VOWEL EPENTHESIS AND CONSONANT DELETION IN LOANWORDS: A STUDY OF AKAN By Kwasi Adomako Master`s Thesis in Theoretical Linguistics LIN - 3990 Department of Language and Linguistics Faculty of Humanities University of Tromsø Spring 2008 Supervised by: Dr. Martin Krämer TABLE OF CONTENTS Table of contents.............................................................................ii - iv Acknowledgement...........................................................................v Chapter 1 1.1. Introduction...........................................................................1-2 1.2. General discussion on loanword phonology.........................2-4 1.3. Speaker groups: Monolinguals compared with bilinguals....4-5 Chapter 2 2.1 Background information on Akan..........................................6 2.2 Segment inventory.................................................................6-10 2.3 Consonant- vowel combination.............................................10-12 2.4 Vowel harmony.....................................................................12 2.4.1 Stem harmony...............................................................13 2.4.2 Prefix harmony .............................................................13 2.4.3 Suffix harmony..............................................................14 2.4.4 Rounding harmony........................................................14-15 2.5 Syllable structure....................................................................15-19 2.6 The default vowel ..................................................................19-21 Chapter 3 Database and methodology...........................................................22-24 Chapter 4 4.0 The general repair strategies in Akan loanwords....................25 4.1 Vowel epenthesis.....................................................................26 4.1.1 Epenthesis into initial consonant clusters........................26-27 4.1.2 Epenthesis into medial consonant clusters ......................27 4.1.3 Epenthesis into final obstruent position...........................28 4.1.4 Epenthesis into word-final clusters ..................................28-30 4.1.5 Default vowel insertion....................................................30 4.1.6 The special case of palatals ..............................................30-32 4.2 Deletion ..................................................................................32-34 4.3 Vowel lengthening...................................................................34 4.3.1 Vowel lengthening after consonant deletion....................34 4.3.2 No vowel lengthening after consonant deletion...............34-35 ii 4.4 Non-native segment adaptation ..............................................35-36 4.5 Perceptual adaptation?.............................................................36-37 4.6 Interim summary.....................................................................37 Chapter 5 5.0 Formalization of major repair strategies in Akan loanword adaptation.............................................................................38 5.1 Segmental representations of place features............................38-56 5.2 Adaptation strategies in OT.....................................................56 5.2.1 Constraints on the quality of the epenthetic vowels.........56-61 5.2.2 OT on vowel epenthesis...................................................62-64 5.2.3 Skipping intervening consonants......................................64-68 5.2.4 Default vowel insertion.....................................................68-70 5.3 Vowel epenthesis into consonant clusters................................70-71 5.3.1 Vowel epenthesis into word-initial clusters ......................70-73 5.3.2 Vowel epenthesis into word-medial clusters.....................73-74 5.4 Consonant cluster reduction and vowel epenthesis in OT........75-82 5.5 Open issues...............................................................................82-84 Chapter 6 6.1 Akan compared with Sesotho and Shona..................................85-86 6.1.1 Akan compared with Sesotho ...............................................86-91 6.1.2 Akan compared with Shona....................................................91-95 6.2 The quality of the epenthetic vowels into clusters.....................95-96 6.2.1 Akan compared with Shona...............................................96 6.2.2 Akan compared with Sesotho............................................96-97 6.3 The quality of the epenthetic vowels in OT...............................97-106 6.4 Open issues, etc. Chapter 7 General discussions and conclusions..............................................107-109 References.......................................................................................110-113 Appendices: 1. Akan loanword corpus (dataset).................................................114-119 2. Consonant- vowel combination in Akan.....................................120-122 iii iv Acknowledgement First and foremost, I wish to express my deep appreciate to my supervisor, Dr. Martin Krämer for his patience, determination, and guidance that have seen me through this project. His