University of Ghana http://ugspace.ug.edu.gh i SEGMENTAL AND PROSODIC STRUCTURES IN ƐSAHIE BY VICTORIA OWUSU ANSAH (10063507) THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA IN PARTIAL FULFILMENT OF THE REQUIREMENT FOR THE AWARD OF DOCTOR OF PHILOSOPHY IN LINGUISTICS DEGREE JULY 2019 University of Ghana http://ugspace.ug.edu.gh ii DECLARATION I, Victoria Owusu Ansah, declare that this research work has not been submitted, either in whole or in part, for the award of any degree elsewhere. All references used in the work have been duly acknowledged, and I am solely responsible for any error detected in the work. ………………………………………. ………………………… VICTORIA OWUSU ANSAH DATE CANDIDATE …………………………………………… ……………………………… DR. GEORGE AKANLIG-PARE DATE SUPERVISOR ……………………………………………… …………………………… DR. SETH ANTWI OFORI DATE SUPERVISOR …………………………………………… ………………………… DR. CLEMENT K. I. APPAH DATE SUPERVISOR University of Ghana http://ugspace.ug.edu.gh DEDICATION THIS THESIS IS DEDICATED TO MY HUSBAND AND CHILDREN, MY SIBLINGS, AND MY MOTHER THANK YOU FOR BELIEVING IN ME. University of Ghana http://ugspace.ug.edu.gh ACKNOWLEDGEMENT I am highly indebted to Jehovah God for helping me to complete this project, but for HIS mercies, this work will not have materialized. I am very thankful for how far He has brought me. I am also sincerely grateful to my able supervisors, Dr. George Akanlig-Pare, Dr. Seth Antwi Ofori, and Dr. Clement K. I. Appah. You were always there for me. Thank you very much for your efficient and effective supervision, timely and insightful feedbacks which propelled this work to its completion. You actually spurred me on. I recall times when I will send messages at odd times, but you were never angry and promptly responded. You were not just supervisors, but friends and mentors, and I have learnt a lot from you. God richly bless you. To my family, especially my husband and children, you do all! Like we casually say, thank you “seriously”. Mummy is really grateful for your understanding, sacrifices and prayers. Thank you, Kay for regulating me, letting me know when to rest and when to continue. I think I would have broken down if not for those professional, but lovely counsel. I am also particularly thankful to Professor Kweku Osam. This whole Ɛsahie concept started with your advice, and you never left me alone in the journey. I really appreciate your encouragement and great concern. University of Ghana http://ugspace.ug.edu.gh I am equally grateful to the entire faculty and staff at the Department of Linguistics, University of Ghana, especially to Professor Saah, Professor Agyekum, Dr. Hudu, Dr. Duah and Dr. Bubuafor. Your regular checkups and phrases such as, “how far; how is the work going; are you reading” etc. put me on my toes. God bless you. I am also grateful to the European Union through its Eramus+ Mobilty programme for the funding in completing this work. Further, to Professor Adams Bodomo of University of Vienna, and Dr. Saanchi, thank you for the resources and support during the Erasmus+ Mobility programme. It offered me more time and resources to complete this work. To all the staff at the department of African Studies, University of Vienna, especially the PhD students, Katharina Gartner, Immanuel Harrish, Dominic, the PhD peer review meetings proved very helpful. Thank you. I am indebted to my course mates, Patience Asare, Emma Sarah Eshun, Partrica Awo Amartefio, Winnifred Otoo, Anthony Klidza, and Asunka. You proved true the saying that ‘if you want to go fast, go alone, but if you want to go far, go together. You were the team I needed to go far on this project. God bless you. How can I forget the sacrifices of my language consultants and resource persons, especially, Mr. Evans Adu Gyamfi, who took time to provide insightful comments and feedback in person and through other social media platforms; to my people at the Bible Society of Ghana, Mr. Thomas Gyedu, and Mr. Wonderful Arthur, who introduced me to reliable people during my field trip, you made this work University of Ghana http://ugspace.ug.edu.gh possible, God bless you. To all my respondents at Sehwi, especially, Mr. Nelson of Sehwi Wiawso, Elder Gyedu of Sehwi Anhwiaso Methodist church, Mr. Anthony Ahi of the Sehwi Wiawso education office, Madam Efe of Sehwi Wiawso, and all the peole of Sehwi, God bless you all. To the staff, faculty, and Pro-Chancellor of Knutsford University College, THANK YOU very much for your numerous support. Finally, to all who believed in me and graciously contributed in diverse ways to complete this work, I say thank you very much and God bless you all. University of Ghana http://ugspace.ug.edu.gh ABSTRACT Studies show that one quarter of the world’s languages are spoken in Africa, yet most of these languages remain unstudied or not fully described (Blench 1998). This may have driven (Crystal 2000) to caution that most languages will be in extinction by 2020, hence the need for inter-generational transmission to help preserve them. One such means of preservation is documentation. Unfortunately, for Ɛsahie, a Kwa language with over 580,000 speakers (2010 National Population Census) located in the Western North Region of Ghana, not much documentation is available, compared to other languages such as Akan, Ewe, Ga and Nzema. Accordingly, using the qualitative research methodology, this study examines some phonological issues in the language. It investigates the principles that underlie and therefore regulate the organization of segments and prosodic units in Ɛsahie. The study discusses the inherent features of Ɛsahie speech sounds and the constraints on their distribution and sequencing, and establishes the repair strategies available in Ɛsahie for dealing with impermissible sound distributions and sequences. The study further examines the tonal structure of nouns and verbs, both in the basic forms and the non-basic forms. On nouns, the thesis establishes three tonal groups for basic nouns and examines the tonal structure of affixed nouns, deverbal nouns, and reduplicated nouns. The study shows that while nominal suffixes trigger tonal change in the base nouns, nominal prefixes do not. University of Ghana http://ugspace.ug.edu.gh For verbs, the thesis looks at the tone of inflected as well as reduplicated verbs to identify the tonal processes that occur when tones interact and how they influence lexical and grammatical patterns in the language. It establishes three tonal groups for un-inflected verb forms in Ɛsahie. It further shows that tone influences the inflection of verbs in the language and that tone assignment varies, depending on the verb’s tense, aspect or mood but there is no attendant segmental alternation in the lexical verb. The study is based on data from both primary and secondary sources. The primary data is from field trips to some selected towns in the language community, while the secondary data is from literature written in the language. This thesis is instructional as it will add to the linguistic information available on Ɛsahie and contribute to our understanding of the segmental and prosodic systems of African languages. University of Ghana http://ugspace.ug.edu.gh TABLE OF CONTENTS DECLARATION .........................................................................................................ii DEDICATION ........................................................................................................... iii ACKNOWLEDGEMENT .......................................................................................... iv ABSTRACT ...............................................................................................................vii TABLE OF CONTENTS .............................................................................................. i LIST OF TABLES .....................................................................................................vii LIST OF FIGURES .................................................................................................. viii LIST OF MAPS ........................................................................................................... x LIST OF ABBREVIATIONS ..................................................................................... xi CHAPTER ONE .......................................................................................................... 1 GENERAL INTRODUCTION .................................................................................... 1 1.1 Introduction ........................................................................................................ 1 1.2 Background of the Language and Ethnographic Information ............................ 1 1.3 Geographical Location of Sehwi .................................................................. 7 1.4 Previous Studies ......................................................................................... 13 1.5 Problem Statement ..................................................................................... 16 1.6 Objectives ................................................................................................... 17 1.7 Research Questions .................................................................................... 18 1.8 Methodology .............................................................................................. 18 1.9 Ethical Considerations ................................................................................ 19 1.10 Organisation of Thesis ............................................................................... 20 1.11 Conclusion .................................................................................................. 20 CHAPTER TWO ........................................................................................................ 22 THEORETICAL FRAMEWORK AND LITERATURE REVIEW ......................... 22 2.1 Introduction ...................................................................................................... 22 2.2 The Distinctive Feature Theory ..................................................................... 23 2.2.1 Classification of the distinctive features .................................................... 24 University of Ghana http://ugspace.ug.edu.gh 2.2.2 Relevance of the distinctive feature theory ................................................ 26 2.3 The Linear theory ............................................................................................. 30 2.3.1 Lapses of the Linear theory ........................................................................ 30 2.4 Autosegmental Theory ..................................................................................... 32 2.4.1 Basic Tenets of Autosegmental Phonology ............................................. 35 2.3.1.1 Universal Association Convention ..................................................... 36 2.3.1.2 Linkage Condition .............................................................................. 37 2.3.1.3 Obligatory Contour Principle ............................................................. 37 2.3.1.4 Well Formedness Condition ............................................................... 39 2.4.2 Principles of Autosegmental Theory........................................................ 40 2.4.3 Tiers of Representation .............................................................................. 42 2.4.4 Motivation for the Introduction of the Autosegmental Framework ........... 44 2.3.4.1 Floating Tones .................................................................................... 44 2.3.4.2 Tone Stability ..................................................................................... 45 2.3.4.3 Tonal Melodies ................................................................................... 45 2.3.4.4 Contour Tones .................................................................................... 46 2.4.5 Goals of Autosegmental Framework ......................................................... 47 2.4.6 Lapses of the Autosegmental Framework .................................................. 48 2.4.7 Summary .................................................................................................... 49 2.5 LITERATURE REVIEW ................................................................................. 50 2.6 Elision ............................................................................................................... 50 2.7 Reduplication .................................................................................................... 52 2.8 Prosodic Structure of Nouns ............................................................................. 57 2.8.1 Prosodic structure of nominalised words ................................................... 65 2.8.2 Borrowed Nouns ........................................................................................ 71 2.9 Prosodic Structure of Verbs .............................................................................. 