃ aá̃ ,́̃ kĺań̃ aá̃ ́̃ Quiet, silent, without activity 432. ?æʨir i (back, krì̀ Dizziness, dizzy spells; details University of Ghana http://ugspace.ug.edu.gh 288 behind) 433. kronᶣ ia kĺónùá A kind of sandals with wooden sole and raffia toe strings (Obsolesc) 434. ŋk̀ùrò fʊ ́ klò́ŋf́ó townsmen 435. kuroti a (entrance klò́tíá Turn-off, turning point, junction to a town) 436. ?kŕʊ̃́ŋ ḱŕʊ̃ŋ́ ́ kŕɔŋ́̃ ḱŕɔŋ́̃ ,́ kĺɔŋ́̃ ḱĺɔŋ́̃ ́ Holy 437. krɔŋŋ kŕɔŋ́ŋ́ ́ Pure, unadulterated 438. kukur udu kùkrù̀dù, kùklù̀dù Earthquake, earth tremor 439. ?kuku kúꜝkwɛí́ Pot, cooking pot 440. kũma á̃ ́̃ kũma ́̃ Junior 441. kùma á̃ ś̃ ɪ ́ kù̃maá̃ ś̃ é The city of Kumasi; small coloured beads 442. kurun ku (Fn) kũŋꜝḱũḿ, kũŋ́ꜝ kũŋ ́ A kind of music 443. ŋkuni m kùníḿ Success, victory 444. ?kunt ann kuǹ̀̃ taŋ̀̃ ŋ̀,̀ kuŋ̃̀ k̀aŋ̀̃ ŋ̀,̀ Pompous, of dressing; unsuitably kɛǹ̀̃ ʃɛŋ̀̃ ŋ̀,̀ kũǹ̀ ʧɛŋ̀̃ ŋ̀ ̀ expensively dressed, overdressed 445. kʊ̀rá kùrá, kùlá Preserve, look after, manage, control 446. kʊ̀ràfa ́̃ kùràfa ́̃ Sister-in-law or brother-in-law 447. kʷaɲʨ ɪrɛ, kʷààʃɛŕí, kʷààʃɛĺí Happiness, long life ??ŋkwaʨɛrɪ 448. ?kusu u kùsùù Sombre, drab, dull; thick, overgrown 449. kóńtó ḿpóꜝfʊ́ kútúḿpɔf́o, Liar làblà̀báfo 450. konto mpo kútúḿpɔ,́ làblà̀bá Lie, counterfeit 451. ?kúẃ kúú Group, council, society 452. kʷáꜝɪ ́ kʷáe Thick forest 453. ?ɔk̀ʷæ̀ kú kʷàkúó Mona monkey 454. ?kʊŕ á á kʷláá, kʷráá, kúláá Intens. All, at all, completely, entirely 455. kʷàm aŋ́̃ ́ kʷàmàŋ̀ Crowd, community, the commonality (populace) 456. ?kʷaw kʷàò, kʷàà Smear clay on a surface e.g. a fireplace, to polish it 457. ɔk̀ʷàs ɪà́ kʷàʃìá Stupid person 458. kʷàtá kʷàtá Leprosy 459. m̀màá má Bamboo clappers, for beating rhythm while singing University of Ghana http://ugspace.ug.edu.gh 289 460. abaaw a àbààwá, mààwá Maid, female servant 461. ɔmam pam maḿ́̃ ꜝpáŋ ́ Land crocodile, monitor lizard 462. mamp robi mam̀̀̃ prò̀bí, abɛntia Royal cemetry 463. maɲam aɲa maɲ̀̃ am̀̃ aɲ̀̃ a ̀̃ Disorderly, mixed up, chaotic, confused 464. ɔm̀aŋ́̃ ́ maŋ̀̃ ̀ Town, country, state 465. ḿmaŋ̀̃ ́ maŋ̀̃ ̀ Herring, sardine 466. mań́̃ ꜝs ʊ ́̃ mań́̃ sɔ ̃ Quarrel, strife, ill-feeling; litigation 467. mantɪ asɪ màŋt̀ìàsé Central town, capital town 468. m̀mìr é mlɛ̀,́ mrɛ̀ ́ Mushroom 469. ɱ̀fʊá̀ ɱ̀fóá Pimple 470. ?ɱ̀fòn íní ɱ̀fònírí, ɱ̀fòníí Picture, photograph 471. mià̃ ́̃ miá̃ ̀̃ Squeeze, pinch 472. m̀màr á mlà́ Rule, rules, law 473. ḿmó mó, omomoomo Excl. congratulations! 474. ḿmòḿ mòŋ ́ Rather 475. m̀mɔb́ ɔ ́ mɔb́ɔ ̀ Pity 476. m̀mɔb̀ ɔm̀mɔb́ɔ ́ mɔb́ɔm̀ɔb́ɔ ̀ Very pitiful 477. m̀mɔd́ ɪŋ́̃ ́ mɔd́̃ ɛŋ́ ́ Effort, esp. successful or laudable effort 478. ? m̀m ɔd́ɪḿ́̃ mɔ ́ mɔd́̃ ɛŋ́b́ɔɔ̀ ̀ Effort, achievement, success 479. mpɪna bedi m̀pɛǹab́̃ ɛd́í Name of a variety of cassava 480. mpɪna ba m̀plàb́̃ í, hemɔbi Love child, offspring of a love affair 481. m̀pʊ̀á m̀pòá Challenge (non-competitive, task) 482. m̀pʊ̀à ʨɥá m̀pòáʧʷàà Challenging 483. mmʷʊ awa múáwá Germ, virus 484. mù̃na ́̃ mũǹ á Be moody 485. m̀mùs ú mũ̀sú Ill luck, abomination, evil, sacrilege 486. ?m̀mù sùbɔ ́ mù̃súbɔɔ̀ ̀ Sacrilege 487. ?bùsù ɛf́ʊ́ mùsùɛf́ó Accursed person, bringer of bad luck 488. ?mṹ mũ̀u ̃̀ All, whole, full 489. ?nan naǹ̃ è Leg, including foot 490. ?nantu naǹ̀̃ tú Calf of the leg 491. naŋka fʊ (No Ak. naŋ̀̃ k̀áꜝfó Gossip, blasphemer word) University of Ghana http://ugspace.ug.edu.gh 290 492. ǹnòm̀ múḿ nòḿ Slave, servant 493. nsama wa (no Ak. ǹsàmàwá Bad body odour (obsolesce.) word) 494. nsra ǹsrà́, ǹslà́ Camp; marching, parade; siege 495. nsaɕɥ a (no Ak. ǹsàʃùà Fisherman’s soup, fish and pepper word) boiled in sea water which is then discarded; casually prepared soup with fresh fish 496. nsufra ma (Fn) ǹsùfrà̀má Light, showery raindrops blown by the wind 497. nsùkò ókó ǹsùkòóꜝkó A plant, dumb cane 498. asutɪŋ ǹsùtɛŋ́̃ ́ Waterfall 499. ǹtá ǹtá Double 500. ǹtam̀̃ a ɡ̀̃ ó ǹtàmàɡó A person’s daily work clothes, put in his coffin when he dies 501. taǹ̃ naà ̃ ́̃ ǹtànàí An area which has been partially burnt over to clear it; heap of dry farm trash, left from clearing 502. ntajaa ǹtájàá Roofing tiles 503. nteme fua ǹtèmèfúà Mashed plantain mixed with palm oil 504. tìá ǹtìá Kick 505. ntobro ǹtòbĺó, ǹtòbŕó Measles 506. ǹtɔḿ́m ɛ ́ ǹtɔḿɛ,̀ ǹtɔb́ɛ ̀ A kind of shrub, used to demarcate indicate a boundary 507. eʨím ǹʧíŋ ́ Blood soup 508. ǹʨɥʊm̀̃ a ́̃ ǹʧù̃ma ́̃ Red clay 509. ntumũ rúm ǹtúmlṹꜝmṹ An herb 510. ǹtũm̀m ʊ́á ǹtúmúá A tiny bitting fly, sunfly 511. ònípá (human ꜝnúk̃ ͡pá, ꜝŋḱ͡pá Senior, elder (Semantic shift) being) 512. ɲaà̃ ̀̃ ɲaà̃ ̀̃ Slowly 513. oɲaḿ́̃ ɲaḿ́̃ Praise, glory, enhancement, honour 514. ɲ̀ɲaà̃ ǹ̃ ɪ ́̃ ɲaǹ̃ é A type of bead 515. ɲaŋ̀̃ k̀ʊ ǹ̃̀ tɔŋ́̃ ́ ɲaŋ̀̃ k̀ùǹtɔŋ́̃ ́ Rainbow; a deity, associated with the rainbow 516. ɲ̀ɲèɲ̀ɲ èɲ̀ʨɥí ɲɛɲ̀̃ ɛŋ̀̃ ʧ̀ʷí Loose conduct, misconduct, profligacy, incl. sexual 517. ɲ̀ɲiǹ̃ a m̀̀̃ ɲiǹ̃ am̀̀̃ ɲràm̀̀̃ ɲràm̀̀̃ Brightly, sparkling, sending out rays 518. ŋk̀aḿ́̃ fʊ́ ŋk̀aḿ́̃ fó Name of a variety of cassava University of Ghana http://ugspace.ug.edu.gh 291 519. ŋkann arɪ ŋk̀áꜝ̃ nalé́̃ Rust 520. ŋkaǹ̃ í ŋk̀aǹ̃ í Name of a variety of yam 521. ŋkan (Fn) ŋḱàŋ,̀ ŋ́ꜝ káńꜝtíá, Weakfish or croaker, cassava fish ŋkaŋpu 522. ŋkatɪb ɛ ŋk̀àtìbɛ ́ Soup made with groundnut paste and palm-nut juice 523. ŋk̀áꜝtɪ ɛ́ ́ ŋk̀áꜝtíɛ,́ ŋk̀ɛ́ꜝ tíɛ ́ Groundnut, peanut 524. ŋk̀àtɪk̀ òǹtó ŋk̀àtìkòǹtó Soup made from groundnut paste and cocoyam leaves 525. ŋk̀ŕák ŕá ŋk̀ĺákĺá Light (unthickened) soup 526. ŋkran ŋk̀ĺꜝáń Army ant 527. ŋkɔ(fi e) ŋk̀ɔ ́ Yam mould, mould in which yam is planted 528. ŋk̀ɔf̀ìb ɔ ́ ŋk̀ɔb́ɔɔ̀ ̀ Making yam moulds 529. ŋk̀ɔm̀ ɔ ́ ŋk̀ɔm̀ɔ ́ Problem, worry, lament 530. m̀páꜝɪ ́ ŋ͡m̀ k͡páì Libation 531. mpok uwa ŋ͡m̀ k͡pòkùá Breast of a young girl; nipple 532. ŋk̀ʊrɔ m ŋk̀rɔŋ́ʃ́ímɔ ̀̃ Snoring 533. ŋk̀ú ŋk̀ú Shea butter 534. kusuk um (Fn) ŋk̀ùs̃ ũk̀ ù̃ḿ A musical style, of Fante origin 535. ǹsɛǹíá ǹsɛǹíì Scales, balance 536. ǹsékú ŋʃ̀ékú, ǹʃékú, òʃékú Back-bitting, slander 537. ?ǹsùp ɔ ́ ŋʃ̀ɔk̀͡pɔ ́ Island 538. ɔt̀áŋ ́ ŋt̀áŋ,́ ǹtáŋ ́ Carrying net, sack; fishing net; web, cobweb 539. ǹtɔŋ́ ́ ŋt̀ɔŋ́ ́ A tree, screw pine 540. wawa ʥan (Fn) ŋʷaŋ̀̃ ̫̀ aʤ̀̃ àŋ,̀ Hair-tail or ribbon fish hʷah̀̃ ʷaʤ̀̃ àŋ ̀ 541. ɔb̀rà́ òblà́ Life; a puberty custom for girls; fashion, fashionable swagger 542. ʥìná òbɛʤ̀ìna ́̃ A customary rite performed at naming ceremonies; a greeting to a newborn baby 543. mpint im òbiǹ̀̃ tíḿ, òbiǹ̀̃ tiŋ́̃ ́ A kind of drum; a style of drumming 544. ɔb̀rà́ꜝf ʊ́ òblà́fó Executioner 545. ɔbraŋ òblà̀ŋ ́ Youthful, vigorous person; strongly built person 546. obere ku òblè̀kú A bird, Senegal coucal 547. opere nten òblɛ̀ǹ̀tɛŋ́ ́ A drum, made in pairs called “male” University of Ghana http://ugspace.ug.edu.gh 292 and “female” 548. ɔbiriʨ i òblì̀ʧí A river fish 549. ɔbʊsʊ masɪ òbósɔŋ́ásé A shrine house 550. ɔb̀ʊꜝ́ sʊ ́̃ḿ òbósúḿ Fetish, spirit (obsolesce.) 