LEARNING TO EXPRESS MOTION EVENTS IN EWE BY FRANCISCA ADZO ADJEI (10046632) THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN PARTIAL FULFILMENT OF THE REQUIREMENT FOR THE AWARD OF PhD LINGUISTICS DEGREE. JULY, 2013 University of Ghana http://ugspace.ug.edu.gh i DECLARATION I declare that no portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning. References to other works have been duly acknowledged. ………………………… …………………………… Franscica Adzo Adjei Date Supervisors: ………………………… …………………………… Prof. K.K. Saah Date ………………………… …………………………… Prof. Akosua Anyidoho Date ………………………… …………………………… Dr. Evershed Kwasi Amuzu Date University of Ghana http://ugspace.ug.edu.gh ii DEDICATION To my family University of Ghana http://ugspace.ug.edu.gh iii ACKNOWLEDGEMENTS The completion of this thesis has been possible through the assistance and support of many people who set off upon this long journey with me as well as people whom I encountered along the way. After five years of countless hours of reading, writing, rewriting, procrastinating and finally writing again, this is it. A piece of work that asks for thanking people around me for teaching, guiding, encouraging, urging, and showing me that there is life beyond the PhD, and reminding me that I have a PhD to finish. Very well, I oblige. First of all, I would like to express my most sincere thanks to my supervisors, Prof. K.K. Saah, Prof. Akosua Anyidoho and Dr. Evershed Amuzu for their constant support, encouragement, guidance and understanding from the start up to the very end. I am most grateful for their “surgical” precision in “dissecting” and commenting on my work and their ability to keep just the right balance between providing helpful supervision and letting me make my own choices. A “huge” thanks to all the participants who kindly volunteered to take part in the study; the children and their parents in Akrofu and Sokode, akpé ná mi loo! Special thanks to the 2012 graduating Ewe class of the University of Education, Winneba. My thanks go to the staff of the Multimedia studio of the ICT Department of the University of Education, Winneba, for allowing me to use their facilities in recording the adult data. I must also acknowledge the contribution of Mr. Koto Odoku, a level 300 student of the Department of Art Education University of Education, Winneba for creating the picture book, Klo afí ka nèle? University of Ghana http://ugspace.ug.edu.gh iv My research was generously supported financially by the University of Education, Winneba and the American Council of Learned Societies (ACLS), who through their African Humanities Programme (AHP) secured a one year leave (2012-2013) for me from the University. ACLS also supported my research stay at the University of Ibadan, Nigeria. A big thank you to the Director of the Office of International Programmes of the University of Ibadan and her staff – Edith, Segou, Anuohuwapo, Mr. Falola, Mr. Effiong Akpan of the Linkhouse, as well as the head of the Department of African Languages and Linguistics, her staff and students for their administrative support during my research stay. I am also grateful to my Eck family in Ibadan, especially the Iboks and Mr. William Mejuru who made me feel at home, away from home. Many thanks to Mr. Aspect Caiquo and Ms. Felicia Takrama. Your suggestions and encouragement helped me decide I could take on this adventure. I am also grateful to Dr. Patrick Ekow Bankah (neorosurgeon) and Mr. Ulric Sena Abonie (physiotherapist) all of the Korle-Bu Teaching Hospital who showed me so much love and gave me the necessary treatment and attention when my spine nearly gave up because of the prolonged period of sitting up in the writing of this thesis. I must acknowledge the contribution of all the lecturers in the Linguistics Department of the University of Ghana and other linguists – Prof. Emeritas Mary Esther Kropp-Dakubu, Prof. Felix Ameka and Prof. James Essegbey who inspired me a lot with their selfless service. I am thankful. I must thank Mrs. Agatha Augustt for typing the work. University of Ghana http://ugspace.ug.edu.gh v Many thanks go out to my numerous friends, including Regina, Cecilia and especially Dr. Samuel Atintono who, though in Manchester University, always made sure I had access to the necessary and current materials for my work. My sincere thanks also go to my research assistants, Mr. Emmanuel Dogbe and Mr. Elvis Yevudey. Finally, but most importantly, I thank my family dearly for their support, love and unshakeable trust during all this time. My husband Peter, and my children (biological and non-biological) provided me with the necessary love, faith, strength and courage to go through all the stages of this journey. They were my best fellow travellers along the uphills, downhills and twisted paths. Thank you all. Mia wóé wɔ dɔ. Miawé sé ŋú. Akpe dzáá. Se ne tu tsi na mi. University of Ghana http://ugspace.ug.edu.gh vi ABSTRACT The focus of this thesis is to examine Ewe speakers’ linguistic organization of motion events and how such language patterns develop in Ewe- speaking children. The work is situated within Talmy’s Theory of Lexicalization Patterns (which examines the conceptual structure of motion events as well as the