75 2.10 Conclusion ...................................................................................................... 87 CHAPTER THREE .................................................................................................... 89 SEGMENT AND SEGMENTAL STRUCTURES IN ƐSAHIE ................................ 89 3. 1 Introduction ..................................................................................................... 89 University of Ghana http://ugspace.ug.edu.gh 3.2 Sounds in Ɛsahie: A descriptive Account ......................................................... 89 3.2.1 A review of Frimpong (2009) on consonant and vowel sounds in Ɛsahie . 90 3.2.2 Description of Ɛsahie Sounds: A Distinctive Feature Account ............ 101 3.2.2.1 Major Class Features ...................................................................... 102 3.2.2.2 Cavity/ Place of Articulation Features ......................................... 104 3.2.2.3 Tongue Body Features ..................................................................... 105 3.2.2.4 Manner of Articulation/Stricture ...................................................... 107 3.4 Sound Sequencing in Ɛsahie ........................................................................... 111 3.4.1 Vowel Sequence ....................................................................................... 111 3.4.1.1 Vowel Sequence across Word Boundary ......................................... 111 3.4.1.2 Vowel Sequence in Compounds ...................................................... 113 3.4.1.3 Summary on Ɛsahie Vowel Sequence .............................................. 115 3.5 Conclusion ...................................................................................................... 116 CHAPTER 4 ............................................................................................................. 117 THE SYLLABLE AND SYLLABLE STRUCTURE PROCESSES IN ƐSAHIE .. 117 4.1 Introduction .................................................................................................... 117 4.2 The Nature of a Syllable ................................................................................. 117 4.3 The Syllable Structure of Ɛsahie ..................................................................... 122 4.3.2 The CRV and CVRV Word Structure ..................................................... 123 4.3.3 Syllabic Distribution of Speech Sounds in Ɛsahie ................................... 127 4.3.4 The Ɛsahie Vowel Distribution ................................................................ 127 4.3.4.1 Vowels in CV Syllable Type ........................................................ 128 4.3.4.2 The V Syllable Structure .............................................................. 128 4.3.4 Ɛsahie Consonant Distribution ................................................................. 131 4.3.4.1 The Syllable Onset Position ........................................................ 132 4.3.4.2 The Syllable Nucleus Position ....................................................... 133 4.3.4.3 The Syllable Coda Position ............................................................ 134 4.3.4.5 Generalizations on Ɛsahie consonant distribution ........................ 136 4.4 Syllable Structure Processes ........................................................................... 137 University of Ghana http://ugspace.ug.edu.gh 4.4.1 Elision ...................................................................................................... 137 4.4.1.1 Vowel Elision .................................................................................. 138 4.4.1.2 [V₂] Elision in Compounds ............................................................. 139 4.4.1.3 V₁ elision in Possessive Constructions .......................................... 140 4.4.1.4 Elision in a Perfective construction ................................................. 142 4.4.1.5 Pre-sonorant High Vowel Elision .................................................... 143 4.4.1.6 Final /n/ Deletion .......................................................................... 146 4.4.1.7 Syllable Loss .................................................................................... 147 4.4.2 Summary .................................................................................................. 151 4.5 Syllable Structure Processes in English borrowed nouns in Ɛsahie ............... 153 4.5.1 Segmental changes ...................................................................................... 155 4.5.1.1 Environment Conditioned Voicing ....................................................... 155 4.5.1.2 ATR Synchronization ........................................................................... 157 4.5.1.3 Diphthong Replacement ........................................................................ 158 4.5.1.4 Unstressed Vowel Replacement ........................................................... 159 4.5.1.5 Consonant Substitution ......................................................................... 160 4.5.2 Syllable Modification ............................................................................... 161 4.5.2.1 Epenthesis ......................................................................................... 161 4.5.2.2 Syllable- onset Epenthesis /ʊ, ɪ / ...................................................... 163 4.5.2.3 Syllable-coda Epenthesis- /ʊ/, /ɪ/...................................................... 164 4.5.2.2 Deletion ............................................................................................ 166 4.5.2.2.1 Consonant Deletion ....................................................................... 166 4.5.2.2.2 Vowel Deletion ............................................................................. 168 4. 6 Conclusion ..................................................................................................... 169 CHAPTER 5 ............................................................................................................. 171 TONE AND TONAL PROCESSES IN ƐSAHIE NOUNS & VERBS ................... 171 5.1 Introduction .................................................................................................... 171 5.2 Pitch and Tone ................................................................................................ 171 5.2.1 Tone Bearing Unit in Ɛsahie .................................................................... 175 5.3 Tone of Basic Nouns and Verbs in Ɛsahie ...................................................... 176 University of Ghana http://ugspace.ug.edu.gh 5. 3.1 Tone of Basic Nouns in Ɛsahie ............................................................... 176 5.3.1.1Low tone Nouns ................................................................................ 178 5.3.1.2 LH tone Nouns ................................................................................ 178 5.3.1.3 High Tone Nouns ............................................................................. 179 5.3.2 Tone of Basic Verbs in Ɛsahie ................................................................. 179 5.3.2.1 Group I Verbs ................................................................................... 180 5.3.2.2 Group II Verbs ................................................................................. 180 5.3.2.3 Group III Verbs ................................................................................ 181 5.3.3 Tone in non-basic nouns in Ɛsahie ........................................................... 181 5.3.3.1 Tone in Affixation in Ɛsahie ............................................................. 181 5.3.3.1.1 Tone in Nominal Prefixes ......................................................... 182 5.3.3.1.1.1 Vocalic Prefixes ................................................................. 182 5.3.3.1.1.2 The Nasal Prefix ................................................................. 185 5.3.3.2 Tone in nominal suffixes in Ɛsahie .................................................. 187 5.3.3.2.1 The /-nɪɛ/ suffix ......................................................................... 187 5.3.3.2.2 The /-fʊɛ/ suffix ......................................................................... 190 5.3.3.2.3 The /-mɔ/ Suffix ........................................................................ 192 5.3.4 Summary .................................................................................................. 195 5.3.5 Tone in Derived Nouns in Ɛsahie ............................................................. 196 5.3.5.1 Tone in Deverbal Nouns .................................................................. 198 5.3.5.2 Tone in action nominals in Ɛsahie ................................................ 198 5.3.5.3 Tone of gerundive nominals ............................................................. 201 5.3.5.4 Tone of Product Nominals ............................................................... 204 5.3.5.5 Tone of participant nominals ............................................................ 206 5.3.5.6 Tone of nominalised nouns .......................................................... 208 5.3.5.7 Tonal Structure of Compounds .................................................... 210 5.3.5.7.1 Noun –Noun Compound ........................................................... 212 5.3.5.7.2 Noun-Verb Compounds ............................................................ 218 5.3.5.7.3 Noun-Adjective Compound ...................................................... 221 5.3.5.7.4 Verb-Verb Compound ............................................................... 221 5.3.6 Tone in Reduplicated Nouns .................................................................... 223 University of Ghana http://ugspace.ug.edu.gh 5.4 Tone in Verbal Forms ..................................................................................... 226 5.4.1 Tone of Inflected Verbs in Ɛsahie ............................................................ 226 5.4.1.1 The Perfective Verb ..................................................................... 229 5.4.1.2 Habitual Verb ................................................................................... 233 5.4.1.3 The Future Tense ............................................................................. 235 5.4.1.4 The Progressive Verb ....................................................................... 237 5.4.1.5 The Imperative Verb ....................................................................... 240 5.4.1.6 Past Tense Formation ...................................................................... 243 5.4.2 Tone in Reduplicated Verbs ..................................................................... 246 5.4.2.1 Tone in Reduplicated Monosyllabic Verbs .................................. 246 5.4.2.2 Tone in Reduplicated Disyllabic Verbs ....................................... 247 5.5 Conclusion ...................................................................................................... 248 CHAPTER 6 ............................................................................................................. 250 SUMMARY, FINDINGS, RECOMMENDATION AND CONCLUSION ............ 250 6.1 Introduction .................................................................................................... 250 6.2 Chapter Summary ........................................................................................... 250 6.3 Significance of the study ................................................................................ 253 6.4 Summary of Findings ..................................................................................... 254 6.5 Delimitations of the Study .............................................................................. 255 6.6 Recommendation for future research ............................................................. 255 6.7 Conclusion ...................................................................................................... 256 References ............................................................................................................ 257 University of Ghana http://ugspace.ug.edu.gh LIST OF TABLES Table 3 1: Summary of Vowel Distribution in Ɛsahie ............................................... 95 Table 3 2: Pronominal Prefixes .................................................................................. 96 Table 3 3 : Tense/ Aspect Affixes .............................................................................. 96 Table 3 4: Ɛsahie Consonant Chart ............................................................................ 99 Table 3 5: Summary of Ɛsahie Consonant Distribution ........................................... 100 Table 3 6: Distinctive Feature Matrix for Ɛsahie Vowels ........................................ 109 Table 3 7: Distinctive Feature Matrix for Ɛsahie Consonants .................................. 110 Table 3 8: Ɛsahie Vowel Sequence Chart ................................................................. 