comments and clarifications have been of immense help to me in writing this thesis. I also thank Dr. Christian Uffmann, Dr. Bruce Morén and the members of the CASTL Phonology reading group 2007/8 for their very helpful comments and suggestions towards this project. Second, I thank all who contributed in the data collection both in Ghana and in Norway. Special mentioning goes to my grannies, Nana Afia Nyarko, Nana Akua Afriyie, and Nana Ama Serwah, who doubled as part of my informants. Also I would like to acknowledge the contribution of my colleague Ghanaian students at the University of Tromsø who in diverse ways helped in my data collection. Finally, to my entire family and friends whose support, encouragement, and inspiration have seen my through all my education up to this level. v CHAPTER 1 1.1 Introduction The study of loanword phonology holds a key to our understanding of many phonological processes and theories. This field has over the years received some attention in terms of research and scholarship. Akan is one of the major African (Niger-Congo) languages that have enjoyed some appreciable amount of study especially in the areas on vowel harmony and reduplication. However, there is another area in Akan which has the potential to contribute immensely to the current research on loanword phonology, but has not yet been research into. Akan is observed to have extensive borrowings from many European languages, particularly English integrated into its vocabulary to the extent that hardly does a sentence or a phrase end without one noticing a loaned word being used. It is the aim of the current study to shed light on the loanword adaptation phenomenon in Akan, a majority language spoken in the West African country of Ghana. Akan has a strict syllable structure with preference for the basic open syllable structure (CV) or at its tolerating best, CrV shape (Dolphyne, 1988). This presupposes that foreign words with sequences of obstruents as in /st/, /sp/, /pl/, /sk/, etc. and word-final obstruents, being adapted into the language have to go through some repair processes to conform to this structural well- formedness requirement in the native grammar. The repair strategies that apply to these illicit- foreign words include vowel epenthesis, consonant deletion, etc. Among these strategies, vowel epenthesis is the dominant strategy that usually applies in the process. In this thesis, I discuss two of such repair strategies such as vowel epenthesis and consonant deletion in Akan. I begin the discussion on this subject by presenting the loanword corpus in Akan, showing the various major phonological processes that adapted forms (‘nativized’ source/foreign words) go through. I will make generalizations thereof based upon which I will present formalizations of such generalizations to be made. Regarding the issues of the quality of the epenthetic vowel in Akan loanword adaptation, it will be observed that the high vowels (both front and back) are the preferred epenthetic vowels based on the data to be discussed in this thesis. However, between the two high vowels, it is will be observed that the labial/round vowel is favoured in many environments as the epenthetic vowel contrary to the high front 1 vowel which is universally attested as the epenthetic vowel even in many cases in loanword adaptation. These contexts or environments include the epenthesis strategies such as the consonant to vowel feature spreading and vowel to vowel feature spreading, even across some intervening coronal consonants (except palatals), though consonantal assimilation is observed to be the preferred epenthesis strategy in Akan loanword adaptation. A reverse of the vowel harmony process by coronal vowels across an intervening labial consonant, that is, coronal vowels spreading across labial consonant is almost impossible, from the data. The analysis of the epenthesis strategies observed in the Akan loanword adaptation processes will be compared with what has already been done in loanword adaptations in other languages such as Shona (Uffmann, 2001, 2004, 2006), and Sesotho (Rose & Demuth, 2006) in chapter 6. It will be concluded that the three languages exhibit more similar patterning than they differ and that the major differences that exist between them are accounted for in the language-particular phonological system differences. In OT terms, the language-specific ranking of the same set of constraints, and in some cases, adding extra constraints accounts for the major differences in strategies employed by the three languages that will be compared in this thesis. Organization of the thesis: Chapter 1 will give a general introduction to the topic of this thesis with brief discussion on loanword phonology in general. Chapter 2 will give brief background information about the Akan language and its phonology with specific emphasis on its sound inventory and syllable structure. In this chapter also, I will discuss the different speaker groups of Akan loanwords including monolinguals and bilinguals who constituted my informants for the corpus being analyzed in this thesis. Their main difference lies in the way they adapt source words into their lexicon. Chapter 3 will summarize the database/corpus and the methodology I used in performing the study with discussion on issues such as the background of the informants used, the mode of data collection used, and how the loanword corpus was analyzed. The results of the study done will be analyzed in this chapter. 