typological patterns in which this conceptual structure is parceled out in languages), Slobin’s Thinking-for-Speaking Hypothesis (which explores how particular typological properties will lead Ewe children to learn a particular way of thinking-for-speaking) and the Cognitive and Language-Specific Hypotheses. The cognitive hypothesis claims that children come to the task of language learning with a pre-existent cognitive representation of the world. In contrast, the Language-specific hypothesis claims that the language learning process is often under the semantic structure of the input language and that such influence begins from the very beginnings of language acquisition. Elicited production tasks with fifty 3-, 4-, 5-, 7- and 9 year olds (10 participants in each age group) as well as a group of 10 adults were carried out using three elicitation tools developed for research into motion expression. Findings of the study support the claim that typological properties constrain how speakers of Ewe talk about motion from early acquisition phases to adulthood. At age three, Ewe-speaking children used more path verbs than manner verbs in the expression of motion events. From four years onwards, they used the typical SVC constructions, a combination of Manner and Path verbs, to express motion events. They also mentioned only one piece of information about ground of movement in individual clauses. The children neither showed any University of Ghana http://ugspace.ug.edu.gh vii ability at describing the physical setting in which movement takes place (until 9 years of age) nor fully develop the narrative habit of describing complex motion events. Ewe-speaking children’s performance in motion event description has been found to grow gradually with increasing age and adult performance is always more extensive than that of children at any age. These results also suggest that while Ewe children follow equipollently-framed structural pattern when talking about motion events at a tender age of three, equipollently-framed discourse characteristics in Ewe-speaking children do not achieve maturity until adulthood. The thesis provides evidence for some possible early cognitive tendencies and the place of language specific hypothesis in language development. It also lends support to the typological categorization of Ewe within the Talmian and Slobin’s frameworks which can be used in other comparative studies in future research. University of Ghana http://ugspace.ug.edu.gh viii TABLE OF CONTENT DECLARATION .................................................................................................... i DEDICATION ...................................................................................................... ii ACKNOWLEDGEMENTS ................................................................................. iii ABSTRACT .......................................................................................................... vi TABLE OF CONTENT ..................................................................................... viii LIST OF TABLES .............................................................................................. xvi LIST OF FIGURES ......................................................................................... xviii ABBREVIATIONS IN INTERLINEAR GLOSSES .......................................... xx CHAPTER ONE .................................................................................................... 1 GENERAL INTRODUCTION .............................................................................. 1 1.0 Introduction ............................................................................................. 1 1.1 Aims of the Study .................................................................................... 4 1.2 Research Questions ................................................................................. 4 1.3 Methodology ........................................................................................... 5 1.3.1 Participants ....................................................................................... 5 1.3.2 Data Collection Tools ...................................................................... 7 1.3.3 Transcription Format and Data Analysis ......................................... 8 University of Ghana http://ugspace.ug.edu.gh ix 1.3.4 Tape Identification ........................................................................... 9 1.3.5 Reference to the Data ....................................................................... 9 1.4 Theoretical Considerations ...................................................................... 9 1.4.1 Talmy’s Typology ............................................................................ 9 1.4.2 Thinking-for-Speaking Hypothesis ................................................ 11 1.4.3 Cognitive Hypothesis versus the Language-specific Hypothesis .. 12 1.5 The Outline of the Thesis ...................................................................... 14 CHAPTER TWO ................................................................................................. 16 LITERATURE REVIEW..................................................................................... 