114 Table 4 1: Summary of the Syllable Distribution of Ɛsahie Consonants ................. 135 Table 4 2: [V₂] Elision in Compounds ..................................................................... 139 Table 5. 1: Nominalised Nouns ................................................................................ 209 Table 5. 5: Noun-Adjective Compound ................................................................... 221 University of Ghana http://ugspace.ug.edu.gh LIST OF FIGURES Figure 1. 1: Kwa Language Family Tree (Dolphyne & Dakubu 1988) ..................... 10 Figure 2 1 Voicing rule 29 Figure 3. 1: Ɛsahie Vowel Chart (Source: Frimpong 2009) ............ 94 Figure 4 1: Onset-Rhyme Representation of the Syllable ........................................ 119 Figure 4 2: Moraic Representation of the Syllable................................................... 120 Figure 4 3: Illustrates V₂ elision ............................................................................... 140 Figure 4 4: V₁ Elision in a possessive construction ................................................. 141 Figure 4 5: Pre-sonorant High Vowel Elision .......................................................... 145 Figure 4 6: Autosegmental representation final nasal deletion in 4.27 .................... 147 Figure 4 7: Autosegmental representation of syllable loss ...................................... 149 Figure 4 8: Autosegmental representation of environmental voicing change in example 4.30a. .......................................................................................................... 157 Figure 4 9: Consonant deletion in English borrowed nouns. ................................... 168 University of Ghana http://ugspace.ug.edu.gh Figure 5 1: Low Tone Spread in example 5.14 ........................................................ 189 Figure 5 2 : Low Tone Spread in suffixation of fʊ̀ɛ ́................................................ 192 Figure 5 3: Autosegmental representation of LT insertion and LTS in example 5.14b. ........................................................................................................................ 194 Figure 5 4: Autosegmental representation of example 5.39a. .................................. 226 University of Ghana http://ugspace.ug.edu.gh LIST OF MAPS Map 1. 1: Language Map of Ghana (SIL 2012) ........................................................... 4 Map 1. 2: Geographical Map of Ghana (pulse.com.gh) ............................................... 8 Map 1. 3: The Kwa Languages (Hartley 2005) ............................................................ 9 Map 1. 4: A Geographical Map Showing Ɛsahie Speaking Areas (Frimpong 2009) 12 University of Ghana http://ugspace.ug.edu.gh LIST OF ABBREVIATIONS 1 First Person 2 Second Person 3 Third Person ADJ Adjective ADV Adverb ATR Advanced Tongue Root C Consonant C₁ First Constituent C₂ Second Constituent DEF Definite Article DEM Demonstrative FUT Future H High HAB Habitual HNA Homorganic Nasal Assimilation HT High tone HTS High Tone Spread IMP Imperative L Low LT Low tone LTS Low Tone Spread N Noun NEG Negative NMLZ. PRFX. Nominalisation Prefix University of Ghana http://ugspace.ug.edu.gh NMLZ. SUFX Nominalisation Suffix NP Noun Phrase OCP Obligatory Contour Principle PERF Perfective PL Plural PRFX Prefix PROG Progressive IMP Imperative PST Past RED Reduplicated SF Surface Form SG Singular SUBJ Subject SUFX Suffix TBU Tone Bearing Unit UR Underlying Representation V Vowel V₁ First Vowel V₂ Second Vowel WFC Well Formedness Condition University of Ghana http://ugspace.ug.edu.gh 1 CHAPTER ONE GENERAL INTRODUCTION 1.1 Introduction The aim of this dissertation is to investigate the principles that underlie and therefore regulate the organization of segmental and prosodic structures in Ɛsahie by bringing to light the processes involved in the syllable and tonal structure of words in the language. This chapter introduces the thesis and is structured as follows; section (1.2) is on the background of the language and its ethnographic information, section (1.3) is on the geographical location of Sehwi, sections (1.4) review previous studies in the language, while section (1.5) presents the problem statement. In the rest of the chapter, we present the objectives of the study (section 1.6), the research questions guiding the study (section 1.7), the methodology (section 1.8), ethical considerations (section 19) and the organisation of the study (section 1.10). Section (1.11) concludes the chapter. 1.2 Background of the Language and Ethnographic Information History has it that the people of Sehwi once lived with other Akan tribes and groups before moving to their current location. Their migration was necessitated by war, famine and lack of farmland. Their first settlement was at Takyiman, then Afiena (present Wassa Amenfi) before they finally settled at Enuwomaso where they established the Sehwi state. Presently, pockets of people from every tribe in Ghana University of Ghana http://ugspace.ug.edu.gh 2 may be found in Sehwi, with Akans being the predominant group of settlers. Ɛsahie is the language spoken by the people of Sehwi. The word “Ɛsahie” is derived from the phrase ɛsa awie ‘war has finished’ (Daaku 1971; Kobiri 1988;2014; Ntumy & Boafo 2002b). The language is referred to in other circles as Asahyue, Samvi, Sehwi, and is coded in ethnologue as [ISO 639-3] with the identifier [sfw] ( Eberhard et al. 2019)1 History has it that the language evolved from Anyi, Bawule and Akan as a result of intermarriage (Kobiri 2014). Ɛsahie is a Kwa language and belongs to the Northern Bia language group in the Central Comoé or Tano subgroup (Dolphyne & Dakubu 1988). Speakers of Ɛsahie in Ghana number about 580,000 and they live mostly in the Western North Region of the country (Ghana Statistical Service Report 2012, 2010 National Population Census).2 The region is located within the tropical rain forest belt and is endowed with natural resources and has very fertile lands. It produces large quantities of cash and food crops, including cocoa. The Sehwis are mostly farmers and grow mainly cocoa. The region is also rich in gold deposit and it hosts the second largest gold mining company in Ghana – Bibiani Gold Mines, in addition to the Kyirano Gold Mines. Again, the region boasts of the only Bauxite mining company in Ghana, Awaso Bauxite. Though rich in natural resources, the literacy rate of Ɛsahie is 1 Sehwi is usually wrongly spelt Sefwi. This error is believed to be due to the absence of ‘hw’ in the orthography of the pioneer German writers of the language. 2 The number given here as speakers are indeed all people living in Sehwi. The population census did not include language identification. It is therefore very challenging to identify people who speak only Ɛsahie using the census report. University of Ghana http://ugspace.ug.edu.gh 3 very limited. According to the World Report (World Report 386-June/July), only 53.3% of the Ɛsahie population are literate either in English or a known Ghanaian language. Linguistically, Ɛsahie is proximate to Nzema, Ahanta, Brosa (Enchi), Chakosi and Sanvi (spoken in La Cote D’voire), as seen in the Language Map of Ghana in map 1.1. University of Ghana http://ugspace.ug.edu.gh 4 Map 1. 1: Language Map of Ghana (SIL 2012) . Ɛsahie is greatly influenced by Asante Twi and many Sehwis are bilinguals in Asante Twi and Ɛsahie. Some of them prefer to speak Asante Twi when they are University of Ghana http://ugspace.ug.edu.gh 5 outside of their home region. Until the introduction of FM stations in the Western North Region of Ghana, Ɛsahie was not used for any official or educational purpose. In schools and churches, Asante Twi books and scriptures were in use until the mid- 1990, when the Bible Society of Ghana (BSG) developed an orthography for Ɛsahie. Available linguistic studies on the language are also few, as compared to neighbouring languages such as Akan, Nzema, and Ahanta. In their discussion of Kwa languages, and focusing on Tano languages in Ghana, Bendor-Samuel and Hartell (1989) did not mention Ɛsahie (Sehwi), even though they mentioned sister languages like Nzema, Ahanta, Baule and Anyi. The Sehwi people distinguish between two varieties of Ɛsahie – the Anhwiaso variety and the Wiawso variety. The two varieties are mutually intelligible and considered the same language by the speakers of each variety. The minor differences between the two varieties are few vocabulary and phonological items, as exemplified in Table 1 below, which do not cause any intelligibility problems. Table 1 Example of Words with Dialectal Difference Anwiaso Wiawso English ebure ebunaen Charcoal binzua brenzua Male nzasrɛ nnasrɛ Towel nnalıɛ lalɪɛ Dream University of Ghana http://ugspace.ug.edu.gh 6 mmɔwerɛ boen Fingernails boni beni which one The Anhwiaso variety is spoken within Bibiani-Anhwiaso, Awaso and their surrounding communities to the extreme east of the area, which is the east of the River Subri, in towns such as Sehwi Bekwai, Sehwi Anhwiaso and Asawinso. The Wiawso variety, which is the major variety in use, is spoken in the west of River Subri. The dialect area stretches over the Sehwi Wiawso traditional area. Generally, the people believe that the Wiawso variety is the older and authentic variety. The Anhwiaso dialect has some influence from Asante because it shares a boundary with it. There is also Sanvi (Anyi), a dialect cluster of Sehwi spoken in La Cote d’Ivoire, which is mutually intelligible with Ɛsahie, though the people feel it is an Ivorian language and as such must not be encouraged. Explaining the reason for the mutual intelligibility, Kobiri (2014) reports that Anyi was once part of Sehwi, but was divided by the colonial rulers for political expediency; Anyi went to the French and Sehwi to the British. Data for this study is taken from the predominant Wiawso dialect. In terms of education, the colonial government in 1915 established the first school in Sehwi at Wiawso. The English Church Mission (Anglican) later started a primary school at Bodi and subsequently at Akontombra, Benkyema and other villages. The Methodist church started their schools at Bekwai, Bibiani, Anhwiaso, Adwoafua and Aferi. Ntumy and Boafo (2002b) reports that the Roman Catholic Mission established University of Ghana http://ugspace.ug.edu.gh 7 schools at Asafo, Bibiani, Anhwiaso, later spreading to several other places. Today, there are many schools in Sehwi, with each paramountcy having at least one government senior high school. Bibiani has one senior high school; Wiawso has two with one each at Asawinso, Bekwai, Juaboso, Akontombra and Debiso. The region also boasts of a teacher training college, the Wiawso Training College. 1.3 Geographical Location of Sehwi Sehwi is located in the Western North region of Ghana. Map 1.2 shows the towns and boundaries of Sehwi. Ashanti and Ahafo Regions border Sehwi to the north. To the southeast, it is bordered by Western Region and stretches to the Western part of the Central Region and lies approximately between latitudes 6º13' and 6º20' (Frimpong 2009; Ntumy & Boafo 2002b). The western boundary of Sehwi also extends from the Ghana-La Cote d’Ivoire border (approximately along the 6º00' latitude) and cuts across the eastern tributary of the Tano and the Subraw rivers, and then stretches eastwards towards the environs of the Ankobra (cf. map1.2). University of Ghana http://ugspace.ug.edu.gh 8 Map 1. 2: Geographical Map of Ghana (pulse.com.gh) Ɛsahie is a Kwa language belonging to the Northern Bia language group in the Central Comoé or Tano subgroup (Dolphyne & Dakubu 1988). The classification of University of Ghana http://ugspace.ug.edu.gh 9 the Kwa languages appeared in Greenberg (1948) when he created the Niger-Congo phylum where the Kwa languages belong (cf. map 1.3). Map 1. 3: The Kwa Languages (Hartley 2005) University of Ghana http://ugspace.ug.edu.gh 10 The language family tree in Figure 1 shows the language family of Ɛsahie and the related languages. The diagram shows two language families - Bia and Akan. Under the Bia language group, there was an initial split between Nzema and Ahanta on one hand, and Anyi and Baule on the other hand. Thereafter, Anyi, Baule and Chakosi split from each other. Anyi further split into Anyi (Aowin) and Sehwi (Ɛsahie) (Frimpong 2009). Proto Central Comoe ‘Proto-Tano’ Bia Akan (Nzema-Anyi-Baule) Bron –Wassa Asante- Akuapem- Fante Nzema-Ahanta Anyi-Baule Anyi Baule Chakosi (Anufo) Nzema Evalue Anyi (Aowin) Sehwi (Sanvi) Figure 1. 1: Kwa Language Family Tree (Dolphyne & Dakubu 1988) University of Ghana http://ugspace.ug.edu.gh 11 Administratively, Sehwi is divided into seven districts. These are Bibiani- Anhwiaso-Bekwai, Sehwi Akontombra, Juaboso, Essam Debiso, Sehwi Wiawso, Bia East and Bia West municipality. Politically, it has five constituencies: Bibiani- Anhwiaso-Bekwai, Wiawso, Juaboso, Essam–Debiso and Akontombra. In terms of paramountcy, Sehwi has three paramount areas – Anhwiaso, Bekwai and Wiawso. Some major towns include Yamatwa, Kaase, Asafo, Nsinsin, Osei Kwadwo, Bodi, Bekwai, Akontombra and Asawinso. Map 1. 