551. ɔbʊsʊ mfʊ òbòsùm̀fó Title for the priest of certain deities incl. Tigali, Akɔŋ, Kpele 552. ɔbɔad ɪ òbɔá̀dé Creation of the world, origin; tradition 553. ɔbɔŋk ɔ ́ òbɔŋ̀̃ k̀ɔ ́ River prawn 554. òbrò̀n íbèdíàwùò òblò̀níbɛd̀íàwùɔ,̀ Name of a variety of cassava òbrò̀níbɛd̀íàwùɔ ̀ 555. òbúbú áꜝfʊ́ òbúbúáfò Cripple, paralysed person 556. ɔd̀àbɔ ́ odabɔ A type of antelope, Bay duiker 557. ɔd̀áꜝsɪ ́ òdásè Testimony 558. ɔdasɪf ʊ òdàsèfóɲò Witness 559. ɔd̀ɪ ́ꜝ ɕɪ ́ òdéhè, òdéè Noble, free born person 560. odi am rɔ òdémŕɔ,́ òdémĺɔ ́ Boiled fresh corn on the cob 561. ɔd̀ɛɲ́́ꜝʨ ɛḿ́ òdɛń́ʧɛ ̀ A small type of crocodile 562. adí-àm mã-wʊ́-ba odiamaraba, afaseo Water-yam 563. ɔdʊ́dʊ ̀bɛŋ̀ ̀ òdòdòbɛŋ́̃ ́ Reed; tube, usu. Bamboo, used in tapping palm wine 564. odoɛ, odoe (Fn) ódóì, ódóè A fish, barracuda when small 565. òdòǹn ó (Mande òdònó A small double-headed pressure dunu) drum, the heads joined by strings 566. ɔd̀ɔŋ̀k̀ ɔ ́ òdɔŋ̀k̀ɔ ́ Slave; derogatory term for stranger, foreigner esp. originating from northern Ghana; facial scarifications 567. odúku òdúkù Barrel 568. odum òdúḿ, òdúŋ ́ A tree; its timber 569. ɛf̀ʊ́ òfó White-thighed Colobus monkey, a variety of “hɔi” 570. ɔf̀rɔ̀t̀ɪ ́ òflò́tè A kind of antelope 571. ɔh̀á òhá Hundred 572. ɔhɪnɪb andi (no òhénébàǹdí A fish, the ten-pounder Ak. word) 573. ɛɕ̀ɪ-́dᶣ ìà òhɛ ́ Gum copal; gum copal tree; a bead, of gum copal; an incense, believed to ward off evil spirits when it is burnt 574. òhià̃ ́̃ òhià̃ ́̃ Poverty, need University of Ghana http://ugspace.ug.edu.gh 293 575. òhià̃ ́ꜝ̃ f ʊ́ òhià̃ ̃́ꜝ fó Poor person, pauper 576. ɔhɪńɪb a òhíníbá Traditional (Ashanti style) sandals 577. oʥɥir á òʤʷrà́, òʤil̀̃ a ́̃ A festival, celebrated by the Otublohum and Ngleshi Alata sections of Accra 578. kɛs̀ɪ ́ òkɛs̀é Fashion, chic, prestigious manner 579. ɔk̀ɔŕɪ ́ òkɔl̀è, kɔɔle eagle 580. ɔkɔmf ʊ ́ òkɔm̀̀fó (traditional) priest 581. ɔkɔmf ʊwa òkɔŋ̀fówá Traditional priest in training 582. ɔk̀ɔ́ꜝ tɔ ́ òkɔt́ɔ ̀ River crab, Potamon 583. ɔprɛm òkp͡ lɛ̀m̀́, òk͡plɛ̀ŋ̀ ́ Cannon, bomb, explosive 584. ɔp̀ʊ́ŋ ́ òkp͡ lɔ̀ ́̃ Table 585. ɔpɔŋk ɔ (Mande òkp͡ ɔŋ̀̃ ɔ ́̃ Horse punu) 586. ɔkʊ̃rʊ pɔŋ òklò̀pɔŋ̀,̀ òkrò̀pɔŋ̀ ̀ Eagle 587. òkúkɔ ḿ́ òkùkóḿ Name of a variety of cassava 588. ɔkúku bàŋ ́(a òkúkúꜝbá Name of a small animal squirrel) 589. kũna ́̃ òkùl̃ a ̀̃ The state of being widowed; a funeral custom, performed by a bereaved spouse 590. ɔk̀ũ̀na f̀̃ ʊ́ òkù̃laf̀̃ ó Widow, widower 591. ɔkum abɔmɔfʊ òkúmàbòmòfó A large variety of cassava 592. ɔkum bɔmmɔfʊɔ òkúḿbɔb̀ɔf̀óɔ ́ A very poisonous green snake, the green mamba 593. òkúá, òkᶣíá òkʷàá Farming, agriculture 594. òkᶣìàf ʊ́ òkʷààfóɲò Farmer 595. ɔm̀aŋ́̃ f́ ʊ ́ òmaǹ̀̃ fó Relative, countryman 596. ɔmaŋʥ ina òmań́̃ jĺa,̀̃ òmɛí́̃ jĺa ̀̃ A shrine, either made of cement, similar to otutu, or consisting of trees surrounded by a wall 597. ɛm̀ʊ́ ̃ òmɔ ́̃ Rice; small blue and white waist beads, for babies 598. àpààfʊ ́ òpàáꜝfó Day-labourer 599. opete òpéꜝté Vulture; soldier fish, squirrel fish 600. òpɛh̀é nádì òpɛh̀énádì Morning star 601. pòkúp òkú (Fn) òpòkú, òkp͡ òkú A fish, frigate mackerel 602. òpúró ẃ òpĺòŋ ̀ Any red squirrel, also the red-headed forest squirrel University of Ghana http://ugspace.ug.edu.gh 294 603. ɔs̀ábɪŋ̀̃ ̀ òsáábɛŋ́,́ sábɛŋ́́ Lightening, discolouration of skin due to bleaching, burns, etc. 604. ɔs̀àɪ ́ òsàí Eunuch 605. ɔs̀áńsá ní òsáńsá Apathy 606. ɔs̀áǹsá òsáŋśá, òsáḿsá A large bird, the black kite 607. ɔs̀ɛśɛ ́ òsɛśɛ,́ sɛsɛo A large tree 608. òsé òʃé Acclamation 609. osii òʃìí Limp 610. ?òsíʥ á òʃíʤá State of being unmarried; spinister, bachelor 611. ?òsíʥ áꜝfʊ́ òʃíʤáꜝfó An unmarried person of either sex 612. sísìbɪŕ ɪ ́ òʃíʃíbĺíìʃì An animal, the honey rattel; bear 613. ?òsóò dé òsóòdé Name of a type of music 614. ɔs̀ɔ́ꜝ fʊ ́ òsɔf́ò Christian priest, minister; title for a reverend minister 615. súpur upù òsúpŕúꜝpú River turtle 616. ɔt̀ɛń̀tɛ ́ òtɛń́tɛ ́ A kind of drum 617. ètí òtí Main point, as in a speech, a book 618. òtià́̃ ꜝfʊ ́ òtìàfó Person sitting side-by-side, abreast with another, esp. in a canoe 619. tírídàà sɪ ̀ ótírídààsè Response to greeting “tirinkwa” 620. tíríŋꜝk ʷa ́̃ ótíríŋꜝkʷa ́̃ A greeting, of congratulation on safe arrival, safe delivery, escape from accident, death, etc. 621. ɔtoko (a)akatá otokotaka (no A small spreading tree or large shrub transcript.) 622. ɔt̀ɔ ́ òtɔ ́ A food, of mashed yam, plantain or cocoyam mixed with palm oil 623. ɔt̀ɔ́ꜝ fʊ ́ òtɔ́ꜝ fó Ghost of one who dies unnaturally, ie. Violently, in an accident, unexpectedly, or young 624. ɔʨ̀ɪà̃ ḿ̃ ɪ ́̃ òʧàámí, òʧàámɛ ́ Linguist, spokesman for a chief 625. ɔʨɪrɪb ɛŋ̃ òʧérébɛŋ́,́ òʧíríbɛŋ́ ́ Any of the several kinds of green snake; fierce snake, green-lined snake, emerald snake, green mamba; also African Beauty snake, striped form 626. òʨɥáǹ sa ́̃ òʧʷàǹsa ́̃ Drunkard 627. ɛʨɥiʨ ɥɛkrɔbɔ òʧʷìʧʷákrɔ̀b̀ɔɔ́ ́ A fish, blanquillo (Fn) 628. ?ɔɥ̀ɪŕɛ ́ òwélè Revenge, vengeance University of Ghana http://ugspace.ug.edu.gh 295 629. ?àɥɪr̀ɛ t̀ɔ ́ òwélètɔɔ̀ ̀ Avenging 630. oɥiak ɔtɔ òwìákɔt̀ɔ ̀ Fish eagle, palm nut vulture 631. ɔɥ̀ìrèŋ f̀ʊ ́ òwìrèŋf̀ó, òwìŋf̀ó Sinner, breaker of taboo 632. òwúrà ówúlà, ówúrà Title for a gentleman, Mister, sir; gentleman, man of substance 633. ɔj̀ɔ ́ ójɔ ́ One good turn deserves another (obsolesce.) 634. paa sʊ pàásó Add, do in addition 635. pàɪ ́ pàá, pàé, pàí Call upon, call upon by an appellation or nick-name, mention 636. páɪ ́ pàí Be upset, be angry at sb. 637. pæʥa pàʤá Lift with both hands, lift with effort (obsolesce.) 638. m̀pam̀̀̃ pam̀́̃ pam̀̀̃ pam̀́̃ Crown of the head 639. pamp amsuadɪ pam̀̀̃ paŋ̀̃ śúádé, A type of fishing net paprasuade 640. pæ̀̃m̀p íḿ pam̀̀̃ pí Trap, snare; obstruction to progress 641. pàpà pàpà, k͡pàk͡pà Fan 642. mpʊ̀p àhʊ̃́ pap̀̃ aŋ̀̃ ,́ pap̀̃ aù́̃ Towel 643. àpásá (falsehood, pásá Tell lies; gossip lie) 644. batakæ rí pàtàkàlí Smock of hand-woven cotton, as made in northern Ghana; war smock 645. pàtæ̀k ú pàtàkú An animal, the spotted hyena, a variety of “klaŋ” 646. pæ̀tú pàtú A bird, the owl; the unexpected, sudden 647. ?pefee (or Eng. péféé Perfectly, plainly, exactly; perfect, 'perfect’) plain 648. pɪŕɪ ́ plè́ Make for, struggle for; scramble, struggle, quarrel over 649. pesú p pèsú Chicken coop 650. pé pɛ́ꜝ í Chisel 651. pɛs̀ɛm̀ ɪŋ́̃ ḱʊm̃́ ɪɲ́̃ a ́̃ pɛs̀ɛk̀úméɲá Selfishness 652. pì pì Indeed 653. pià́̃ pià̃ ́̃ Push from below 654. pìá pìá Bedroom, bed chamber 655. pii pìì Many 656. pìrá pìlá Hurt, injure; get hurt University of Ghana http://ugspace.ug.edu.gh 296 657. pené ( groan with pií̃ ̀̃ Suffer; torture, cause to suffer, pain) mentally or physically 658. prà́ád úrù plà́ádúrù An erect herb 659. pràm plàm̀̃ a ́̃ Courtyard 660. pó, ?? m̀pó pó Very, even, at all, indeed 661. pùé púè Appear, be out, manifest; appearance, appearing 662. sà sà Syringe, purge, administer an enema 663. sà sà Castrate 664. sá sá A cylindrical implement used in making fishing nets, to regulate the size of the mesh 665. ɔs̀á sá Access road or path, to a farm or hunting grounds 666. (n)sas̀̃ ɛŋ́̃ ́ sààsɛ ́̃ A pot for collecting palm wine 667. ǹsab̀̃ r àŋ̀̃ ́ sàblà,́̃ sàbrà ́̃ Appellation, nickname; a formal appellation corresponding to one’s lineage or asafo name 668. sakasa ka sàkàsàkà Disorder, roughness 669. sãm ( figures sàma ́̃ Designs, painted on canoes, clothing, made by