16 2.0 Introduction ........................................................................................... 16 2.1 Talmy’s Work ........................................................................................ 16 2.1.1 Talmy’s Theory of Lexicalization and Motion Events......................... 16 2.1.2 The Motion Event .................................................................................. 18 2.1.3 Diversity of Patterns in Motion Expression .......................................... 23 2.1.3.1 The Three-Way Typology: The Verb Root .................................... 24 2.1.3.2 Lexicalisation Pattern: Motion + Co-event .................................... 24 2.1.3.3 Lexicalisation Pattern: Motion + Path ............................................. 26 2.1.3.4 Lexicalisation Pattern: Motion + Figure ........................................ 28 University of Ghana http://ugspace.ug.edu.gh x 2.1.3.5 Other Minor Patterns ...................................................................... 31 2.1.3.6 Split and Parallel Systems of Conflation ....................................... 32 2.1.4 The Two-way Typology: Path of Motion .............................................. 34 2.1.5 Typological Shifts ................................................................................. 38 2.1.6 Typology Revisited (A Third Category) ............................................... 40 2.2 The works of Slobin and others ......................................................... 42 2.2.2 Cline of Manner Salience ............................................................... 46 2.2.3 Cline of Path Salience .................................................................... 51 2.2.4 Linguistic Relativity and Slobin’s Proposal to Shift the Subject from “Thought and Language” to “Thinking and Speaking” ................................... 54 2.3 Cross-linguistic studies on motion events ............................................. 61 2.3.1 Introduction .................................................................................... 61 2.3.2 Thinking-for-Speaking Research ................................................... 61 2.3.3 Narrative Style ............................................................................... 62 2.3.4 First Language Acquisition ............................................................ 71 2.4 Linguistic Relativity Research .............................................................. 79 2.4.1 Positive Evidence ........................................................................... 80 2.4.2 Conflicting Evidence ...................................................................... 83 University of Ghana http://ugspace.ug.edu.gh xi CHAPTER THREE .............................................................................................. 88 FORMAL ENCODING OF MOTION EVENTS IN EWE ................................. 88 3.0 Introduction ........................................................................................... 88 3.1 The Language and Its Speakers ............................................................. 88 3.2 Phonology and Morphology .................................................................. 89 3.3 The Syntax of Ewe ................................................................................ 91 3.4 Ewe Verbal System ............................................................................... 97 3.5 Ewe Adpositions .................................................................................. 100 3.6 The Linguistic Encoding of Motion Events ........................................ 106 3.6.1 Introduction .................................................................................. 106 3.6.2 The Linguistic Encoding of Figure and Ground in Ewe .............. 106 3.6.3 Ground ......................................................................................... 107 3.6.4 Goal and Source ........................................................................... 110 3.6.5 Motion Interpretation from Non-motion Expressions.................. 113 3.7 Verbs that have a Motion Semantics ................................................... 116 3.8 Verbs that Express Direction ............................................................... 119 3.8.1 Verbs that Conflate the Fact of Motion and the Speed of Movement 119 University of Ghana http://ugspace.ug.edu.gh xii 3.9 Motion-cum Purpose Constructions .................................................... 119 3.10 The Linguistic Encoding of Path ......................................................... 120 3.10.2 Boundary Crossing Verbs ............................................................ 122 3.10.3 The Directional Complement Verbs ............................................ 124 3.10.4 Deictic Path Verbs in Ewe ........................................................... 125 3.11 The Linguistic Encoding of Manner.................................................... 125 3.11.1 Some Semantic Categories of Ewe Manner of Motion Verb ....... 126 3.12 The Linguistic Encoding of Cause of Motion ..................................... 131 3.13 The Place of Ewe in Motion Event Typology ..................................... 