4 shows the Ɛsahie speaking communities. University of Ghana http://ugspace.ug.edu.gh 12 Map 1. 4: A Geographical Map Showing Ɛsahie Speaking Areas (Frimpong 2009) University of Ghana http://ugspace.ug.edu.gh 13 1.4 Previous Studies As noted above, prior to linguistic documentation, T.Y Ennin had written a number of unpublished books and other materials, including ‘Me kyire wɔ Ɛsahie’ a translation of the New testament into Ɛsahie, set of primary school readers, a dictionary and spelling book (Kobiri 1988). The Bible Society of Ghana, Ghana Institute of Linguistics, Literacy and Bible Translation (GILLBT), and other religious organizations had also translated parts of the Bible into Ɛsahie, and also written other primers in the language. A set of primers, ‘Ɛsahie Kengalɛ 1-3’, was published by GILLBT and Bible Society of Ghana in 2000 to teach the natives how to read and write in Ɛsahie. This was followed by a translation of the New Testament of the Bible into Ɛsahie to be used in churches. The few published linguistic studies are by (Broohm & Rabanus 2018), (Broohm 2017), (Ntumy & Boafo 2002a). There are however other unpublished linguistic research works done on the language. These include (Broohm 2019) (Boateng 2017), (Andam 2017), (Broohm 2014), and (Frimpong 2009). Other anthropological publications such as ‘Me kyire wɔ Sehwi’ (Kobiri 1988;2014) are available. Kobiri (1988) highlights some rudiments of Ɛsahie. For instance, it mentions twenty-six (26) alphabets for Ɛsahie although phonologically, the language has more phonemic sounds than those outlined above. Ntumy and Boafo (2002a) conducted a sociolinguistic survey on the people of Sehwi. On the phonology of Ɛsahie, Frimpong (2009) provides a general description of the sound system and syllable structure in Ɛsahie. The work gives a sense of the University of Ghana http://ugspace.ug.edu.gh 14 sound segments and words in which these sounds occur. She identifies ten (10) phonetic oral vowels, four contrastive nasal vowels and twenty-four (24) consonants for the language. Frimpong further discusses some aspects of the segmental processes in Ɛsahie. Among them are vowel harmony, reduplication, elision and nasalization, where she provides a general description of these processes to show their presence in the language. While the work shows the presence of these phonological processes, it is not sufficiently detailed in showing the constraints and repair strategies used in handling the constraints. Important phonological features such as syllable structure processes and tonal processes are also not discussed. In this study, we build on Frimpong (2009), using additional data to establish specific cases of sounds constraints and how sounds are distributed at the syllable level. Also, in his discussion of information structures in Ɛsahie, Broohm (2014) provides a comprehensive description of how information is packaged in Ɛsahie. The work mainly discusses how focus and topic constructions are presented in Ɛsahie. It shows, among other things, that Ɛsahie combines both syntactic strategies (fronting and or clefting) and lexical strategies to signal information structure. He further shows that Ɛsahie has a rich inventory of lexical items/markers, which guide the interlocutor to various aspects of the context of the discourse. These markers include the focus marker yéyɛ́, and focus-sensitive particles like the inclusive markers kósó ‘also’ and pó ‘even’, and the exclusive markers ŋgóm ‘only’, déín and álà ‘just’. syntactically, a focus marker or a focus-sensitive particle (including yéyɛ́, or kósó, pó, ŋgóm, déín, and University of Ghana http://ugspace.ug.edu.gh 15 álà) immediately follows the constituent in focus and it can appear either at the left periphery of the sentence before a complement clause or after the object. Ɛsahie can encode focus through two strategies; pre-posing and clefting. With the fronting or pre- posing strategy, he explained that the constituent being focused is usually fronted to the left periphery of the sentence, the clause-initial position, and is immediately followed by the focus marker. He adds that to show that a fronted argument was actually moved from its canonical position; an overt copy of the argument in the form of an anaphoric pronoun is left in its canonical position. He, however, observes that the fronting strategy is not enough to encode focus, and that it obligatorily has to be complemented by a lexical strategy where the focus marker yéyɛ́ immediately follows the fronted constituent. In Ɛsahie, clefting involves placing a sequence of words within the structure beginning with ɔ-te ‘it is’. Boateng (2017) studies the morphological structure of nouns in Ɛsahie and identifies three noun formation processes in Ɛsahie – affixation, nominalisation and compounding. On affixation, his study shows that nouns in Ɛsahie are made up of stems and affixes, which are mainly prefixes and suffixes. He identifies three nominal prefixes [a-], [-ɛ] and [n-] and two nominal suffixes - [-niɛ], and [-fʊɛ] with different morphological functions. He further identifies five (5) noun classes using number as the criterion for the classification in the language. He summarises the five noun classes as V-/N-, Ø-/N-, Ø-/Ø-, [-niɛ] [-fʊɛ], and [-lɛ]. Phonologically, he identifies three phonological processes that are very productive in the language. These are homorganic University of Ghana http://ugspace.ug.edu.gh 16 nasal assimilation (HNA), vowel harmony (VH), and consonant mutation (voicing assimilation). While the work highlights some useful features of nouns in Ɛsahie, it makes some claims that are contestable. Contrary to the claim by Boateng (2017) that the nominalization of verbal forms is always done by the suffixation of [-lɛ] to the verb stem, we show in this study that nominals can also be formed by prefixation and parasynthesis (Broohm 2019). We establish that while [-lɛ] is one of the forms for affixation, deverbal nominals can also be formed by prefixing [-a] or the affixes [a- - lɛ] which must attach simultaneously to the verb root. Preliminary study for this work confirms Boateng (2017) claim that compounds can be Noun-Noun, Noun-Adjective and Noun-Verb. We add to his analysis by looking also at the prosodic changes inherent in the derivation process. 1.5 Problem Statement Studies show that one quarter of the world’s languages are spoken in Africa and that most of these languages remain unstudied or not fully described (Blench 1998; Polomé & Winter 2011). This may have driven Crystal (2000) to caution that most languages will be in extinction by 2020, hence the need for inter-generation transmission to help preserve them. Sadly, not much documentation is available on Ɛsahie linguistically, compared to such neighbouring languages as Akan, Ewe, Ga, Nzema and others. As noted above, Bendor-Samuel and Hartell (1989) overlooked Ɛsahie in their discussion University of Ghana http://ugspace.ug.edu.gh 17 of Kwa languages and Tano languages found in Ghana. It is against the foregoing that we study the Segmental and Prosodic Structures in the language to complement the existing literature. The study reviews the segments and segmental processes of the language, bringing to bear some syllable structure changes such as elision that take place when segments occur in morphological and syntactic structures in the language. It also discusses the tonal pattern of nouns and verbs and shows the tonal processes that occur when tones interact in the language. 1.6 Objectives Despite the linguistic in-roads made in the past years into Ɛsahie, the language continues to remain one of the under-described languages in the Kwa family. This inspired us to study the chosen aspect of the language. The objectives of the study are to identify: 1. The constraints in sound distribution and sound sequencing in Ɛsahie and the accompanying repair strategies. 2. The constraints on syllabification and some of the syllable structure repair mechanisms in the language. 3. The tonal structure of nouns and verbs in the language, and the tonal changes associated with noun and verb forms in the language. University of Ghana http://ugspace.ug.edu.gh 18 1.7 Research Questions To realise the objectives stated above, the study explores these questions regarding the language: 1. What is the nature of the sound system of Ɛsahie, and how are the sounds distributed and organized in the language? 2. What are the repair strategies available in Ɛsahie for resolving impermissible sound sequencing? 3. What are the permissible syllable types in the language and how are impermissible syllables dealt with in the language? 4. What is the tonal structure of Ɛsahie nouns; what tone changes exist in the formation of noun forms in the language? 5. What is the tonal structure of Ɛsahie verbs; what tone changes exist in the formation of verb forms in the language? 1.8 Methodology Data for the study came from both primary and secondary sources. The primary naturalistic data was elicited from native speakers in Sehwi, using ethnographic and stimuli methods. The tools used were the Ibadan Word list, SIL picture story, participant observation and focus group discussion. Forty people were consulted in the data collection, made up of twenty people from each dialect area, i.e. Anhwiaso and Wiawso. Four towns were selected in each dialect area. From each town, five University of Ghana http://ugspace.ug.edu.gh 19 respondents were selected comprising of two males and three females, young and old, aging between eighteen and sixty-five. All respondents were native speakers of Ɛsahie. However, speakers of Ɛsahie were excluded from the study, if they were non-natives of Sehwi. An unstructured interview guide was also used. These un-structured interviews were used as follow-ups on the elicitation. The interviews were conducted in Ɛsahie with the aid of a translator. The issues covered bordered on their life, general issues, festivals, historical events, list of animals, events, food, days of the week, months and birthdays and other salient issues. These were recorded using an audio recorder and transcribed. The transcribed data were crosschecked with four different native speakers for consistency and accurateness. Additionally, a copy of the Ɛsahie New Testament of the Bible and some literature in Ɛsahie were acquired for consultation. The work is largely descriptive, so the data recorded was transcribed and described. 1.9 Ethical Considerations According to Berg (2007), researchers owe professional and ethical obligations to the human subject and the real world they collect data from in order to honour and ensure confidentiality made to them. Supporting this claim, Bryman (2007), states that researchers cannot overlook ethical issues because it holds the integrity of the research they conduct. Following these arguments, ethical issues were given due consideration, as all the protocols of the study were subjected to institutional ethical review by the University of Ghana http://ugspace.ug.edu.gh 20 College of Humanities of the University of Ghana. Additionally, the study adheres to situational ethics embedded in the qualitative research tradition. Thus, we sought the consent of respondents to participate in the study and future related work. 1.10 Organisation of Thesis The thesis is organized into six (6) chapters. Chapter one introduces the subject of the study and provides information pertaining to the ethnographic background of the study, the geographical information of the study area, the statement of the problem, objectives of the study, research questions and the methodology used for the study. Chapter two highlights the theoretical framework used for the study and provides a review of relevant literature related to the study. Chapter three describes the segments in the language and examines some segmental processes in the language. Chapter four focuses on the syllable and syllable structure processes such deletion and insertion in the language. Chapter five focuses on tone and tone processes in noun and verb forms in the language. Chapter six provides a summary of findings and conclusions of the research and make recommendations for future research. 