cutting etc. in relief) 670. saman adzɪ (Fn) saḿ̃ aà̃ d̀̃ é, sã́ꜝ maá̃ d́̃ é The world of ghosts (hell) 671. sæ̀mín á (Arab. sàmaĺ̃ a,̀̃ sam̀̀̃ la ̀̃ Soap sabun) 672. sæʨim adɪ (Fn) saŋ́̃ ḱũmad́̃ é A fish, small sea-perch 673. àsáńtŕ òfí sáńtŕófí A bird, the long-tailed nightjar 674. sáɲaà̃ ́̃ sáɲaà̃ ́̃ A metal; silver 675. ?sànɪ ́̃ saŋ́̃ è Be relieved, feel better 676. ɔs̀aŋ̀̃ k̀ ú saŋ̀̃ k̀ṹ Any keyed or stringed musical instrument: thumb piano, organ, piano, accordion, guitar, violin, harp 677. asapa até sàpàté, sàpàtí Assistant to a traditional priest or priestess; leader of a religious group; member of a group, peer, righthand associate 678. sra slà́, srà́ Visit with a purpose; give the second round of drinks at a customary function 679. sárɪśʊ ́ bàɲ̀ʨɪ ̀ sàrésò-bàǹʧí A wild, inedible kind of cassava University of Ghana http://ugspace.ug.edu.gh 297 680. ɔs̀átàd ᶣíàà sátáꜝdúá, tátádùá A shrub, with a tough flexible wood; name of a variety of cassava 681. ɔs̀àɥè é sàwìé Sponge chewed and dried 682. sɛʨ́ɛ sɛḱɛ ́ anchor 683. sɛŋ́̃ ̀ sɛŋ̀̃ ̀ Hang; kill by hanging or crucifixion 684. ɔsɛŋ́̃ ́ sɛŋ́̃ ́ Pot 685. sadé ʃàdé Blessing, luck, good fortune; destiny, fate 686. sàdɪb́ʊ á́ ʃàdébóá An animal, hedgehog or brush-tailed porcupine 687. sià́̃ w ʃáó Cluster, bunch e.g. of bananas 688. ?aɕɥɪr ɪw ʃɛ ́̃ Sugar cane 689. sika ʃìká Money; precious metal 690. sikasi ka (Fn) ʃìkáʃìká Fish, golden-headed sea bream 691. sìw ʃìù Suppress; replace, displace; be bogged down, get stuck; sink 692. ɕɥɛd́ɪ ́ ʃʷɛd́é binoculars 693. ɕɥɛɕɥ ɛ ́ ʃʷɛʃ́ʷɛ ́ A fresh water fish 694. ɕɥie ʃʷìê, ʃʷìé Spill, pour something; fall, pour; to start, commence; set out for a destionation; put into a sequence, etc. 695. ɕɥíḿ ʃʷíḿ Fast, in a flash 696. ɛs̀a ́̃ sísà Ghost; shade of one who died a natural death, in sickness or old age 697. sò só Hit the back of an opponent’s hand in a boy’s game 698. sʊ̀pá sópà Insult 699. sʊ̀nʊ̀ sòlò, sòrò, sòlè, sèlè Differentiate; be different 700. ɛs̀ʊ̃́ sɔ ́̃ Blame; responsibility 701. sɔr̀ɪ ́ sɔl̀è Worship, pray to; baptize 702. sũḿ sɔḿ̃ ɔ,́̃ sũmɔ ́̃ Work for someone as a live-in house servant; serve, work to repay someone’s debt to the employer 703. sɔẁ sɔɔ̀ ̀ Catch from below; catch, with a flourish; accept, receive, appreciate 704. sʊ́sɔẃ́ sɔśɔó́ A digging stick, with a wooden handle and metal blade 705. srawa sŕꜝáwá, srầwá, Lightening; electricity sĺꜝáwá 706. sʊrʊŋ kʊ srò̀ŋk̀ṹ Extraordinary (obsolesce.) 707. èsú sú Nature, character, colour, appearance University of Ghana http://ugspace.ug.edu.gh 298 708. sʊ̀á sùá Carry a heavy load (obsolesce.) 709. sùbáŋ ́ súbáŋ ́ Nature, character, habit 710. àsùmà ǹfʊ́ sùmaŋ̀̃ f̀ó A sea fish (Semantic shift?) (owners of charms) 711. sumpi (Port. sù̃mũ̀i ́̃ A metal, lead; lead weights for fishing chumbo) nets 712. sũm sũné Pillow 713. sùsúẃ súsù Measure, calculate; think, imagine, assume 714. sũńsũ ḿ sùsùma ́̃ Spirit, soul, shade; subconscious mind, out-of-body experience 715. nsú sùũ ̃̀ Drown, submerge (semantic shift) 716. tàà táábó Gizzard 717. káfrà̀ʨ ɛ ̀(Haus. tàfĺàʧɛ ̀ Excuse me, pardon me gafara) 718. taŋ̀̃ ̀ taŋ́̃ ́ Impurity, nastiness 719. ǹtáḿp é táàŋk̀͡pèé Sisal, a beach plant; a long heavy rope or chain 720. taŋ̀̃ t̀aŋ́̃ ́ taŋ́̃ t́aŋ́̃ ́ Impurities 721. tààtáá tàtáá, tàtŕá With hesitant steps, as of children learning to walk 722. ?tápòr í tàpòlí A small wooden hourglass shaped implement for grinding vegetables, used with ‘ka’, grinding bowl 723. ?tɪt̀ɪt̀ɪt̀ ɪ ̀ tètèètè Time immemorial 724. tɛkrɛm a ́̃ (tongue) tɛḱrɛ̀m̀á (no nasal) Cantor in “otu” singing (semantic shift) 725. ɔt̀ɪm̀̀̃ p ɔŋ̀̃ ́(main tɛḿ́pɔŋ́̃ ́ Path, way (slight semantic shift) road) 726. tɛẁ tɛò̀ Transplant, plant trees 727. tɛt̀ŕɛɛ́ ́ tɛt́ŕɛɛ́ ́ Flat, wide, broad 728. tɪ ̀(apɔ ) tì (apɔ) Muscular, well-built (restricted usage) 729. tià̃ ́̃ tiá̃ ̀̃ Save 730. tìàfi ́̃ tìàfí (no nasal) Toilet, latrine 731. tɪá̃ t́̃ ɪá̃ ́̃ tiá̃ t́̃ iá̃ ,́̃ tiá̃ á̃ ́̃ Skinny 732. tìpɛŋ́̃ ,́ tìpɛŋ̀̃ f̀ʊ́ tìpɛŋ́,́ tùpɛŋ́ ́(no Peer, equal, mate nasal) University of Ghana http://ugspace.ug.edu.gh 299 733. tit̀̃ i ́̃ tit̀̃ i ́̃ Scratch 734. titiriw (main, tíꜝtŕí Important, special, particular principal) 735. tiw tìù, tùù Follow, pursue, catch up with; bother about take an interest in, take trouble over 736. ntʊ̀kʊ ̀tá (?Haus. tòkòtá Sandals takalmi) 737. tòǹtó tòtó, tɔǹ̀̃ tɔ ́̃ Name of a variety of yam, that matures very slowly 738. túródó ó tŕódóó Plain, straightforward 739. tɔŋ̀̃ ŋ̀ ̀ tɔŋ̀̃ ŋ̀,̀ tɔɔ̀̃ ̀̃ Off, away 740. ǹtʊń́̃ tʊ ̃ḿ́ tɔŋ́̃ t́ɔŋ́̃ ,́ tɔŋ́̃ t́ŕɔ ́̃ Mosquito 741. ɛt̀ɔẃ́ ( lump) tɔó́, tɔ ́ bunch 742. tŕʊt́ ŕʊ ́ tŕótŕó Smooth and slippery 743. ʨèʨè kú (Fn) ʧèʧèkúʧèkú A fish, triggerfish 744. ʨɛ ́ ʧɛ ̀ Last, be long lasting; spend time, delay 745. àʨɪà̃ ́̃ ʧià̃ ́̃ Swagger 746. ʨɪń́̃ ʨɪ ń̃ ɪá̃ ʧíńʧíníí (no nasal) By all means, at all cost 747. ʨɪʨ́ɛ-̀ pɛ-̀àwàrɪ ́ ʧíʧɛṕɛá̀wàlé, The planet Venus ʧíʧɛṕɛá̀wàré 748. ʨɥɪm̀̃ ʧʷɛm̀̀ All together 749. ʨɥɪǹ̃ ɪ s̀̃ iŋ́̃ ́ ʧʷɛŋ̀ʃ̀iŋ́̃ ,́ ʧòìʃiŋ́̃ ́ A type of medium-sized drum; the music played on it 750. tᶣìá túà Become strong; attack; rape; punish, esp. by inserting hot spice into the anus; drink 751. tᶣìé túè Gush out 752. ǹtù̃ŋk̀ ũḿ tũŋ̀ ɡ̀ùḿ̃ , tuŋ̃̀ ɡ̀ũ̀ṹ New palm wine, palm wine from a freshly felled tree 753. tù̃m̀pa ŋ́̃ ́ tù̃ŋk̀͡pa ́̃ Bottle, liquor bottle 754. tù̃m̀m̀ tũŋ̀ ŋ̀ ̀ Dark, black 755. tùtúẃ tùtú Ache 756. ?tuǹ̀̃ tu ̀̃ḿ tùùtùù, tùù Darkness, dark of the evening 757. ɔẁæ̀d úrú wàdúrú Mortar, for pounding fufui 758. ɥiriɥi riwa (Fn) wìíꜝwíá, wìíjúꜝwá, A fish, golden-headed sea-bream, jùújúꜝwá when small 759. wʊ̀sʊ ́ wósò, hósò Shake University of Ghana http://ugspace.ug.edu.gh 300 760. wɔf̀à (mother’s wɔf̀à Uncle, father or mother’s brother (semantic extension) brother) 761. wɔf̀àà sɪ ́ wɔf̀ààsé Nephew, niece 762. ɥírídú ú wúrúdúú, wúlúdúú Cool; safe, secure; quiet; bland; fluent 763. jìjí jìjí Reason, remonstrate 764. jóma ̀̃ jóómà Camel University of Ghana http://ugspace.ug.edu.gh 301 Appendix 5: List of Akan loanwords in Dangme (Approximately 329 suspected Akan source entries in Dangme) NO. AKAN Dangme GLOSS 1. àpàplàǹ̀̃ sa ́̃ àbàblàsa,́̃ àpàplàsa,́̃ A food made of milled roasted àbɛb̀lɛs̀a ́̃ corn cooked in palm soup 2. àbàsá àbàsá Yard, arm-spread; fathom 3. àbààwá àbàwá, àblàwá Servant 4. ɛb̀ɛ ́ àbɛ ́ Proverb 5. àbɛb́ú àbɛb́ù Proverbial, formal language 6. abɛti àbɛt̀i;́̃ àbɛt̀í A very tall mature palm tree 7. abradɪ àblàdè, òdéhē A royal, noble; a respectable person 8. abaawa àblàwá, màwá Maid servant, also slave or servant of either sex 9. àbɪr̀ɪẁá àblè̀wá Old lady 10. abɪrɪwabɛwʊ àblèwábɛẃó A measure of drink, about a quarter of a pint. It refers to the bottle of Atwood’s medicine 11. ɔb̀ʊ̀ǹsaḿ́̃ àbòsáḿ, àbùsáḿ, Devil abosiá̃ ḿ́̃ , àbùsiá̃ ḿ́̃ , àbùsiá̃ ḿ̃ i,́̃ àbosiá̃ ḿ̃ i ́̃ 12. abɔntore (no àbɔntole Vine rubber transcript.) 13. Æ̀bùda ́̃ àbùda ̀̃ Foolish man; a hen-pecked man, a man who does women’s chores 14. ɔd̀àbɔ ́ àdàbɔ,́ òdàbɔ ́ Animal, kind of (antelope) 15. àdàɥɪr̀ɪk̀ɔs̀ɔ ̀ ādàwlikɔsɔ Iron bell 16. àdɪ ́ àdé Creation, the created world 17. àdɪ ́ꜝbɔ ́ ādébɔ ̀ Creation, nature 18. àdéŋ́ꜝ kũḿ àdekum A kind of music and dance, played in the coastal areas. 