135 3.14 Summary ............................................................................................. 137 CHAPTER FOUR .............................................................................................. 139 MOTION EVENT EXPRESSION IN NARRATIVES OF THE CONTROL GROUP (ADULT EWE SPEAKERS) .............................................................. 139 4.0 Introduction ......................................................................................... 139 4.1 Methodology ....................................................................................... 140 4.1.1 The Control group and Data ......................................................... 140 4.1.2 Data Collection Procedure ........................................................... 141 4.1.3 Transcription ................................................................................ 141 University of Ghana http://ugspace.ug.edu.gh xiii 4.2 Coding Categories ............................................................................... 142 4.2.1 Motion Verb Categories ............................................................... 142 4.3 Results ................................................................................................. 150 4.3.1 Uses of Motion Verbs .................................................................. 150 4.3.4 Static Setting Versus Dynamic Movement .................................. 167 4.4 General Discussion and Conclusion ................................................... 169 4.4.1 The Use of Motion Verbs ............................................................. 169 4.4.2 Descriptions of Ground Elements per Individual Clauses .............. 184 4.4.3 Dynamic Movement versus Setting ............................................. 187 4.5 Summary of Chapter Four ................................................................... 191 CHAPTER FIVE ................................................................................................ 194 DEVELOPMENT OF MOTION EXPRESSIONS IN EWE-SPEAKING CHILDREN ....................................................................................................... 194 5.0 Introduction ......................................................................................... 194 5.1 Research Participants, Data and Method of Collection ....................... 195 5.1.1 Coding .......................................................................................... 196 5.2 Results and Discussion ........................................................................ 197 5.2.1 Learning to Use Motion Verbs ..................................................... 197 University of Ghana http://ugspace.ug.edu.gh xiv 5.2.2 The Development of Ground Descriptions .................................. 207 5.2.3 Developing the Narrative Habit of Describing Complex Events . 210 5.2.4 The Development of Static Description of Setting versus Dynamic Movement ................................................................................................... 211 5.3 General Discussion of Findings .......................................................... 213 5.3.1 Introduction .................................................................................. 213 5.3.2 Lexical Diversity of Motion Verbs .............................................. 214 5.3.3 Describing Ground Elements in Motion Events .......................... 221 5.3.4 Description of Complex Events ................................................... 228 5.3.5 Description of Dynamic Movement versus Setting ..................... 229 5.4 Summary of the Chapter ...................................................................... 230 CHAPTER SIX .................................................................................................. 232 CONCLUSION .................................................................................................. 232 6.0 Summary Of Main Points ................................................................... 232 6.2 Possible Directions for Future Research ............................................. 235 6.3 Application .......................................................................................... 237 BIBLIOGRAPHY .............................................................................................. 239 APPENDIX A .................................................................................................... 268 University of Ghana http://ugspace.ug.edu.gh xv LIST OF ADULT PARTICIPANTS ................................................................. 268 APPENDIX B .................................................................................................... 271 SAMPLES FROM CHILDREN’S TORTOISE NARRATIVES ...................... 271 APPENDIX C .................................................................................................... 348 WORDLESS PICTURE BOOK ........................................................................ 348 KLO AFI KA NE  LE? (TORTOISE WHERE ARE YOU)? ............................. 348 APPENDIX D1 .................................................................................................. 