1.11 Conclusion In this introductory chapter, I have discussed the ethnographic and ethnolinguistic background of the people of Sehwi and Ɛsahie as a language. We have observed that although Ɛsahie has a considerable number of speakers, the language lacks formal University of Ghana http://ugspace.ug.edu.gh 21 usage. Other issues highlighted in the chapter are the objectives of the study, the methods used in data collection, and a review of previous studies on the language, among others. University of Ghana http://ugspace.ug.edu.gh 22 CHAPTER TWO THEORETICAL FRAMEWORK AND LITERATURE REVIEW 2.1 Introduction This chapter highlights the major theoretical and descriptive concepts that guide the discussion of the data in this work. It also presents a review of relevant literature on the subject matter of the present study. The approach to this study is descriptive. Nonetheless, conducting a description in a theoretical vacuum may be problematic since descriptions sometimes involve some amount of analysis which cannot be carried out without some minimal set of theoretical assumption (Casali 1995;1997). Therefore, the discussion is couched in a manner that is acceptable in the Generative Phonological theory (Chomsky & Halle 1968b; Kenstowicz & Kisseberth 2014; Kisseberth & Kenstowicz 1977). Section 2.2 focuses on the distinctive feature theory and its groupings. In Section 2.3, we discuss the Autosegmental theory with its principles of operation, while section 2.4 through to section 2.8 presents a review of some relevant literature for the study. The section 2.9 concludes the issues discussed in the chapter. University of Ghana http://ugspace.ug.edu.gh 23 2.2 The Distinctive Feature Theory The Distinctive Feature theory was outlined in (Chomsky & Halle 1968a) and modified in (Katamba 1989; Kenstowicz 1994; Kisseberth & Kenstowicz 1977). The theory is an aspect of Generative Grammar (Kisseberth & Kenstowicz 1977) and focuses on the inherent properties of speech sounds. Distinctive Features are the minimal constructive units that make up a sound. They are a set of articulatory and acoustic features that are used to define and distinguish between speech sounds. They have phonetic specifiability, in that they are able to determine the phonetic characteristics of a sound. They show that sound features are universal and are relevant for the description of sounds in all languages. (Fromkin et al. 2011: 278) defines it “as a feature that distinguishes one phoneme from another, hence a word from another word”. This shows that the distinctive features have functional relevance and can create meaning difference in a word. They can also explain the behaviour of sounds in a phonological structure. Some distinctive features are binary in nature. The binarism shows the presence (+) or absence (-) of a feature. Some of the features are, however, intrinsically unary, that is, they have a single value and specify only sounds that have them. The unary feature focuses on the active articulator used in the sound production. Below are the classifications of the features. University of Ghana http://ugspace.ug.edu.gh 24 2.2.1 Classification of the distinctive features The Distinctive features can be classified into five broad groups basically in terms of phonetic specifiability. These groups are Major Class Features, Place of Articulation Features, Manner of Articulation Features, Glottis, and Tongue Body Features. 1. Major Class Features The major class features define the distinctions between vowels and consonants in general and the roles they play in the structure of a language. They are sub-categorised into three groups namely: [+/- syllabic] which describes sounds that occur in the nucleus of a syllable and those that occur at the periphery; [+/- consonantal] used to separate consonants from non-consonants such as consonants from vowels and glides; and [+/- sonorant] which distinguishes sounds that do not exhibit voicing differences and. those that do. The latter are also called obstruents. 2. Place of Articulation Features These categorise sounds based on where they are produced along the vocal tract. The sub groups under this features are [+/-anterior] which describes sounds produced from the alveolar ridge to the lips and those produced behind the alveolar ridge. The rest are [labial] for sounds produced with the lips involved as active articulators, [coronal] for University of Ghana http://ugspace.ug.edu.gh 25 sounds produced with blade/front of tongue as active articulator, and [dorsal] for describing sounds produced with back of tongue as active articulator. 3. Manner of Articulation Features The manner of articulation features specifies types of stricture. They group sounds into [+/- continuant] for sounds produced with a continuous flow of the airstream and others that do not, [+/- delayed release] for sounds produced with a delay in the release of the airstream and others that do not, [+/- lateral] for sounds produced with the airstream exiting at the sides of the point of articulation and others that do not. [+/- Nasal]: sounds produced with the airstream exiting through the nasal cavity and others that do not. 4. Glottis The state of the glottis relates to phonation or the activities of the vocal folds. They distinguish between [+/- voice] for sounds produced with vibration of vocal folds and those without vibration of vocal folds, and [+/- spread glottis] for aspirated and non- aspirated sounds. University of Ghana http://ugspace.ug.edu.gh 26 5. Tongue Body/Root Features The tongue body features are primarily vowel features but also relevant to consonants. They show whether sounds are [+/- High] for height of tongue, [+/- Low] for height of tongue, [+/- Back] for part of tongue, [+/- Round] for the posture of the lips, and [+/- ATR] for the movement in the tongue root in speech production. 2.2.2 Relevance of the distinctive feature theory The distinctive feature theory has brought in its wake many advantages. It can resolve the inadequacies of the phonemic theory especially in differentiating between segments. By means of the features, differences in segments are clearly identified. The presence or absence of a feature can contrast the meaning of words. For instance, using the feature nasal, the meaning of the words in example (2.1) below are distinguished otherwise they would have been the same. Example 2.1 Nasal Oral a. ʨĩ́à ‘day’ ʨɪ́à ‘dog’ b. sṹà ‘to learn’ súà ‘a house’ In the examples above, the difference in the words is made possible by the feature nasal. Also, using this feature, speakers and even non-speakers of the language University of Ghana http://ugspace.ug.edu.gh 27 are able to distinguish between two words in the same context as shown in example (2.2) below. Example 2.2. Ama sṹà nwoma wɔ súà ne mu Ama learn- HAB book room DEF inside ‘Ama learns in the room’ The theory also enables sounds to be grouped into Natural Classes based on their shared features and general statements made about them. For example, [b, d, g] form a natural class of [+voice] stops. We can often refer to these and no others with just two features, [-continuant, +voiced]. The grouping of sounds into natural class shows that sounds in the same natural class can undergo similar phonological processes and have the same effect on sounds adjacent to them. Again, the Distinctive Feature theory makes it possible to give detailed description of sounds by breaking them into smaller components. If we take a sound like /a/, we can tell, using the feature matrix, that it is [+syllabic, -consonantal, +sonorant, +voiced]. Using this representation, we are indicating the phonetic characteristics of the sound /a/. For instance, because it is [+ sonorant] it allows free flow of air in the oral cavity. We can therefore deduce that the differences between segments are featural. For example, the difference between/a/ and /ã/ is the feature [nasal]. University of Ghana http://ugspace.ug.edu.gh 28 The feature theory further makes it possible, with a small number of features, to specify adequately the nature of speech sounds and their behaviour in structure. For example, we can describe these sounds /m, n, ŋ, ɱ, ɲ, ɳ, ã, õ, ĩ, ũ/ as [+Nasal] and predict that they nasalize adjacent segments. Additionally, the distinctive features are used to formulate phonological rules to explain phonological processes that occur in languages. Consider the example (2.3) below. Example 2.3 towaa ‘bottle’ → n-towaa → ndowaa ‘bottles’. In this example, the change in the input form of the word is the feature voicing, where the [-voiced] sound /t/ becomes [+voiced] /d/ due to the presence of the nasal consonant /n/. Using the distinctive features, we can explain that the phonological process taking place in example (2.3) above is a voicing assimilation. The phonological rule that explains the data in (2.3) is as follows: Voicing rule: [-voiced] → [+voiced] [+voiced] This same rule can be recast in the autosegmental way where features and segments are autonomous. In the case of the voicing rule, we can reorganize it as follow: University of Ghana http://ugspace.ug.edu.gh 29 Voiced tier +vd -vd +vd -vd +vd -vd +vd +vd → → → → Skeletal tier x x x x x x x x x Place tier n t n t n t n d n d Figure 2 1 Voicing rule These rules enable us to make general statements to explain several things that occur in a language. Further justification for the feature-based theory is in line with (Fromkin et al. 2011: 586) observation that: feature-based analysis of segments may facilitate the process of language acquisition as it gives the learner the means to generate novel segments sequences which conform to the sound pattern of the language. Hence, using the feature-based theory, it is possible to anticipate how speakers of a language will pronounce certain words. For example, using the voicing rule above, a speaker who may not be aware of the alternations discussed above would however be University of Ghana http://ugspace.ug.edu.gh 30 able to predict how a word as n+ talie will be pronounced without being told how. He will be able to predict the pronunciation of n+talie → ndalie ‘spoons’. 2.3 The Linear theory The Linear theory is a type of the Generative phonological theory which assumes that phonological descriptions are dependent on information from other linguistics levels and phonological rules map underlying representations onto surface pronunciations. The theory considers phonological representation as consisting of strings of segments and sounds as a feature bundle. The segments are strictly ordered, but the features are unordered. The Linear theory futher assumes that speech is produced in a sequential fashion and that all distinctive features of sounds are equal. Under this theory, speech segments (phonemes) are arranged in sequential order that convey a specific message as done in the distinctive feature theory. 2.3.1 Lapses of the Linear theory The Linear theory was fraught with some challenges. The theory could not account for the autonomy of tones. Under distinctive feature theory for example, tones were considered part of a segment that bears it. Hence, the deletion of the segment meant the loss of the tone too. It failed to give meaning to floating tones, which are independent from the vowels. Thus, it could not give meaning to and preserve tone stability. University of Ghana http://ugspace.ug.edu.gh 31 Also, contour tones were thought of as a combination or sequence of two simple tones. The Absolute Slicing Hypothesis (Goldsmith 1990; Katamba 1989) of traditional SPE-style phonology, regarded tone as a feature belonging to the vowel. Under it, a segment such as /â/ would be represented as: â +syll -con -back -ATR +H +L In this representation, tone is seen as a built-in feature of the segment. Contour tones behave as if they were a combination of two simple tones. The linear repressentation could, therefore, not account for contour tones especially on a single syllable. This is because under linear phonology, segments are allotted features. Each feature specification corresponds to one segment (Kenstowicz 1994; Kenstowicz & Kisseberth 2014). A feature has two values [±] and a segment can have only one value, so contour tones could not be assigned two values [+H, +L]. If [+H] and [+L] are built into the matrix, it will be contradictory because it will mean the vowel has a High tone and Low tone at the same time. University of Ghana http://ugspace.ug.edu.gh 32 Another challenge with the linear theory has to do with the feature arrangement. The linear theory fails to provide an organized arrangement of the features. However, as Clements and Humes (1995), Sagey (1986) indicate through the feature geometry theory, the features that make up the sounds are not arbitrarily ordered, but are organised in a hierarchical order shows the coordination between the features. Again, in the Sound Pattern of English (SPE) model, morphemes were represented as Underlying Representations (URs) that consist of units that are defined by distinctive feature matrixes; hence segments were thought of as bundles of features. These inadequacies of the linear repressentation called for an alternate way under generative phonology to resolve the challenges identified in the linear phonology, hence the Autosegmental theory. 