19. àdʊẁ á àdòwá A kind of music or dance form, played mainly by women in coastal areas. 20. àdʊ̀wá àdòwá royal antelope University of Ghana http://ugspace.ug.edu.gh 302 21. ??edufa (Fn) àdúfá A medicine, made by grinding herbs and formed into balls before drying 22. æ̀dìfùdɪ ́ àdùfùdé Windfall, sth. received without working for it; greed, wanting things cheap 23. æ̀dúòkú àdùòkú A nickname for the rat 24. ɔfa,́̃ àfa ́̃ àfa ́̃ soothsaying 25. wɔf̀ààsɪ ́(no āfas̀̃ é Matrilateral relative: sister’s child nasal) of a man; matrilateral cousin 26. àfàsɪẃ́ (Ewe?) àfàsé, àfàsó water yam 27. àɕɥɪɕ̀ɥɛ ́ àhìhʷíɛ,́ àhùhúɛ,́ Mirror àjìhúɛ ́ 28. æ̀hũḿ àhũmì Storm, gale 29. ??æ̀wúɲad́̃ ɪ ́ àhṹɲad́̃ é Sale by auction 30. æ̀ɥísiá̃ á̃ ́̃ àhũsa;́̃ àhu͂̃́sa ́̃ Orphan, orphanage 31. àjá àjá A man who is sponsoring a dipo candidate or has a fiancée among them; a masculine name 32. àɕɥɪŕɪẃ́ àhléú Sugarcane 33. æ̀ʥìna ́̃ àʤìna,́̃ àʤil̀̃ a,́̃ àʤrà,́̃ Meeting, private consultation, àʤlaà̃ ́̃ conference 34. æ̀ʥìnàmʊá̀ àʤlàmuá̃̀ Domestic cat 35. æ̀ʥɥàmań́̃ àʤùam̀̃ a ́̃ Prostitute; fornication 36. æ̀ʥɥúma ́̃ āʤṹma;́̃ ɛʤṹma ́̃ Work, labour 37. æ̀ʥɥúmaʥ́̃ ań́̃ àʤúmaʤ́̃ aí̃ ,́̃ Unprofitable labour, a wild goose àʤṹmaʤ́̃ aŋ́̃ ́ chase; in vain 38. àkàsaẁ̃ á àkàsàwá Cymbal 39. àkàtàɥìá àkàtàwiíá Umbrella 40. ŋk̀áꜝtɪ ́ ākáté, káꜝtíɛ ́ Groundnut 41. àkæ̀ǽꜝtíá àkèétiā, àklɛ̀t́īa Chimpanzee, baboon 42. ??æ̀kùròwá àklòwá village 43. áꜝkó ákō A bird, parrot 44. kʷæ̀dú àkɔd̀ṹ, kɔd̀ṹ Banana 45. àkʊ́kɔʥ́ań́̃ àkɔḱɔʤ́aí̃ ́̃ A kind of domestic fowl that often has very few feathers 46. àkʊḱ ɔb́ɛ́ꜝ sa ́̃ àkɔḱɔb́ɛśa ̀̃ A climbing or straggling thorny University of Ghana http://ugspace.ug.edu.gh 303 shrub 47. àkɔń́nɔ ́ àkɔńɔ ́ Lust, strong desire 48. kɔt́ɔḱŕɔ ́ àkɔt́ɔḱlɔ ́ Hooked drumstick; a wooden fork used in thatching 49. ??ɛp̀á àk͡pá, Trap for catching animals 50. àpàʨɪ ́ àkp͡ àké Crutch; any disease that renders the victim lame 51. ɔp̀àpʊ́ àk͡pak̀̃ ͡pó, àk͡pākp͡ ó Male goat, billy goat, (fig.)sex maniac 52. àpɪḿ́̃ àk͡pé thousand 53. æ̀kùtú àkù̃tú, àkù̃tṹ; àkùtú Sweet orange 54. àmaŋ̀̃ k̀aǹ̃ i ́̃ àmak̀̃ aǹ̃ i,́̃ àmaŋ̀̃ k̀aǹ̃ i,́̃ Cocoyam, taro maǹ̀̃ kaǹ̃ i,́̃ mak̀̃ aǹ̃ i ́̃ 55. àmaǹ̃ ɪ ́̃ àmaǹ̃ é; àmaǹ̃ è Death, trouble, calamity 56. àmaǹ̃ ɪh̀̃ ṹ àmaǹ̃ èhlṹ, àmaǹ̃ èhṹ Affliction, suffering, unpleasant occurrence 57. àmaǹ̃ ɪɛ̀̃ ́̃ àmaǹ̃ ìɛ ́ Story, case, event, matter 58. ??àpæ̀̃ǹtú ámútu ̃̄ Twin plantain, plantain fingers joined down the side 59. ??ànaǹ̀̃ sɪ ̀ ànas̀̃ è, Disease, shingles 60. æ̀nih́̃ áẃ ànih́̃ áó Laziness 61. ànʊk̀̃ ʷárɪ ́ ànòkʷádé Fact, truth 62. anʊkʷarɪ ànɔk̀uálé Truth, fact 63. ànʊk̃̀ ʷárɪ ́ ànɔk̀̃ úalé True, correct; truly, indeed 64. æ̀pítí àk͡pítí; àbìtì A plantain cake, steamed or baked. 65. pàá àpàá Hire, labour, job; wages 66. ??áḿpaǹ̃ ɪ ́̃ ápāní Edible species of bat, esp. rosette bat, Gambian fruit bat 67. ??àpààwá àpàwá Flat round wooden tray, carried on the head, for displaying wares 68. àpɪḿ́̃ àpíḿ A kind of plantain, with thin fingers 69. àm̀pèsí àpèsí Boiled green plantain 70. àbíréʨí àpléʧí Goat 71. àsàbrà̀nɪ ́ àsàblà́; sã̀blã̀í Appellation, title 72. àsàfʊ̀ àsàfò Group, company, congregation 73. àsàrɪ ̀ àsàlè Personal name 74. asɛw (no asɛu Misfortune University of Ghana http://ugspace.ug.edu.gh 304 transcript) 75. asrɛ (no àsìlɛ,́ àslɛ̀,́ àslɛɛ̀ ́ comb transcript.) 76. àsímæ̀sí àsímas̀̃ í So-and-so, what’s-his-name, somebody 77. ??ɔs̀rà́àfʊ́ áslàáfó soldier 78. ??ǹsrà̀-bań́̃ asra, sala Camp, military camp 79. ??ǹtáfŕɛ ́ àtɛṕlɛɛ̀ ́ Cockroach 80. ??àtɪá̃ ́̃ àtiá̃ ,̀̃ àtsiá̃ ̀̃ Cashew nut tree; cashew nut 81. atrɔmbaa (no àtlɔm̀bàá, atlombi A seed transc.) 82. æ̀tírìdìí àtŕìdìí, àtĺìdìí fever 83. æ̀tᶣià̃ ́̃ àtũ̀a ́̃ Rebellion, defiance 84. ??àʨɥɪr̀ɪ ́ àtsré; àʧùlé Ladle 85. æ̀tùpań́̃ àtũkp͡ á Bottle 86. æ̀tùm̀pań́̃ àtũ̀pàní A large talking drum 87. awaa awaaáwà A kind of music and dance played in coastal areas 88. aɥɪndadɪ awendade A plant 89. àɥɪr̀ɛh̀ʊ́ àwɛr̀ɛh̀ó, àwīlɛh̀ó, Grief, sorrow, sadness àjùrɛh̀ó 90. ??jààwá àjáwá A large metal pan, tray or box 91. àjɪf̀ʊ́rɔ ́ àjɛf̀ló, àjɛf̀lóò, àjìfló, Bride, newly married girl àjìflóò 92. àjɪf́ŕɛ ́ àjífálé An adultery fine, paid by a wife’s lover 93. bàbàsʊ̀ bàbàsò, bàbàzò Venereal disease, gonorrhea 94. ??bàsàà bàsàà Disorderly, haphazardly, confused, reckless 95. bàsàbàsà bàsàbàsà Carelessness, lack of respect, lack of self-discipline and self-control; bad, disorderly, uncultured, chaotic 96. bɪḿ́̃ béḿ Innocence 97. àbèdᶣìá bɛʤ̀ùá A palm tree, when slightly taller than tamlaa and still untrimmed 98. bɛr̀ɛd̀ᶣìá (no bɛl̀ɛd̀ùá A tall straggling shrub transc) University of Ghana http://ugspace.ug.edu.gh 305 99. ?bɛŋ bɛŋ bayonet 100. bísí bàsàà bísíbàsàà Disorderly, in a disorderly manner 101. bɪr̀ɛ ẁ̀ blɛò̀ Slowly, carefully, gently 102. búró ɲà bĺôɲa;̀̃ blóòɲa ̀̃ Christmas, the Christmas season 103. bòǹ sù bòsò Whale 104. ɔbɔ́ꜝ fʊ ́ bɔf́ō Messenger; angel; helper at a critical time 105. bʊ̀ɹɔ f̀ʊ́ blɔ̀f̄óɲò; blɛ̀fónò European, foreign 106. bɔk̀ ɔb̀ɔk̀ɔ ̀ bɔk̀ɔb̀ɔk̀ɔ ̀ Gentle, slow, calm; gently, slowly 107. bɔk̀ ɔɔ̀ ̀ bɔk̀ɔɔ̀ ̀ Slow, calm 108. ??bú bū Listen, obey, be respectful 109. bu bù proclaim 110. ??bú bù Count, work out a problem, check 111. ??bʊ ̀á búá Gather, collect, assemble 112. ??bʊ ̀á búá Help 113. ??bʊ ̀á bùá Lie, pretend; do intentionally, decide to do; take lightly, joke about sth; judge, determine 114. bùtú bútù Capsize, be overset, overthrow, collapse 115. ??dá á dáá Every day; always, often; forever 116. dæ̀æ̀ bí dààbí, dàbí, dɛb̀í Excl. no, negative response, signifying dissent, disagreement 117. dæ̀b ídà dàbídà, dɛb̀ídà No, emphatic negative response 118. ??dà dààdà dàdà, dādā Past times, ancient times 119. dàd ààdà dàdààdà Of old, long ago 120. dàd ɪ ́ dàdé Iron, metal; catlass, sword; money (sem. extension) 121. dàd ɪs̀ɛŋ́ ́ dàdèsɛí̃ ,́̃ dàìsɛí̃ ́̃ Iron pot 122. ??dà nɪ ́ daí̃ ̀̃ Change, turn into, metamorphose 123. ??àd áꜝká ; dákà Box, chest, trunk; square or (Port.?) rectangular shape; lie, liar 124. ɔd̀áꜝ sɪ ́ òdásē, dásē Witness 125. ??dá ńtá dátá Belt, girdle 126. diń́̃ ń ; dĩ̃́ĩ ̃́ Reserved, reluctant to talk, habitually quiet; quietly, silently 127. dɔŋ̀̃ k̀ɔ ́ dɔŋ̀k̀ɔ,́ òdɔk̀ɔ ́ slave University of Ghana http://ugspace.ug.edu.gh 306 128. ɔf̀a ́̃ fa ́̃ Half, piece 129. fɛẃ́ fɛó́, fɛú́ Beauty 130. fìrí flì́ Buy or sell on credit 131. m̀fò ní fòní, òfòní Picture, photo 132. fɔ ́ fɔ ́ Guilt 133. fùfú fùfùí Food, fufu, obtained by pounding boiled yam, cassava, or plantain (or a mixture) in a mortar 134. fusu o (no fùsúo A large antelope, waterbuck transc.) 135. ??fí táá fútaa Very white; dazzling white 136. fʊt̀ ʊ ́ fútŕú, fòtĺó Treasure 137. ??fù tú fútù Mix; be misty, foggy 138. ??æ̀ ʥań́̃ gaí̃ ;̀̃ ɡɛ̀͂ɛ͂̃́; kɛ̀͂ɛ͂̃́ bow 139. ʥɥò ŋk̀ú ɡ͡bòŋù́ Thigh 140. ??ɡú á ɡua, ɡwa assembly 141. ɡua ɡúa Gather 142. ɡʷú áḿ ɡʷúám̄ A gathering 143. ɡùà ẁ ɡwaò Whip, beat 144. háẃ hàò worry 145. ??w òsó hósò shake 146. ??hu hú Shout someone down; blow, of wind 147. ʥɥá ḿ (no ʤūāmì Inside the market transcr.) 