372 ELICITATION TOOLS ..................................................................................... 372 SLOBIN MANNER CLIPS ............................................................................... 372 APPENDIX D2 .................................................................................................. 374 ELICITATION TOOLS ..................................................................................... 374 RUN/WALK/CLIMB/CRAWL CLIPS ............................................................. 374 APPENDIX E .................................................................................................... 382 DATA COLLECTION PROCEDURE .............................................................. 382 APPENDIX F ..................................................................................................... 387 INFORMATION AND CONSENT FORMS .................................................... 387 University of Ghana http://ugspace.ug.edu.gh xvi LIST OF TABLES Table 1: Ewe Spatial Prepositions...................................................................... 102 Table 2: Simple Postpositions ............................................................................ 104 Table 3: Complex Postpositions......................................................................... 105 Table 4: Motion verbs used: Manner of Motion Verbs extracted from Adult Tortoise Stories .................................................................................................. 152 Table 5: Path Verbs extracted from Adult Stories ............................................. 153 Table 6: Neutral Verbs extracted from Adult Stories ........................................ 154 Table 7: Distribution of Verb Patterns in Ewe Tortoise Stories ........................ 155 Table 8: Distribution of Plus-Ground Clauses by the Type of Ground Elements ............................................................................................................................ 161 Table 9: Number of Event Segments Mentioned by Each Adult Ewe Narrator of the antelope Scene .............................................................................................. 163 Table 10: The Entire Collection of Motion Verbs in Ewe – Adult Tortoise Stories ............................................................................................................................ 171 Table 11: The Entire Collection of Motion Verbs in English and Spanish Languages .......................................................................................................... 172 Table 12: Proportion of Plus and Minus Ground Motion Expressions in Adult Spoken Narratives in Ewe, Mandarin, Spanish and English6 ............................ 185 Table 13: Description in English, Spanish, Chinese and Ewe ........................... 192 University of Ghana http://ugspace.ug.edu.gh xvii Table 14:Manner of Motion Verbs in Tortoise Stories: Frequency and Distribution ........................................................................................................ 199 Table 15: Path of Motion Verbs in Tortoise Stories: Frequency and Distribution ............................................................................................................................ 200 Table 16: Neutral Motion Verbs in Tortoise Stories: Frequency and Distribution ............................................................................................................................ 202 Table 17: Distribution of Different Motion Verb Combinations in SVCs ........ 205 Table 18: Development of Ground Elements..................................................... 207 Table 19: Number of Event Segments Mentioned by Each Ewe Tortoise Narrator in the Antelope Scene......................................................................................... 210 Table 20: Goal Source and Medium Clauses in Ewe Tortoise Stories .............. 224 University of Ghana http://ugspace.ug.edu.gh xviii LIST OF FIGURES Figure 1: Talmy’s (1985) typology of path encoding .......................................... 10 Figure 2: Satellite-framed languages ................................................................... 34 Figure 3: Verb-framed Languages ....................................................................... 35 Figure 4: Frequency distribution of manner, path and neutral expressions ....... 159 Figure 5: Frequency distribution of manner + manner ...................................... 160 Figure 6: Total Type of Frequencies of Different Motion Verb Categories for All Speakers of Each Age Group ........................................................................... 203 Figure 7: Token Frequencies of Different Motion Verb Categories for Each Group ................................................................................................................. 204 Figure 8: Total Token Frequencies of Different Verb Combination Types (in SVCs) for Each Age Group................................................................................ 205 Figure 9: Percentage Distribution of Plus Ground Clauses in Ewe Tortoise Stories ................................................................................................................. 