2.4 Autosegmental Theory Autosegmental theory was introduced by Goldsmith (1976) as a framework which gives independent representation to segments and suprasegments. Autosegmental theory is an offshoot of Generative Phonology (Roca 2003) which had its foundations in the Sound Pattern of English (Chomsky & Halle 1968b; Goldsmith & Laks 2016). It came about due to inadequacies identified in the Linear Phonology model. In the Sound Pattern of English (SPE) model, morphemes are represented as Underlying University of Ghana http://ugspace.ug.edu.gh 33 Representations (URs) that consist of units that are defined by distinctive feature matrixes. One of the primary properties of the SPE model was the postulation that the deepest and most interesting principles of universal grammar would be found in the form and function of phonological rules. The segments were strictly ordered but the features were unordered. For instance, ‘The Absolute Slicing Hypothesis’ “claims that speech can be exhaustively sliced into segments which consists of unordered bundle of features which are linearly ordered” (Katamba 1989: 191). Phonology in this model was linear and supra segments like tone were properties of the sound segments. This means that anytime a tone-bearing unit was deleted, the tone was also deleted. The inadequacy of the linear representation in capturing the independency of supra-segments such as tone culminated in the formulation of Autosegmental Phonology. The theory recognized the autonomy of tone. In this theory, phonological features are not represented in a linear order, but in a graphical way that shows the relationship that exists between the features that make up sound segments and supra segments. As Goldsmith (1990: 137) observes, Autosegmental Phonology “is a theory of how the various components of the articulatory apparatus, i.e. the tongue, the lips, the velum are coordinated.” Katamba (1989: 190) adds that “in principle the various articulatory parameters such as aspiration, nasalization, voicing and tone are autonomous and the articulations that result from them are in principle independent”. The theory also considers how phonological rules change the organization of phonological representation. In its representation, two or more parallel tiers of University of Ghana http://ugspace.ug.edu.gh 34 phonological representation are posited. The tiers are anchor points for elements on the various tiers. Each tier carries a string of segments but differs in feature specification. Association lines link the separate tiers. As Goldsmith (1990) explains, the tiers and the association lines make up a chart and the tiers take their names from the feature found on them as shown in example (2.4) below. Example 2.4 Tonal Tier H L Segmental Tier b i a Each tier represents a sequence of gestures, viewed from an articulatory point of view, or a distinct acoustic transition, viewed from an acoustic point of view, (Goldsmith 1990). Though interconnected, the tiers are in principle autonomous. Segments are linked to tiers using ‘Association lines’, which are used to show simultaneity in time or co-registration. That is to say, the features and segments are realized at the same time and they occur concurrently (Goldsmith 1990). The autosegments are on independent tiers (tonal tier, segmental tier etc.). When phonological processes take place, segments may be modified. These modifications can be reduced to two processes: linking and delinking (i.e. addition or removal of an association line). It is University of Ghana http://ugspace.ug.edu.gh 35 for this reason that the Autosegmental theory provides ways of representing these idiosyncratic changes that take place on different tiers. 2.4.1 Basic Tenets of Autosegmental Phonology Goldsmith (1990) observes that the successful operation of Autosegmental theory hinges on a set of principles, including the Obligatory Contour Principle and the Well Formedness Condition. I present these in turn. Before that, I introduce the notation that are employed in the theory in example (2.5). Example 2.5 Autosegmental Notations for Rule Writing T V A vowel linked to a tone T A floating tone which is not linked to a vowel T Establishing a link between tone and vowel V V Free vowel slot not linked to any tone University of Ghana http://ugspace.ug.edu.gh 36 T V Delink the tone from the vowel 2.3.1.1 Universal Association Convention The Universal Association Convention (UAC) states that “when unassociated vowels and tones appear on the same side of an association line, they will be automatically associated in one-to-one fashion radiating outward from the association line” (Goldsmith 1990: 14). The UAC thus helps one to realize the relationship of the elements on each tier to the other. Such relationships are seen after applying the UAC. In other words, the UAC maps tones to the TBU one-to-one, left to right, as demonstrated with the Ɛsahie example in (2.6). Example 2.6 tìré ‘head’ Tonal Tier L H Skeletal Tier x x Segmental Tier t i r e University of Ghana http://ugspace.ug.edu.gh 37 2.3.1.2 Linkage Condition Goldsmith (1990: 53) observes that “a segment that is not linked to a position on the skeletal tier will not be phonetically realized”. All segments must, therefore be associated. This condition explains why floating tones are not phonologically realized unless linked to a skeletal position. 2.3.1.3 Obligatory Contour Principle First proposed by Leben (1973), the Obligatory Contour Principle (OCP) restricts tone association. This principle prevents two identical features from being adjacent to each other. The principle came about due to “the need to streamline some of the descriptive devices of the theory” (Abakah 2004b: 46). In Autosegmental theory, there was indeterminacy in the structural representation of segment and suprasegments resulting in representations like (2.7a). Thus, with the OCP, when sounds appear in succession to each other, they must differ in at least one tone feature. Where adjacent, the tones must be of different values. OCP is illustrated in (2.7b). University of Ghana http://ugspace.ug.edu.gh 38 Example 2.7 a. pέέ ‘likely’ * H H x x pɛ ɛ b. pέέ ‘likely’ H x x pɛ ɛ In explaining whether the identical autosegments will be automatically or universally reduced to one, Goldsmith (1990: 311) clarifies that “there are good and general reasons to expect that the OCP will appear to operate as a design strategy in the phonology of natural language; not because the principle is overtly present, but because it follows from basic considerations of how phonology is learned.” So, the idea of a single representation for identical and adjacent tones is, as Bota (2002) explains, to demonstrate that their production involves the same gestural activity. Though regulatory, the OCP has not been spared criticism over the place or otherwise of the principle in phonological theory, either as a rule or a principle University of Ghana http://ugspace.ug.edu.gh 39 (cf.Kenstowicz & Banksira 1999; McCarthy 1986; Odden 1986;1988a;1988b; 1995a: ,inter alia). To Roca and Johnson (1999), OCP goes beyond a rule or principle. It is rather a latent force that sparks off the application of rules and principle. They write, “the OCP is not a principle as such …but rather a latent force motivating some of the rules and principles of language, somewhat as the shifting of the earth’s inner matter motivates volcanoes to erupt: Shifting is not eruption” (Roca & Johnson 1999: 401). After applying the OCP to data from Akan, Abakah (2004b) corroborates Roca and Johnson (1999) assertion that the OCP is not a rule but a constraint. He explains that “… OCP is not a rule. It is a constraint whose obedience or violation does not affect the phonetic representation of a lexical item or tone or phrase. […] its violation does not neutralize the fact that, it is one and the same autosegment that runs over pertinent segmental anchors on the segmental tier” (Abakah 2004b: 22). He continues that “its obedience… does not change the phonetic shape of a lexical item or tone phrase, although it makes for an elegant presentation of phonetic facts” (Abakah 2004b: 22). These counter debates reinforce the notion that OCP is not a rule, but also confirms the utility of OCP as a unifier of segments with the same suprasegmental features. 2.3.1.4 Well Formedness Condition The Well Formedness Condition (WFC) governs the linking and association of elements on different tiers. The WFC states that: University of Ghana http://ugspace.ug.edu.gh 40 a. Each vowel must be associated with at least one tone. Tonal Tier H Segmental Tier d i ‘ to eat’ b. Each tone must be associated with at least one vowel. Segmental Tier H Tonal Tier g o ‘to dance’ c. No association lines may cross. * V V L H WFC provides sanity in the association of autosegments as associations are not done haphazardly. 2.4.2 Principles of Autosegmental Theory The application of the Autosegmental theory is governed by some principles. These principles regulate the theory. They aid in the successful application of WFC by solving any problem that might crop up from its application. The principles are mapping, dumping, and spreading, and are described and exemplified below. University of Ghana http://ugspace.ug.edu.gh 41 a. Mapping—Associate vowels with tones in a one-to-one fashion left to right until one runs out of tones or vowels. One-to-one mapping V V V T T T b. Dumping— If in applying (a), some tones are still free, that is unassociated, link them to the last vowel to the right. Dumping of remaining tones V V T T T T T c. Spreading—If in applying (a) some vowels are still free, link them to the last tone on the right. Spreading of remaining vowels V V V V T University of Ghana http://ugspace.ug.edu.gh 42 The spreading principle can be equated to one-to-many association; and the dumping to many-to-one association. 2.4.3 Tiers of Representation Some of the tiers that may be used in the representation of segments and autosegments are provided in (2.8). Skeletal Tier: Acts as an anchoring device for elements on various tiers. This is sometimes referred to as CV Tier or Timing Tier. Skeletal Tier The Skeletal tier acts as an anchoring device for elements on various tiers. This is sometimes referred to as CV Tier or Timing Tier. Example 2.8. Skeletal Tier x x Segmental Tier d a ‘to sleep’ Segmental Tier Carries the segments. Tonal Tier For the representation of tonal features ATR Tier Used to show the position of the tongue root in sound production; either pushed forward or retracted. Nasal Tier To represent nasal features Height Tier To present the feature ‘high’ University of Ghana http://ugspace.ug.edu.gh 43 Backness Tier To carry the feature ‘back’ Place of Articulation To represent features that specify the articulation point of a segment Phonation Tier Indicate whether a segment is voiced or voiceless The use of these tiers are illustrated in (2.9) below. Example 2.9 a. dor → Place of articulation tier +ant ŋ g o → Segmental tier C C V → Skeletal tier [+nas][-nas] → Nasal tier b. [+vd] → Phonation tier b u e [+ATR] → ATR tier University of Ghana http://ugspace.ug.edu.gh 44 2.4.4 Motivation for the Introduction of the Autosegmental Framework The autosegmental theory was inspired by certain issues, some of which are highlighted below. 2.3.4.1 Floating Tones Goldsmith (1990: 40) refers to floating tones as “segments which, at any given moment in the derivation are not associated with any vowel.” They come about when a tone is left after the deletion of a vowel linked to it. Under Linear phonology, tones were considered part of a segment that bears it. Hence, the deletion of the segment meant the loss of the tone too. However, the Autosegmental theory showed that tones are independent of the TBU (Clements et al. 2011; Hyman 2014a;2017a). When the TBU is deleted, the tone is left to float until it docks on an adjacent TBU. This is illustrated with this Akan data in example (2.10) (Leben 2006). Example 2.10 a. hɛn adan ---‘their house ’ H L H b. hɛn adan→ hέń ↓ ádáń ‘their house’ University of Ghana http://ugspace.ug.edu.gh 45 H L H From the example, the L tone of the prefix survives even when the segmental portion of the prefix assimilates to the following H tone. 2.3.4.2 Tone Stability Tone stability shows that the deletion of a segment does not affect the tone associated with it, and vice versa. The tone thus resists deletion. Goldsmith (1990: 40) remarked that, “if a segment was deleted on one tier, it may not affect a segment with which it was formally associated. The tone stays put until it is re-associated with the closest segment”. This accounts for the autonomy of tones. 2.3.4.3 Tonal Melodies Goldsmith (1990: 40) explains melody levels as “linguistically significant levels in the grammar which refers to just one or two features in the utterance”. Katamba (1989: 195) adds that “tone melody is concerned with languages where a given pitch configuration is linked to certain morphemes or words or grammatical constructions regardless of the number of consonant and vowel segments which they contain”. Katamba illustrates this point with an example from Mende, a language spoken in Sierra Leone adapted from (Leben 1973;1978). In the language, a monomorphemic University of Ghana http://ugspace.ug.edu.gh 46 word could have the following tone melodies: H, L, HL, LH, LHL irrespective of the number of syllables it contains as shown in example (2.11). Example 2.11 a. H—pɛ́lɛ́—house’ b. L—bɛ̀lɛ̀—‘trousers’ c. HL—ǹgílà—‘dog’ d. LH—fàndé—‘cotton’ e. LHL— nyàha - ‘woman’ He continues that it can be assumed and captured that there are given melodies on each class of words regardless of the number of segments present. Tone is melodic, hence autonomous. 