148. ʥɥɪ ŋ́̃ ̀ ʤù̃ɛ ̀̃ Reason, think 149. ka ́̃ ka ́̃ Swear, invoke 150. ka ́̃ ka ́̃ Turn fufu as it is pounded 151. kàɪ ́ káì, Remind, remember, recollect 152. káp rɛ̀ ́(Eng. káplɛ̀ ́ Any coin of very low value copper) 153. kato (no kat̀̃ ó Shin bone transcr.) 154. ʨɪń aḿ́̃ kénaḿ̃ i ́̃ Fried fish 155. kese ?? késé Tell University of Ghana http://ugspace.ug.edu.gh 307 156. kó klésìà Wild date palm 157. krɔɔ (no klɔɔ̀̃ ,́̃ kʷlɔ̀,́̃ kùlɔ ́̃ table transc.) 158. ??kʊ ̀tʊ́ kótò Bend over; roll around flat on the ground 159. ??kó ḿḿ kóóm Alone, quietly 160. ??kò tòkú kòtòkú Sack, pocket 161. kòǹ trò̀m̀fì kòtròfì Chimpanzee 162. ??kɔ k̀ɔb̀ʊ́ kɔk̀ɔb̀ó Mongoose; fox 163. kɔǹ̀̃ kɔǹ̀̃ sa ́̃ kɔk̀ɔs̀á Slander; damaging report 164. kɔk̀ ɔs̀æ̀ʨì kɔk̀ɔs̀ɛʧ̀ì A title for the vulture 165. kɔk̀ ɔt̀ɛ ̀(Fn) kɔk̀ɔt̀ɛ ̀ Sea catfish 166. kɔt̀ɔ ́ꜝ kɔ ́ kɔt̀ɔḱɔ ́ Crested porcupine 167. ??pá pá kp͡ ák͡pá Goodness, hospitality, correct thing; good; well, successfully, truly, indeed, really 168. ɔp̀àp ʊ́ àkp͡ ak̀̃ p͡ ó Male goat 169. ??pà tá k͡pat́̃ a ̀̃ Pacify, mollify, settle a misunderstanding 170. àpæ̀̃ ǹtú k͡pat̀̃ ṹ big-fingered plantain. Often used for fufu 171. pɛp̀ɛ ɛ̀ ́ k͡pɛk̀p͡ ɛɛ̀ ́ stinginess 172. ɛpra kp͡ laa armadillo 173. pòk úpòkú (Fn) k͡pókú, k͡pókp͡ ókú, A river fish, tuna k͡pókp͡ ĺókú, k͡pōkp͡ lōkú, pópókú 174. kʷà sɪà́ kùàsìá A foolish, stupid person, a common insult 175. ŋk̀u ǹ̃ íḿ kuñ̀ iḿ́̃ , kũ̀niḿ̃ i ́̃ Victory, success 176. ??kʊ ̀sɛ ̀ kùsìɛ ̀ God forbid! 177. ??ku suu kùsùù Dull, of weather 178. ŋk̀ó ńtóḿpó kútúḿpɔ,́ kútúpɔ ́ Undermining, intrigue (Semantic shift) 179. mm àá màá, àmàá Clappers used for musical time keeping, made of slip bamboo; nickname for a goat 180. maǹ̃ ɪ ́̃ mané, manê Send sth. via a messenger; order sth. 181. ??m aḿ́̃ pam̀́̃ maṕ̃ ām̄ Monitor lizard University of Ghana http://ugspace.ug.edu.gh 308 182. m̀m ààwá maẁ̃ á, àblàwá Servant 183. m̀m àrá mlà́, mlàá Law 184. ??ḿ mó mó Of congratulations, well done! 185. m̀m ùsú mùsú Abomination, taboo, ill luck 186. m̀m ɔb́ɔ ́ mɔb́ɔ,̀ mɔḿɔ ̀ Pity, sorrow 187. m̀m ɔb̀ɔm̀̀mɔb́ɔ ́ mɔb́ɔm̀ɔb́ɔ ̀ Pitifully, sadly 188. m̀m ɔd́ɪń́̃ mɔd́é, mádé, mɔd́í Effort, esp. successful effort 189. ??m ṹ mù̃; mlùkú Whole 190. ??na n naǹ̃ ē, naǹ̃ è Leg, foot; hind leg of an animal 191. ??òn i ́̃ ni ̄̃ Person 192. òbú búáꜝfʊ́ óbúbúafō Person crippled by an accident 193. ɔd̀áꜝ sɪ ́ òdásē Testimony, witness 194. ɔd̀ɪ ́ꜝ ɕɪ ́ òdéhē, àblàdè Person of royal lineage, noble 195. ɔdo dobɛŋ ododobɛŋ́ A tall grass, reed 196. ɔd̀ò ǹnó òdònó Hour-glass drum (Mande?) 197. ??ɔh̀ a ́̃ òha ́̃ Flying squirrel 198. òhià̃ ́̃ òhià̃ ́̃ poverty 199. òhià̃ ̃́ꜝ fʊ́ òhià̃ f́̃ ō Poor person 200. òʥù ma ́̃ òʤùma ́̃ Umbrella tree 201. ɔkrɔ ɔ òklɔɔ̀̃ ;́̃ òkplɔɔ̀̃ ́̃ Table 202. ɔk̀ɔ́ꜝ tɔ ́ òkɔ́ꜝ tɔ ́ A small greenish river crab 203. ɔk̀ɔm̀̀ fʊ ́ òkùm̀fó soothsayer 204. ɔku mbɔmmɔfʊɔ òkúḿ-mɔm̀ɔf̀ūɔ The poisonous twig or vine snake 205. ɔp̀ɛ́ꜝ sɛ ́ òpɛs̀ɛ ́ Brush-tailed porcupine; hedgehog 206. ò̀pú ró òpléú squirrel 207. ɔp̀rɛ̀ m̀ʊ̃́ òplɛm̀i ́̃ Cannons, heavy artillery, very big guns 208. ɔprɛ tɛ (no òplɛ̀t́ɛ ̄ A Klama drum, beaten with the transcr) hand 209. ?ɔsɛ ɛ osɛɛ God 210. osis ibisi òsísíblísì, òsísíbìsí Honey rattel, bear, wolf 211. ɔs̀ɔ́ꜝ fʊ ́ òsɔf́ō Priest (christian) 212. ??ɔt̀ ɔ ́ òtɔ ́ A ritual food made of mashed yam or other tuber mixed with palm University of Ghana http://ugspace.ug.edu.gh 309 kernel or palm oil. 213. ɔt̀ɔ́ꜝ f ʊ ́ òtɔf́ō Accidental death; ghost of one who died an accidental death 214. ɔʨ̀ɪà ̃ ̃́ꜝ mɪ ́̃ òʧìāꜝmɛ,́ òʧaà̃ ḿ̃ i,̄̃ Spokesman, linguist òʧaà̃ m̄́̃ , òʧià̃ ḿ̃ i ̄̃ 215. ??ɔʨ̀ ɪr̀ɛm̀a ́̃ òʧrè̀má A short drum (Semantic shift) 216. owu ɲʷuka (no òwùɲũ̄kā An evergreen shrub transcr.) 217. pàp à pàpà Fan 218. pàp àhʊ̃́ pàpàmí Towel 219. pæ̀t ú pàtú owl 220. ??pɛ ṕɛɛ́ṕɛ ́ pɛṕɛɛ́ṕɛ ́ Exactly, precisely, orderly 221. pɛs̀ɛ m̀ɪŋ́̃ ḱʊ̃́mɪɲ́̃ a ́̃ pɛs̀ɛk̀úmíɲa ́̃ Egotism, self-centredness 222. pàtæ̀ kú pɛt́ɛḱú Hyena 223. ɔb̀rà́ ꜝfʊ́ plàáfō State executioner 224. pow a pòwá A bead 225. ??sá sá Draw water 226. sæ̀̃ŋ k̀ú saŋ̀̃ k̀ṹ piano 227. sìàd ɪb́ʊ́á síadébóa, sìalèbúa hedgehog 228. sìká sìká money 229. ǹsék úró sìékú, sékú Slander, gossip 230. sìsí sīsí Cheat 231. ??sr a srà́, slà́, slàá Visit; watch, be alert, vigilant 232. èsú sú Character, habit, behaviour, nature 233. sùba ń́̃ súbaí̃ ́̃ Character 234. ??sù sú súsù Think about, measure, imagine 235. súńs ꜝúḿ sūsūmá Spirit, image 236. ??ta tá Patch 237. ??àt àdɪ ́ tàdé Clothing to cover the upper part of the body 238. ??ta ń́̃ taí̃ ,́̃ taŋ́̃ ́ Unpleasant, ugly 239. tɪǹa ̀̃ àsɪ ̀ tí nàsè; tí dàsè Response to ‘tlikua’ 240. tɪá̃ á̃ ́̃ tiá̃ á̃ ́̃ Thin and long, slim, narrow 241. tɪɪ̀̃ ̀̃ tiì̃ ̀̃ Straight, directly 242. ??tɪ t̀ɪ ̀ tìtì The past, ancient times University of Ghana http://ugspace.ug.edu.gh 310 243. títír íẃ títlī, títrī Mainly, chiefly, especially 244. tírí ŋḱʷa ́̃ tlí kũ̄a ̄̃ Expr. Of congratulations 245. ??æ̀ tᶣìá tóá Attack (Semantic shift?) 246. ʨɥɪ ǹ̃ ɪs̀̃ iń́̃ tōìsiń́̃ A kind of drum and its music, used for chiefs 247. tɔm tɔḿ̃ , tɔń̃ Screw pine 248. ?ètú tṹ Gun, shooting 249. wáw á wáwá A tree 250. ?wʊ ̀sʊ́ wósò; hósò Shake 251. wɔf̀ à wɔf̀à Maternal uncle 252. wɔf̀ àsɪ ́ wɔf̀àsé Maternal uncle; niece or nephew, child of a man’s sister University of Ghana http://ugspace.ug.edu.gh 311 Appendix 6: Ga war song This is an Asafo song sung when feud or dispute arises. It is sung to warn the opponents or enemies of the preparedness and might of the Asafo member coming into the ‘war’. Song Transcription 1. Agya Yaw eeei, Yaw o ya! [aʥa jaw eeei jao o ja] Father Yaw, Yaw! (Crying out Father Yaw’s name!) 2. Agya Yaw eeeei, Agya Yaw eeei! [aʥa jaw eeei aʥa jao eeei] Father Yaw, Father Yaw! (Crying out Father Yaw’s name) 3. Obi a-kɔ-fa biribi a-ba, Yaw oo, Yaw ooo eya! [Obi akɔfa biribi aba jao oo, jao oo eja] Somebody PERF-go-bring something PERF-come, Yaw oo, Yaw ooo! Yaw, somebody has caused a problem! 4. Agya Yaw eeei, Agya Yaw a-kɔ-fa manso aba ooo! [aʥa jao eeei aʥa akɔfa mãnso aba oo] Father Yaw, Father Yaw PERF-go-bring litigation! Father Yaw has caused litigation! 