2 Chapter 4 will present the Akan loanword data and generalizations to be made about the data presented based on the main strategies used in repairing illegalities in the foreign/source words. The outcome of the generalizations to be made will be applied to the formalizations of feature representations in chapter 5. Chapter 5 is divided into two main parts. The first part discusses the formalization of the various repair strategies observed in Akan loanword corpus in chapter 4 through feature representations. For the second part, I will present the optimality-theoretic (OT) analysis of the general repair strategies found in Akan loanword adaptation which will later on in chapter 6, be compared with similar analyses already done in two Bantu languages such as Shona and Sesotho to determine the universality of the patternings in the loanword phonology. In Chapter 6, I present the general similarities and differences observed among three languages such as Akan, Shona, and Sesotho in both their native and loanword phonologies/grammars. I also discuss the issue of directionality effects found in Akan loanword adaptation where it will be observed that Akan exhibits two directions in its spreading of features to the epenthetic site. It displays right-to-left direction as its main directionality effect to repair initial and medial cluster illegalities, while it shows left-to-right directionality in spreading place feature to repair illegal codas. The latter directionality effect is also predictably attested in Shona and Sesotho in which spreading to all positions is basically unidirectional. I also do major comparisons of strategies in adapting from donor languages into their phonologies between Akan and Shona on the one hand, and Akan and Sesotho on the other hand. In addition to this, I will attempt an OT account for the vowel insertion into clusters (word-initial) observed in the three languages. Chapter 7 will summarize the findings to be made in the study and the general conclusions to the present work. 1.2 General discussion on loanword phonology In this chapter, I give a very brief discussion on loanword phonology in general with focus on a quick discussion of what researchers are arguing should be the approach or model in analyzing loanword adaptation. 2 The study of loanword phonology has gained some appreciable prominence in the field of phonology in general because of the crucial role it plays in our comprehension of many phonology processes and their interaction in particular languages and cross-linguistically, which leads to making phonological theorization. This assertion is supported by all loanword specialists with data from language-particular and cross-linguistic evidence. However, there is sharp division among experts in this field due to differences in opinion as to which approach/model is best in analyzing the loanword adaptation phenomenon. To date, three main different opinions have emerged as to which approach comprehensively accounts for the loanword adaptation phenomenon. One of the extreme sides simply postulates that from language-particular and cross-linguistic evidence, loanword adaptation processes are purely phonological/representational and that they have little or nothing to do with phonetics/perception. Notable researchers with this opinion are Jacobs and Gussenhoven (2000), Uffmann (2001, 2004, 2006), Paradis and La Charité (1997, 2005), etc. At the diametrically opposite side are views by such notable experts as Silverman (1992), Dupoux and Peperkamp (2003), Vendelin and Peperkamp (2006), Davidson (2007), Kenstowicz (2005), etc. that loanword adaptation is mostly or wholly influenced by phonetic/perceptual factors and as a result, it has very little or nothing to do with phonology/representation. The third opinion, which juxtaposes between the former and the latter views postulates that both phonology/representation and phonetics/perception play equally important roles in explaining loanword adaptation. Scholars such as Kenstowicz (2003), Heffernan (2005), Rose and Demuth (2006), Yip (2006), Kenstowicz and Suchato (2006), among others are prominent with this intermediate opinion on loanword adaptation model. In this thesis, I favour the third approach, which seeks to draw explanations from both representational and perceptual means of analyzing loanword adaptation and in this thesis, it will be realized, with the evidence from the loanword corpus from Akan, that neither pure phonology only nor pure phonetics only can adequately and comprehensively account for the all processes that loanwords undergo in their adaptation in a particular language or across languages. Rather, there is the need to draw explanations from both fields. The question that seems to separate those two extremist sides such as those who argue that either phonology or phonetics has a ‘minimal’ role to play is, how much of phonology is considered enough to 3