2.3.4.4 Contour Tones The Linear Phonology was unable to account for contour tones especially on a single syllable. This is because segments are allotted features. Autosegmental representation came to the rescue of contour tone representation. (Goldsmith 1990: 143; Goldsmith & Laks 2016) explained, “tonal features are properties of another level; feature specifications on the other level constitute segments, but their relation to the vowels University of Ghana http://ugspace.ug.edu.gh 47 with which they are associated is merely one of simultaneity.” It indicated that contour tone is two tones on a segment which are produced in tandem and that one is not rooted in the other as illustrated in (2.12) below: Example 2.12 a. â— *a a b. a H L H L The illustration shows that in example (2.12a) /â/ was represented as a combination of two segments, /à/ and /á/ under the linear theory, but autosegmentally, it is represented as one segment with two tones/ â/ in example (2.12b). 2.4.5 Goals of Autosegmental Framework The goals of the Autosegmental framework can be summed up in the following: i. To depart from the linear representation where segments were thought of as bundles of features. ii. To give meaning to and preserve tone stability. iii. To give autonomy to tones. University of Ghana http://ugspace.ug.edu.gh 48 iv. To give representation to contour tones, which are perceived as multiple tones linked to a vowel. v. To make abstraction of tones from Tone Bearing Unit (TBU) possible. vi. To give meaning to floating tones, which are seen as independent from the vowels. 2.4.6 Lapses of the Autosegmental Framework Despite the effectiveness of the Autosegmental theory in tightening some of the loopholes in the linear theory, especially in respect of tone and other suprasegments, some scholars have identified some loose fitting ends in it (Katamba 1989; Leben 1973;1978; Leben & Ahoua 2006; Odden 1995b). Odden (1995b: 465) notes that “the strongest possible version of OCP at this point is that these may be a dispreference for adjacent identical tones; languages are free to express this dispreference by constraining lexical representation by adding rules of tone fusion or tone deletion, or by putting conditions on tone spreading rules. Ultimately, languages retain the portion of doing nothing about OCP violations”. Odden concludes that the Obligatory Contour Principle (OCP), which reduces both H and HH to the same melody is not applicable universally. Katamba (1989: 195) notes that tone stability is not a universal phenomenon. He explains that in languages where only heavy syllables can bear contour tones, the deletion of the tone bearing unit spells doom. This is because such languages allow for only one-to-one association of tones and TBUs. University of Ghana http://ugspace.ug.edu.gh 49 Despite these lapses, the usefulness of the theory can never be challenged as it formed the springboard for the formulating of the theories of lexical phonology, optimality theory, etc. (Pulleyblank 1986), Yip (1989b), (De Lacy 2002a; De Lacy 2002b), Zoll (2003) along with many others. Because tone and other phonological processes form an integral part of the analysis in this thesis, it is vital to describe the changes or differences in the process. It is in light of this that the autosegmental theory suffices. 2.4.7 Summary The review of the theoretical frame has expanded our understanding of the inherent features of speech sounds and how supra segmental features such as tone are distinctly treated from segments. As indicated earlier, this study is descriptive. However, the framework above will serve as a guide when linguistic theory becomes imperative in our data analysis. In the next section, we review some literature relevant to the study. University of Ghana http://ugspace.ug.edu.gh 50 2.5 LITERATURE REVIEW The study takes inspiration from researches done in other languages that have some relevance on the issues to be discussed. The literature review focuses on three major themes: (i) Segmental processes, (ii) Prosodic structure of nouns and (iii) Prosodic structure of verbs. On the segmental front, we briefly touch on Elision and Reduplication. 2.6 Elision Elision is the loss of a sound segment or a syllable in a word. The study of elision has been approached from varied angles in the literature. Matthews (1997: 11) indicates that elision occurs at word final position only when he writes that “elision is a process by which a vowel at the end of a word is lost, or elided, before another vowel at the beginning of a word.” This definition identifies just one context for elision and excludes other context of elision in other languages such as Ɛsahie. It also limits elision to vowels alone and ignores consonants. Data from Akan and other languages show that elision extends beyond word final to mid and initial positions and from vowels or University of Ghana http://ugspace.ug.edu.gh 51 consonants. Abakah (2004a: 182) therefore widens the scope and context when he describes elision “as a phonological process by which a vowel, a consonant and sometimes a syllable, which is an intrinsic property of a morpheme in an isolative style, is dropped in a combinative style”. Most languages employ vowel elision to break vowel hiatus (Orie & Pulleyblank 2002), however, there are differences in which vowel is elided and the context of elision. In some languages, the first vowel (V₁) is elided, whiles in others, the second vowel (V₂) is elided. For example, Casali (1997) identifies four contexts in which vowel elision is used to dissolve vowel hiatus in Etsako. These are: at the boundary between two lexical words (dɛ +akpa → dakpa ‘buy a cup’ & owa +ɔda → owɔda ‘a different house’); at the boundary between a lexical word and a functional word, where V₁ elision is more common than V₂. It can also be at the boundary between a CV prefix and a root, where he claims that V₁ occurs; and at the boundary between a root and a suffix, where either V₁ or V₂ elision is possible. Similar to Etsako, Abakah (2004a) shows that vowel elision involves the truncation of one of two contiguous vowel at word boundary. This occurs when a word that ends in a vowel is followed by another word that begins in a vowel. He explains that if the second vowel in the sequence of V₁ # V₂ is [-Low], then it is obligatorily deleted. However, if the first vowel is [-low], then the first vowel is deleted. Abakah (2004a) further shows that in compounding, word boundary vowel sequence may be deleted simultaneously under the condition that the final syllable of the first free form University of Ghana http://ugspace.ug.edu.gh 52 is CV#, where the C is [+Son] and the V₁ is [+high], with the # V₂ underspecified for tongue height position feature. He explains that regardless of the dialectal variations in Fante, the post sonorant word final deletion has to occur intervocalically at the underlying level of representation. Aside sound segments, Abakah notes that syllables are also deleted in rapid speech in Fante. These works are relevant to Ɛsahie as it shares similar context of deletion. Ɛsahie also deletes vowels at word boundary as well as truncating syllable in compounding. These contexts of operation will be studied in our discussion as we analysis the data on vowel elision. 2.7 Reduplication Reduplication has been analysed from different angles. Bodomo (2000: 3) explains it as ‘involving repetition or multiple occurrence of a morphological entity within a larger unit.” Inkelas (2014: 7) observes that “reduplication is capable of conveying derivational as well as inflectional meaning of any sort” and that “it serves a wide variety of functions cross-linguistically and within individual languages”. These functions, obviously, ranges from the standard morphological function of derivation to inflection. This result contributes to our understanding of the role of phonology in determining the realization of reduplicative morphemes. Our observation in Ɛsahie is that reduplication is used to mark plurality of subject, derivation of new words, attenuation, augmentation, emphasis, and particularly for encoding repetition of actions or concepts. Also, it has been shown that reduplication may be used to signal University of Ghana http://ugspace.ug.edu.gh 53 pluralization, iteration or duration, intensification and sarcasm (cf.Abakah 2015; Boakye 2015; Rubino 2005; Winkler & Obeng 2003: , inter alia). Many approaches on the study of reduplication treat it solely as a morphological concept. For instance, (Inkelas & Zoll 2000;2005) in their Morphological Doubling Theory (MDT) perceive reduplication “as the result of concatenation segmentally underspecifies shape by copying procedure” (Inkelas & Zoll 2005: 144). However, the works of Marfo (2005) Osam et al. (2013) and Abakah (2015), inter alia, show that reduplication is one area where phonological information informs morphological conclusions. Through their works, we notice that prosodic information such as tone is key in understanding the changes that take place in reduplicated forms. For instance, Abakah (2015) looks at the segmental and tonal process that reduplicants and the bases of nouns, verbs, and adjectives undergo. He notes that the difference between a base and a reduplicant can be deduce from output tonal patterns/melody. He further observes that reduplicated words from different tonal groups behave differently. For instance, nouns derived from class 1 verbs fails to be reduplicated, while nouns derived from class 2 verbs can be reduplicated. Moreover, he shows that nouns derived from verbal class 2 do not take nominal prefix, but in their reduplicated form, the plural morpheme is prefixed to the reduplicated word. While some verbs and nouns copy both the segmental and suprasegmental features of the base, others copy only the segmental features. Some of the example he gives are in (2.13). University of Ghana http://ugspace.ug.edu.gh 54 2.13 ɕìrá ‘bless’ → ɕìré- ɕìrà wìá ‘steal’ → wìé- wìà ʥí ‘to get’ → ʥì- ʥí Abakah’s claim is corroborated by Osam et al. (2013) when they examined the morphology of reduplicated verb stems in Akan and how morphological information affect target sounds. They show that in Akan, reduplicated verb forms where the full stem is reduplicated, one has to resort to phonological information such as vowel harmony, vowel shift or tonal processes to resolve the difference between the base and the reduplicant as exemplified below in (2.14). 2.14 a. bɔɔ ‘to hit’ → bòbɔ́ → bòbɔ̀- bòbɔ́ b. kùm̀ ‘to kill’ → kùǹ-kùm̀ → kùǹkùm̀ – kùǹkùm̀ (Osam et al. 2013: 46) They show that where the base and the reduplicant are identical, one resorts to phonological information to identify the differences in forms. This phonological information explains the changes in the similarities of the output form. Though reduplication precedes the phonological processes, the latter is key in understanding the inherent changes in the reduplicated forms. Thus, they may apply differently in University of Ghana http://ugspace.ug.edu.gh 55 different context, for example, vowel harmony may apply in some monosyllabic verb form but may fail to apply in disyllabic forms as shown below in (2.15). 