5. Yaw eeei, Yaw eeei! [jao eeei jao eeei] Yaw, Yaw! (Crying out Yaw’s name!) 6. Biribi Akyiampɔn twa gidi gandan! [biribi aʨiampɔn ʧʷa gidi gandan] Something Acheampong crosses gallantly! We are gallantly marching pass our opponents/enemies to scare them! 7. Biribi Akyiampɔn twa gidi gandan! [biribi aʨiampɔn tʧʷa gidi gandan] University of Ghana http://ugspace.ug.edu.gh 312 Something Acheampong crosses gallantly! We are gallantly marching pass our opponents/enemies to scare them! 8. Agya Yaw eeei, Yaw eeei! [aʥa jao eeei jao eeei] Father Yaw, Yaw! (Crying out Father Yaw’s name) 9. Yɔɔ miafo eeei, Yaw eeei! [jɔɔ miafo eeei jao eeei] Retreat, Yaw! Yaw, the person who retreats! University of Ghana http://ugspace.ug.edu.gh 313 Appendix 7: Dangme war song The War Cry of the Warriors of the Kudzragbe Division of the Ada Traditional Area sing during the Ada Asafotufiami Festival and also mainly after the swearing of oath of allegiance to the Paramount Chief and the President of the Ada Traditional Area of the Greater Accra Region, Djetse Nene Abram Kabu Akuaku III. Song Transcription Call Yɛn anim ogya ee, ogya ee (asase)! [jɛn anim oʥa ee, oʥa ee (asase)] There is fire in our eyes, fire [landlord]! We are seriously ready to attack (in the battle)! Response Ee, ogya ee! [ee, oʥa ee] Yes, fire! Seriously ready for war. Repeat Ee, ogya ee! [ee oʥa ee] Yes, fire! Seriously ready for war. Ogyae (ne) su! [oʥa ne su] The fire is set (lit)! Ready. Ogya oo! [oʥa oo] Fire! Serious. Call Asase! [asase] Land (The Landlord/Owner)! Response Hee! [hee] Here am I! Call Asase! [asase] Land (The Landlord/Owner)! Response Hee! [hee] Here am I! Question Wobɛyɛ dɛn? [wobɛjɛ dɛn] What will you do? Action Wazo, wazo, wazo! [wazo wazo wazo] (Stepping of feet on the ground to indicate readiness!) Call Pediatɔ! [pediatɔ] The name of the Chief Warrior/Warlord/Captain/Leader! Response Hee! [hee] Here am I! University of Ghana http://ugspace.ug.edu.gh 314 Call Peditɔ! [pediatɔ] The name of the Chief Warrior/Warlord/Captain/Leader! Response Hee! [hee] Here am I! Question Wobɛyɛ dɛn? [wobɛjɛ dɛn] What will you do? Action Waya, waya, waya! [waja waja waja] (Shaking the chest and weapons very vigorously (bravery)!) Action Waya ee! [waja ee] (Warriors raise their ammunitions/weapons singing!) Yɛn nie oo! [jɛn nie oo] Here we are! Yɛn agya nie oo! [jɛn aʥa nie oo] Here is our father (Leader/Captain/Warlord/Chief)! Yɛkum a yɛnsuro! [jɛkum a jɛnsuro] When we are killing (our adversaries) we are not afraid! Repetition Call Ee, Asase! [ee asase] Land (The Landlord/Owner)! Response Hee! [hee] Here am I! Call Asase! [asase] Land (The Landlord/Owner)! Response Hee! [hee] Here am I! Question Wobɛyɛ dɛn! [wobɛjɛ dɛn] What will you do? Action Wazo, wazo, wazo! [wazo wazo wazo] (Stepping of feet on the ground to indicate readiness!) Call Pediatɔ! [pediatɔ] The name of the Chief Warrior/Warlord/Captain/Leader! Response Hee! [hee] Here am I! Call Peditɔ! [pediatɔ] The name of the Chief Warrior/Warlord/Captain/Leader! Response Hee! [hee] Here am I! Question Wobɛyɛ dɛn? [wobɛjɛ dɛn] What will you do? Action Waya, waya, waya! [waja waja waja] (Shaking the chest and weapons very vigorously (bravery)!) Action Waya ee! [waja ee] (Warriors raise their ammunitions/weapons singing!) Yɛn nie oo! [jɛn nie oo] Here we are! University of Ghana http://ugspace.ug.edu.gh 315 Yɛn agya nie oo! [jɛn aʥa nie oo] Here is our father (Leader/Captain/Warlord/Chief)! Yɛkum a yɛnsuro! [jɛkum a jɛnsuro] When we are killing (our adversaries) we are not afraid! Notes 1. Asase. The Kudzragbe Clan of the Ada Traditional Area are the custodians of the entire land of the traditional home (Hometown) for all Ada natives. They, therefore, use the Asase (land) as the major linguistic identity in their war cry. 2. Pediator. The Stool Name is Pediator [pediatɔ]. An Akan corrupted phrase ‘wopɛ dea, to’ which means if you like, throw (it at me/us) … your weapon… and you will see the action that will follow. The caller calls on him as the leader of the team for action or direction. 3. Ogya. The metaphor for seriousness, war-ready, etc. Fire always means danger! University of Ghana http://ugspace.ug.edu.gh 316 REFERENCES Abakah, Emmanuel N. 2013. Vowel replacement patterns in the Mfantse dialect of Akan. Journal of Universal Language 14(2), 7-51. Abakah, Emmanuel N. 2012. Some assimilatory processes in Akan. Journal of West African Languages 39(2), 47-82. Abakah, Emmanuel N. 2006. Where have all the consonantal phonemes of Akan gone? Journal of Philosophy and Culture 1/2, 21-48. Abakah, Emmanuel N. 2005b. Tone rules in Akan. Journal of West African Languages 32(1), 109-134. Abakah, Emmanuel N. 2005a. Phonological analysis of word-final consonants in Akan. Africa and Asia 5, 47-65. Abakah, Emmanuel N. 2004. Elision in Fante. Africa and Asia 4, 181-213. Abakah, Emmanuel N. 2002. Remarks on the Akan vocalic inventory. In Ameka, F. & E. K. Osam (eds.), New Directions in Ghanaian Linguistics: Essays in honour of the 3Ds, pp. 243-263. Accra: Black Mask Ltd. Abakah, Emmanuel N. 2000. A closer look at downstep in Akan. Afrika und Übersee 83, 1–23. Abakah, Emmanuel N. & Tenteh, A. M. K. 2005. Patterns of vowel epenthesis in University of Ghana http://ugspace.ug.edu.gh 317 Akan loanword phonology. In Napon, A. (ed.) Cahiers du CERLESHS: Actes du colloque interuniversitaire sur la coexistence des langues en Afrique de l’Ouest, pp. 227-251. Ouagadougou, Burkina Faso. Abedi-Boafo, J. 1971/80. Dangme Nyaii (Classical and Idiomatic Dangme) (2nd Ed.). Accra: Bureau of Ghana Languages. Addo, Gladys N. 2002. Lexical innovation in Akan. M.Phil thesis, University of Ghana, Legon. Adomako, Kwasi. 2013. Underapplication in Akan loanword adaptation. International Journal of Linguistics 5(5), 174-196. Adomako, Kwasi. 2008. Vowel epenthesis and consonant deletion in loanwords: a study of Akan. Master’s thesis, University of Tromsø. Adu Manyah, Kofi. 2003a. Vowel quantity contrasts in Twi. 15th International Congress of Phonetic Sciences (ICPhS), pp, 3185-3188. Barcelona. Agbedor, Paul K. 2006. Lexical Borrowing: The Case of Ewe. In Obeng, Samuel G. & C. S. Obeng (eds.), From Linguistics to Cultural Anthropology: Aspects of Language, Culture and Family Issues in Ghana (West Africa), pp. 87-110. Munchen: Lincom. Agyekum, Kofi. 2006. The sociolinguistic of Akan personal names. Nordic Journal of African Studies 15(2), 206–235. Apenteng, Monica A. 2013. A morpho-phonological study of Akan loan words. M.Phil thesis, University of Ghana, Legon. Apenteng, Monica A. & Nana Aba Appiah Amfo. 2014. The form and function of University of Ghana http://ugspace.ug.edu.gh 318 English loanwords in Akan. Nordic Journal of African Studies 23(4), 219- 240. Appah, Clement Kwamina I. 2003. Nominal derivation in Akan: A descriptive analysis. MPhil thesis, Norwegian University of Science and Technology, Trondheim. Apronti, E. O. 1990. Dangme Ngmami bɔ (The writing of Dangme) revised edition. University of Ghana, Legon. Apronti, E. O. 1967. A phonetic and phonological study of the nominal piece in Dangme. PhD thesis, University of London. Apronti, E. O. & Kropp Dakubu, Mary E. 1972. Towards a dialect geography of Dangme. Institute of African Studies Research Review 8(3), 35-46. Awedoba, Albert K. 1980. Borrowed nouns in Kasem nominal class. Anthropological Linguistics 22(6), 248-263. Ballard, Lee W. 2010. Akan vowel harmony in Optimality Theory. Master’s thesis, University of Florida, Florida. Best, Catherine T. 1994. The emergence of native-language phonological influences in infants: A perceptual assimilation model. In Goodman, J. & H. Nusbaum (eds.), The Development of Speech Perception: The Transition from Speech Sounds to Spoken Words, pp. 167-224. Cambridge, MA: MIT Press. Boadi, Lawrence A. 2009. A comparative phonological study of some verbal affixes in seven Volta-Comoe languages of Ghana. Accra: Black Mask Ltd. Boadi, Lawrence A. 1988. Problems of palatalisation in Akan. Journal of West University of Ghana http://ugspace.ug.edu.gh 319 African Languages 38(1), 3-16. Bodomo, Adams & Marfo, Charles. 