2.15 sa ‘to scoop’ →sesa [sɪsa] twa ‘to cut’ → twitwa [ʨwiʨwa] sɔne ‘to seive’ → sɔnesɔne [sɔnɪsɔnɪ] They further observe that tone interacts with morphological forms in reduplication. For instance, a H tone monosyllabic base may surface as L tone in the reduplicated output. Based on these observations, they conclude that morphological changes in reduplication do not happen in isolation. They are highly influenced by prosody. I share their opinion that phonological impact is key in the reduplicated output. Indeed, there are more than enough evidence in Ɛsahie to support the claims identified in these works. Boakye (2015) also studies reduplication in Akan and identifies two types of reduplication, full and partial. He notes that like most languages, Akan reduplication employs copying as its strategy where nouns, adjectives, adverbs and numerals copy the base segments in the reduplicant without any change, but in the reduplication of verbs the height of the base vowel determine whether the vowel segments will be modified or copied as it is in the base. Although the work touches on many aspects of reduplication in Akan, it somehow fall short of elaborate thematic discussion. University of Ghana http://ugspace.ug.edu.gh 56 Some of the examples, for instance, are not elaborate enough to give the reader enough evidence to ascertain some of his claim. However, the work is relevant to this study considering the similar characteristics share by both languages in terms of reduplication. One key aspect of this study is on prosody and an aspect of it is treated in Downing (2011). On the effect of tone on reduplication, Downing (2011) observes that tones do not usually transfer along with the segmental content of the base in Bantu languages. She notes that tone realization in the RED+Base complex falls out if the complex is a compound verb stem. She compares tone realization in the RED+Base complex to accent realization in compounds in other languages and predicts a three way, three stems (RED stem, Base stem, compound stem) as its domain of realization. The analysis, she notes, confirms that both morphological and phonological factors determine the realization of reduplicative morphemes. Inagaki (2011) also discusses reduplication in Kadorih and shows that in Kadorih, a duplicated output (the whole structure derived by means of word duplication) is either a repeated or a reduplicated output phonologically. The repeated outputs are either phonological or intonational phrases, whereas reduplicated outputs must be prosodic words. He explains that a reduplicated output must be a single prosodic word, whereas a repeated output must be a phrase composed of two prosodic words. He shows that some prosodic words have two syllables, which can be regarded as appropriate inputs for reduplication in Kadorih. However, he explains that repeated University of Ghana http://ugspace.ug.edu.gh 57 outputs can receive one or two intonation units, and always express iconic meanings. He continues that monosyllabic inputs cannot be reduplicated. Furthermore, ‘extended prosodic words’ which consist of a clitic and another element, for instance, ngorihih ‘just drink’ (ngorih=ih) cannot be reduplicated either: *ngorihi-ngorihih. Thus, inputs for reduplication in Kadorih must have at least two syllables and must not be ‘extended’ through the addition of a clitic. Shiohara and Furihata (2011) investigates the function of reduplication in some Indonesia languages and notices that the infix [-ar-] occurs in only a limited number of nominals which can be recognized as an established and frequently used word and that other nominals may potentially host the infix [-ar-], but most of them are not recognized as an established word. On verbal bases, he notes that the infix [-ar-] usually indicates the plurality of (i) the actor of the situation, or (ii) the undergoer of the situation expressed by the transitive construction. These findings will enhance our understanding of the data on reduplication in the study. 2.8 Prosodic Structure of Nouns Several works confirm the impact of tone on nouns. The segmental and prosodic melodies of Akan was studied by Abakah (2004b) who makes several observations about segments and their prosodic melody in Akan. Among other things, he observes that Akan has optional nominal prefixes and nominal suffixes which play different morphological roles or functions. While the nominal suffixes are pre-linked for H tone, University of Ghana http://ugspace.ug.edu.gh 58 the prefix is underspecified for tone and takes an L tone by default. He notes that in Akan, tone is functional and a drop in pitch is usually noticed on H₂ which is separated from H1 by a linked or floating L tone at the prosodic surface. Akan, as he notices, does not have lexical contour tone but rather have phonetic contour tone which is the output of juxtaposing two contrastive tones. The study further examines the tonal pattern of verbs in Akan and revealed that verbs in Akan can be sub-divided into three tonal classes with prosodic behavoiur unique to each tonal group. These verbs, he notes, undergo different tonal alternations depending on the syntactic role it plays, i.e. whether marking negation, optative, habitual, stative or imperative. To explain the tonal changes, he postulates 14 tonal rules which account for the differences in tonal behavior. These include Floating L-Spread, Leftward H-Spread, Rightward L-Spread, Rightward H-Spread, Right-Edge Floating L-Spread, Perfect Aspect Verb Root Replacement, Past Tense Tone Replacement, and Right-Edge Floating H-Spread. Abakah’s examination of Akan nouns shows that prosodic patterns have immense influence on segmental behavior. Owusu (2015) corroborates Abakah’s claim on the tone patterns of possessive constructive in the Asante Twi dialect of Akan when he observes that the tone pattern of possessed nouns does not usually change in the possessive NP construction. He explains that “when the possessive noun is replaced with a possessive adjective, the tone of the possessive adjective is determine by the tone pattern of the possessed noun” (Owusu 2015: 98). For example, he notes that inalienable nominal prefixes are realised University of Ghana http://ugspace.ug.edu.gh 59 on low tone when used in isolation, but become downstep high in possessive constructions. Accounting for the tonal change, he explains that after the nominal prefix is deleted in a possessive construction, its low tone is left floating and causes a reduction in the pitch level of a following high tone to be downstep. He further explains that for nouns with L-H tone patterns, if the nominal prefix is low, the vowel in the possessive adjective is deleted if it is a pronoun (mi, wu, ni). This results in the nominal prefix being realised on a low tone as in these examples: Example 2.16 a. /mi/ + anumaa→ manumaa ‘my bird’ b. /mi/ + awuɔ → ma!wuɔ ‘my birth’ (Owusu 2015: 91) However, when the nominal prefix is a syllabic nasal, the possessive construction is realised on a high tone as in (2.17). Example 2.17 a. mi +mpa→ mi!mpa ‘my bed’ b. mi +nsa → min!sa ‘my drink. Similarly, Marfo (2004) provides a prosodic account of tone and segmental process in Akan phrasal words, and observes that there is an interface between prosody and syntax in phrasal construction in Akan. He specifically cites the output forms of University of Ghana http://ugspace.ug.edu.gh 60 noun-noun (N-N) and noun-adjective (N-Adj) phrases as evidence of the interaction between syntax and phonology. He argues that the application of phrasal rules in the construction are triggered by phonological contexts. His position is that “phrasal phonological rules apply only with properties of the prosodic domains, and syntactic constituents structure (along with other grammatical information) only constitutes the primary input base on which the domains of the prosodic structure are defined” (Marfo, 2013:94). He goes on to explain that in the construction of N-N and N-Adj, phrasal rules such as H-deletion, Prefix deletion, Diphthong simplification, H- insertion and boundary assimilation has to apply to support the syntactic components. For example, in the formation of N-N, and N-Adj phrasal words, the N1 loses its non- low stems to become high as shown in example (2.18). Example 2.18 a. òdwáí +òníní → òdwàníní b. àsɛ́ḿ + húnú→ àsɛ̀mhúnú (Marfo 2004: 96) As a diphthong, either the word is shortened or one of its vowels is lengthened as illustrated (2.19). Example 2.19 Diphthong Simplification a. kɔ̀téɛ́ + bɔ́tɔ́ ‘penis, sack’ → kɔ̀tèbɔ́tɔ́ ‘uncircumcised’ b. dàdéɛ́+ kétéwá ‘metal, small’ → dàdè(kété)wá ‘a nail’ University of Ghana http://ugspace.ug.edu.gh 61 (Marfo 2004: 98) Thus, the low tone of N₂ regressively spreads to replace the deleted H tone. It is interesting to note that similar processes such as H-deletion and subsequent L tone spread are observed in Ɛsahie compounds. Marfo’s study obviously is very relevant to our current study, especially in the area of N-N and N-Adj compounds. For Ewe, Capo (1981) shows that noun prefixes are marked for high tone when the stem begins with a sonorant consonant, but a low tone when the stem begins with a voiceless obstruents as in ā-ŋ́bā ‘floor’/ à-gbádzέ ‘reed seive’/ à-sǐkǐέ ‘tail’. Preliminary data analysis show that like Akan, prefixes in Ɛsahie are also underspecified for tone and take low tone by default. Chapters 5 of this study will examine the tonal behavior of nouns and verbs in Ɛsahie and these work will provide immense insight into the data analysis. In Digo and Nguni languages, Odden (1995b) reports that tone and consonants interact and voiced and voiceless consonants are treated alike. Gac (2002) investigates the tonal alternations of nouns in Somali and posits that tonal alternations are triggered by intonative tone. The work gives a phonological account of the alternations in the tone accent and show that the interactive tone interacts with the tone accent to yield different patterns of parameters. He showed that the tones of nouns undergo several complex changes according to the parameters of nominal class and syntactic discursive contexts. He argues that in hierarchical structure, the low tone of focalization is the key factor that triggers tonal accent University of Ghana http://ugspace.ug.edu.gh 62 changes and regulates intonative tones. Somali has two independent tonal tiers – the first tonal tier consists of high tones, and the second tier is intonative tier – which are made up of low or high tones preceded by high tones. Data from Buli also confirm tone-segment interaction in lexical forms. Akanlig-Pare and Kenstowicz (2002) analyse the tonal contrast and alternations in Buli from both the synchronic and diachronic perspective and report that tone is highly relevant in the inflectional morphology of the language. They observe that while nouns and adjectives exhibit all three lexical tone forms, verbs do not. They claim that in Buli, plural suffix ends with [-a] which, when inflected changes to a low, mid or high, as in (2.20). They further observe that the low tone of the emphatic forms of the 1st and 2nd person singular have a spreading effect on the attached nouns and may cause a high tone to become a rising tone as in the examples in (2.20). Example 2.20 a. bí:k → bísà ‘child’ b. bí:ká → bísáŋá ‘children c. pōk → pōbā ‘wife’ Example 2.21 a. bí:k ‘child’ m̀ biˇ:k ‘my child’ b. bísà plural m̀ bı́ ̀sá plural University of Ghana http://ugspace.ug.edu.gh 63 Kenstowicz (2008) studies the origin of tonal classes in Kinande noun stems and notices that the nouns in the language have four contrasting forms – HH, HL, LH, LL. Akinlabi and Liberman (2001) also provide further evidence for the relevance of OCP in morpho-phonology as they examine the tonal phonology of Yoruba clitics and identifies what they call a ‘morphophonemic conspiracy’. They identified this conspiracy after examining the tonal behaviours of six types of enclitics in standard Yoruba namely the subject marking high tone morpheme, the object pronouns, the emphatic particle, the short subject pronoun, the exclamatory or vocative particle and the reduced forms of the possessive pronouns. They observe that in all case of the enclitic, there is an OCP constraint that prevents the last syllable of the host and the adjacent clitic from having the same tone. Example 2.22 Object clitic Emphatic clitic ó kó mi ‘he/she/it taught me’ ó lo ọ ‘he went’ ó kó ehe ‘/she/it taught you’ ó lọ ò ‘he went (emphatic) ó dé ‘he arrived’ ó dé é ‘he arrived (emphatic) ó lé akin ‘he pursued Akin’ ó lé akin in ‘he pursured Akin (emphatic) (Akinlabi & Liberman 2001: 10-11) University of Ghana http://ugspace.ug.edu.gh 64 In these examples, the enclitic and the host carry the same tone which is a violation of the OCP. This is resolved by deleting one of the tones belonging to either the clitic or the host, or inserting a toneless vowel. While the constraint is observed in host-plus-clitic forms, it is inapplicable on other constructional forms such at lexical and phrasal levels or even in compounds. They propose that invoking OCP constraint prevents the last syllable of the host and the adjacent clitic from having the same tone. Akinlabi et al. (2009) study the tonal structure of Nkoroo nominal construction and observe the existence of ‘tonal conspiracy’. In the work, they examine the tonal pattern of nominal constructions in Nkoroo and proclitic-host with other types of noun phrases. They discover that regardless of the multiple phonological processes such as floating tone, tone spread, tonal metathesis, the tonal output is always the same. This, they call tonal conspiracy’. They pair compound of different word groups and of different tonal patterns. In all instances, they notice a sameness in tonal output. The only exception is when the first noun bears a low tone, in which case the ‘conspiracy’ is broken. They explain after pairing nouns with different tonal patterns and getting the same LH output that it appears the language is in need of a specific tonal melody on the output, so it maneuvers to achieve this output regardless of the input tones. This is quite significant to our current study as it brings to the fore the need to assiduously investigate nominal compound pairs to determine similarity in nature. University of Ghana http://ugspace.ug.edu.gh 65 2.8.1 Prosodic structure of nominalised words That languages derive nouns from other word groups, lexical and non-lexical, is widely studied (Akanlig-Pare 1999;2005; Appah 2004; Comrie & Thompson 2007; Gerner 2011; Koptjevskaja-Tamm 2002; Payne 1997; Yap et al. 2011: , inter alia). Although this study is not on the process of nominalization, but on the output of the process, knowledge on the process will aid our understanding of the product and their tonal pattern. Nominalization principally refers to the process by which nominal expressions are derived from other constructions (eat