2006. The morphophonology of noun classes in Dagaare and Akan. Studi Linguistici e Filologici Online 4, 205-244. Boersma, Paul. 2011. Praat Intro: Viewing a spectrogram. Boersma, Paul. 2007. Cue constraints and their interactions in phonological perception and production perception. ROA#944. Boersma, Paul & Hamann, Silke. 2009. Loanword adaptation as first-language phonological perception. ROA#975. Broselow, Ellen. 2000. Stress, epenthesis, and segment transformation in Selayarese loans. In Chang, S., L. Liaw & J. Ruppenhofer (eds.), Proceedings of the Twenty-fifth Annual Meeting of the Berkeley Linguistics Society, pp. 311-325. Berkeley, CA: Berkeley Linguistics Society. Caesar, Regina O. & Adi, Daniel B. 2014. Dangme fonɛtiks kɛ fɔnɔlɔji (An introduction to phonetics and phonology of Dangme). Kumasi: Alpha and Omega Publications. Campbell, Akua A. 2017. A grammar of Gã. PhD thesis, Rice University. Chomsky, Noam & Halle, Morris. 1968. The Sound Pattern of English. New York: Harper and Row. Christaller, Johann G. 1933. A dictionary of the Asante and Fante Language called Tshi (Chwee, Twi) (2nd Ed.). Basel: Basel Evangelical Missionary Society. University of Ghana http://ugspace.ug.edu.gh 320 Clements, George N. 1981. Akan vowel harmony: a nonlinear analysis. Harvard Studies in Phonology 2, 108-77. Bloomington: Indiana University Linguistics Club. Dakubu, Mary Esther K. 2013. Dangme-English dictionary with English-Dangme glossary (draft). University of Ghana, Legon. Dakubu, Mary Esther K. 2009b. Ga-English dictionary with English-Ga index (2nd Ed.). Accra: Black Mask Ltd. Dakubu, Mary Esther K. 2009a. The historical dynamic of multilingualism in Accra. In McLaughlin, F. (ed.), The languages of urban Africa, pp. 19-31. London: Continuum International Publishing Group. Dakubu, Mary Esther K. 2002. Ga phonology. Language Monograph Series No. 6. Accra: Institute of African Studies, University of Ghana, Legon. Dakubu, Mary Esther K. 1999. Ga/English dictionary with English/Ga index. Accra: Black Mask Ltd. Dakubu, Mary Esther K. 1987. The Dangme language: An introductory survey. Accra: Unimax Publishers Ltd. Dakubu, Mary Esther K. 1974. Some features of tone in Dangme. Journal of West African Languages 9(2), 79-86. Dakubu, Mary Esther K. 1973. A survey of borrowed words in Dangme. Research Review, 81-128. De Lacy, Paul. 2001. Markedness in prominent positions. MIT Working Papers in Linguistics 40, 53-66. Dohlus, Katrin. 2010. The role of phonology and phonetics in loanword University of Ghana http://ugspace.ug.edu.gh 321 adaptation: German and French front rounded vowels in Japanese (Vol. 353). Peter Lang. Dolphyne, Florence A. 2006 [1988]. The Akan (Fante-Twi) language: Its sound system and tonal structure. Accra: Woeli Publishing Services. Dolphyne, Florence A. 1967. A phonological analysis of Twi vowels. Journal of West African Languages 4(1), 83-89. Dorvlo, Kofi. 2008. A grammar of Logba (Ikpana). PhD thesis, University of Leiden. Dzameshie. Alex K. 1996. The social dynamics of lexical borrowing. Legon Journal of Humanities 9, 141-157. Flege, James E. 2003. Assessing constraints on second-language segmental production and perception. In Meyer, A. & N. O. Schiller (eds.), Phonetics and Phonology in Language Comprehension and Production: Differences and Similarities, pp. 319–355. Berlin: Mouton de Gruyter. Flege, James E. & MacKay, Ian R. A. 2004. Perceiving vowels in a second language. Studies in Second Language Acquisition 26, 1–34. Genzel, Susanne. 2013. Lexical and post-lexical tones in Akan. Doctoral dissertation, University of Potsdam. Goldsmith, John. 1976. Autosegmental Phonology. Doctoral dissertation, MIT. New York: Garland Press, 1979. Goldsmith, John, & Laks, Bernard. 2010. Generative phonology: its origins, its principles, and its successors. The Cambridge History of Linguistics. Cambridge: Cambridge University Press. [en ligne: http://hum. uchicago. edu. University of Ghana http://ugspace.ug.edu.gh 322 Hayes, Bruce. 2009. Introductory phonology (Blackwell textbooks in linguistics). Sussex: Wiley-Blackwell. Heffernan, Kevin. 2005. Phonetic similarity and phonemic contrast in loanword adaptation. Toronto Working Papers in Linguistics 24, 117-123. Heine, Bernd. 1968. The allocation of loan-words within the nominal class system of some Togo Remnant languages. Journal of African Languages 7, 130- 139. Hudu, Fusheini A. 2002. Phonological integration of English-Dagbani loanwords. BA long essay, University of Ghana. Jacobs, Haike. & Gussenhoven, Carlos. 2000. Loan phonology: perception, salience, the lexicon, and OT. Optimality Theory: Phonology, Syntax, and Acquisition, 193-210. Kager, René. 1999. Optimality Theory. Cambridge Textbooks in Linguistics. Cambridge: Cambridge University Press. Kenstowicz, Michael. 2005. The phonetics and phonology of Korean loanword adaptation. In Proceedings of the first European conference on Korean linguistics1, pp.17-32. Kenstowicz, Michael. 2003b. Salience and similarity in loanword adaptation: A case study from Fijian. Ms. ROA#609. Cambridge, MA: MIT. Kenstowicz, Michael. 2003a. The role of perception in loanword phonology. A review of Les emprunts linguistiques d’origine européenne en Fon by Flavien Gbéto, Köln: Rüdiger Köppe Verlag, 2000. Studies in African Linguistics 32 (1), 95-112. University of Ghana http://ugspace.ug.edu.gh 323 Kenstowicz, Michael. 2001. The role of perception in loanword phonology. Linguistique Africaine 20. Kenstowicz, Michael. 1996. Sonority-Driven Stress. ROA#33. Kenstowicz, Michael & Sohn, Hyang-Sook. 2001. Accentual adaptation in North Kyungsang Korean. Current Studies in Linguistics Series 36, 239-270. Kenstowicz, Michael. & Suchato, Atiwong. 2006. Issues in loanword adaptation: A case study from Thai. Lingua 116, 921-949. Kisseberth, Charles. 1970. On the functional unity of phonological rules. Linguistic Inquiry 1, 291-306. Koelle, Sigismund W. 1854. Polyglotta Africana. London: Church Missionary Society. Cited in Stewart (1966). Koffi, Phil Y. 1998. The expansion and integration of the loanwords in the Togo Remnant Languages: An approach based on the Akebu language. Kotei, Nii A. 1969. A description of modern spoken Ga with reference to tone and intonation. PhD thesis, Northwestern University. Kügler, Frank. 2016. Tone and intonation in Akan. In Downing, L. J. & A. Rialland (eds.), Intonation in African Tone Languages, pp. 89-129. Berlin: De Gruyter Mouton. Kügler, Frank. 2015. Phonological phrasing and ATR vowel harmony in Akan. Phonology 32(1), 177 – 204. Kügler, Frank & Genzel, Susanne. 2011b. Phonetic realisation of automatic (downdrift) and non-automatic downstep in Akan. ICPhS 17, 735–738. Kügler, Frank & Genzel, Susanne. 2011a. On the prosodic expression of pragmatic University of Ghana http://ugspace.ug.edu.gh 324 prominence: the case of pitch register lowering in Akan. Language and Speech 55(3), 331–359. LaCharité, Darlene & Paradis, Carole. 2005. Category preservation and proximity versus phonetic approximation in loanword adaptation. Linguistic Inquiry 36(2), 223-258. LaCharité, Darlene & Paradis, Carole. 2002. Addressing and disconfirming some predictions of phonetic approximation for loanword adaptation. Langues et linguistique, 28, 71-91. Ladefoged, Peter. 1993. A Course in Phonetics (3rd Ed.). London: Harcourt Brace Javanovich Publishers. Larweh, Richard A. 2017. Compounding in Dangme. MPhil thesis, University of Ghana. Mante, Amu A. 1971. Comparative study of the Gã spoken by the older generation of Tɛʃi, Labadi, and Central Accra, particularly the fishermen, with special interest in lexis. BA long essay, University of Ghana, Legon. McCarthy, John. 2003b. Phonology: An overview. In Frawley, William J. (ed.), International Encyclopedia of Linguistics (2nd Ed.), pp. 327-330. Oxford: Oxford University Press. McCarthy, John. 2003a. Optimality Theory: An overview. In Frawley, William J. (ed.), International Encyclopedia of Linguistics (2nd Ed.), pp. 210-212. Oxford: Oxford University Press. McCarthy, John & Prince, Alan. 1993b. Prosodic Morphology I: Constraint University of Ghana http://ugspace.ug.edu.gh 325 interaction and satisfaction. Ms., University of Massachusetts, Amherst and Rutger University. McCarthy, John & Prince, Alan. 1993a. Generalised Alignment. In Booij, Geert E. & J. van Marle (eds.), Yearbook of Morphology 1993, pp. 79-153. Dordrecht: Kluwer. Mensah, E.N.A. 1977. Problems of palatalisation in Akan. Papers in Ghanaian Linguistics 11, 64-76. Nartey, Jonas N.A. 1976. Phonological adaptations of English loan-words in Adanme. An unpublished M.A. research paper, Dept. of Linguistics, Southern Illinois University, Carbondale, Illinois. Osam, Emmanuel K. 2005. Akan word classes. Ms, University of Ghana, Legon. Owulah, Francis T. 2014. Aspects of Dangme phonology. MPhil thesis, University of Education, Winneba. Owusu-Adjei, Akua. 2007. Nativisation of Akan loanwords in Ga. BA long essay, University of Ghana, Legon. Paradis, Carole. 1996. The inadequacy of filters and faithfulness in loanword adaptation. ROA#157 Paradis, Carole. 1995. Derivational constraints in phonology: Evidence from loanwords and implications. In Proceedings of the Chicago Linguistics Society 31, pp. 360-374. Chicago: Chicago Linguistics Society Chicago. Paradis, Carole, & LaCharité, Darlene. 1997. Preservation and minimality in loanword adaptation. Journal of Linguistics 33(2), 379-430. Paradis, Carole & Tremblay, Antoine. 2009. Nondistinctive features in loanword University of Ghana http://ugspace.ug.edu.gh 326 adaptation: The unimportance of English aspiration in Mandarin Chinese phoneme categorisation. In Calabrese, A. & W. Leo Wetzels (eds.), Loan Phonology, pp. 211-224. Amsterdam/Philadelphia: John Benjamins Publishing Co. Peperkamp, Sharon. 2004. A psycholinguistic theory of loanword adaptations. In Annual Meeting of the Berkeley Linguistics Society 30(1), pp. 341-352. Peperkamp, Sharon & Dupoux, Emmanuel. 2003. Reinterpreting loanword adaptations: the role of perception. In M. J. Solé, D. Recasens & J. Romero (eds.), Proceedings of the 15th International Congress of Phonetic Sciences, pp. 367-370. Barcelona: Universitat Autónoma de Barcelona. Peperkamp, Sharon, & Dupoux, Emmanuel. 2001. Loanword adaptations: Three problems forphonology (and a psycholinguistic solution). Unpublished manuscript, Laboratoire de Sciences Cognitives et Pscyholinguistique, Paris & Universite de Paris, 8. Prince, Alan & Smolensky, Paul. 1993. Optimality Theory: constraint interaction in generative grammar. Ms., Rutgers University, New Brunswick and University of Colorado, Boulder. Cambridge, Mass.: MIT Press. Rose, Yvan. 1999. A structural account of Root node deletion in loanword phonology. Canadian Journal of Linguistics 44, 359-404. Rose, Yvan & Demuth, Katherine. 2006. Vowel epenthetic in loanword adaptation: Representational and phonetic considerations. Lingua 116, 1112-1139. Schachter, Paul & Fromkin, Victoria. 1968. A phonology of Akan: Akuapem, University of Ghana http://ugspace.ug.edu.gh 327 Asante & Fante. Working papers in phonetics 9, University of California, Los Angeles. Sey, Mary. 1990. A phonological analysis of Akan loanwords from English. Ms., Institute of Adult Education, University of Ghana, Legon. Silverman, David. 1992. Multiple scansions in loanword phonology: evidence from Cantonese. Phonology 9, 289-328. Simons, Gary F. & Fennig, Charles D. (eds.). 2017. Ethnologue: Languages of the World, Twentieth edition. Dallas, Texas: SIL International. Online version: http://www.ethnologue.com. Snider, Keith L. 1999. Phonetic realisation of downstep in Bimoba. Phonology 15(1), 77-101. Steriade, Donca. 2001. Directional asymmetries in place assimilation: a perceptual account. In Hume, E. & K. Johnson (eds.), The role of speech perception in phonology, pp. 219–250. San Diego (CA): Academic Press. Stewart, John M. 1983. Akan vowel harmony: The word structure conditions and the floating vowels. Studies in African Linguistics 14(2), 111-139. Stewart, John M. 1967. Tongue root position in Akan vowel harmony. Phonetica 16, 185-204. Stewart, John M. 1966. Asante Twi in the Polyglotta Africana. Sierra Leone Language Review 5, 111-115. Uffmann, Christian. 2013. Loanword Adaptation. In Honeybone, P. & J. Salmons (eds.), The Oxford handbook of historical phonology, pp. 644- 666. Oxford: Oxford University Press. Uffmann, Christian. 2007. Intrusive [r] and optimal epenthetic consonants. University of Ghana http://ugspace.ug.edu.gh 328 Language Sciences 29(2-3), 451-476. Uffmann, Christian. 2006. Epenthetic vowel quality in loanwords: Empirical and formal issues. Lingua 116, 1079-1111. Uffmann, Christian. 2001. Patterns of epenthesis (not only) in Shona loanwords. In Fery, C., A. Dubach Green, & R. van de Vijver (eds.), Proceedings of HILPS 5, pp. 193-211. Universitat Potsdam. van der Feest, Susanne V.H., Fikkert, Paula & Davis, Barbara L. 2016. Cross- linguistic differences in the perception of dorsal and coronal CV- combinations: Evidence from English and Dutch. In Scott, J. & D. Waughtal (eds.), Proceedings of the 40th annual Boston University Conference on Language Development, pp. 417-428. Somerville, MA: Cascadilla Press. Vendelin, Inga & Peperkamp, Sharon. 2006. The influence of orthography on loanword adaptations. Lingua 116, 996-1007. Wentum, Comfort. 1997. A lexical tonology of Ga. MPhil thesis, University of Ghana. Yip, Moira. 2006. The symbiosis between perception and grammar in loanword phonology. Lingua 116, 950-975. Yip, Moira. 2003. Phonology: Long-distance processes. In Frawley, William J. (ed.), International Encyclopedia of Linguistics (2nd Ed.), pp. 326-327. Oxford: Oxford University Press. Yip, Moira. 2002. Tone. Cambridge: Cambridge University Press. Yip, Moira. 1993. Cantonese loanword phonology and Optimality Theory. Journal of East Asian Linguistics 2, 261–291. University of Ghana http://ugspace.ug.edu.gh 329