208 Figure 10: The Number of Segment Mentioned for the Antelope Scene in Ewe Tortoise Stories .................................................................................................. 211 Figure 11: The Development of Static Description of Setting ........................... 212 Figure 12: Development of Plus Ground Clauses in English, Spanish, Chinese and Ewe .............................................................................................................. 227 University of Ghana http://ugspace.ug.edu.gh xix Figure 13: Mean Number of Event Segments Mentioned in the Antelope Scene by the Different Age Groups .............................................................................. 229 University of Ghana http://ugspace.ug.edu.gh xx ABBREVIATIONS IN INTERLINEAR GLOSSES ABL = Ablative ACC = Accusative ALL = Allative ALTRI = Altrilocal COMP = Complementizer COP = Copula cf = Confer DAT = Dative DEF = Definiteness marker DEM = Demonstrative DIM = Diminutive DIST = Distal FOC = Focus marker HAB = Habitual INDEF = Indefiniteness marker INT = Intensifier IT = Itive LOC = Locative LOG = Logophoric pronoun NEG = Negative pFOC = Predicate Focus PL = Plural marker POSS = Possessive POSTP = Postposition POT = Potential University of Ghana http://ugspace.ug.edu.gh xxi PRES = Present tense PROG = Progressive PROSP = Prospective PROX = Proximal Q = Question marker RED = Reduplicative REL = Relativiser REP = Repetitive RW = Run Walk Crawl Climb video Clips SG = Singular SUBJV = Subjunctive SVC = Serial Verb Construction TOP = Topicalizer TP = Topic marker TP = Terminal Particle UFP = Utterance Final Particle VENT = Ventive VT = Verb Transitive 1, 2, 3 = First-, Second -, and Third- persons Other Symbols e.g. = for example etc = etcetera i.e = that is klo = Based on a picture book, “Tortoise where are you?” University of Ghana http://ugspace.ug.edu.gh xxii lit = literally p.c = personal communication RW = Run Walk Crawl Climb video Clips SM = Slobin Manner Clips x;y = year; month S-languages = Satellite-framed languages V-languages = Verb-framed languages University of Ghana http://ugspace.ug.edu.gh 1 CHAPTER ONE GENERAL INTRODUCTION 1.0 Introduction The universal character of language in language acquisition has been discussed by examining the issue from multiple directions and dimensions. The most fundamental issue in the study of first language acquisition is to distinguish between two sources of structure and determine how they interact: the capacities and predispositions learners bring to the task themselves on the one hand, and a combination of properties of the language and the contexts (culture, etc) in which the language is used. This thesis takes up a domain that researchers have examined again and again – the expression of motion events. This linguistic representation of motion events and its impact on cognitive processes has been one of the main research topics explored in recent studies. Motion events are among the earliest and most basic and pervasive events in our lives (Johnson 1987, cited in Chen 2005, Chen and Guo 2009, 2010). They typically involve an entity moving from one place to another along a specified path. We move around the world everyday, we experience the movement of ourselves and others from the first day we are born and we talk about movement of objects and animate beings ever since we start to talk. Although the basic elements of motion events are universal around the world, different languages may verbally represent the same motion in different ways. Works on motion event expressions in the field of language acquisition have progressed considerably over the past three decades. Since the 1990’s, a University of Ghana http://ugspace.ug.edu.gh 2 growing number of studies have investigated how adults package and children learn the appropriate packaging of motion event components in their tartget language. These studies include Tzeltal (Brown, 2004), Chinese (Chen, 2005, Chen and Guo, 2009), Mandarin (Guo and Chen, 2010), English (Slobin, 1996a), American Sign Language (Galvan and Taub, 2004), English, Turkish and Spanish (Ozcaliskan and Slobin, 1999) Basque (Ibarretxe-Antunano, 2004), Korean and English (Choi and Bowerman, 1991), English and Spanish (Hohenstein, Naigles and Eisenberg, 2004) etc. To the best of my knowledge, no study has been done on how children express motion events in African languages. There have been some formal study on the acquisition of Bantu languages with phonological issues receiving the most attention. Some of these are Chimombo and Mtenje, (1989), Connelly, (1984), Demuth (1984, 1987a, 1987b, 1988, 1992, 1998, etc), Idiata, (1998), Kunene (1979), Lewis and Roux (1996), Suzman (1980, 1982, 1985, 1987, 1991, 1996), and Tsonope (1987). For the West African languages, a limited number of works on the acquisition of phonology can be found in the literature. These include Dangme (Apronti, 1968), Akan (Kumi, 1988, Dyson, 1980) and Ewe (Akordor, 2002). There is very little investigation of children’s use of syntactic and semantic constructions in the West African languages as well. The works that exist in this domain can be counted on the finger tips. These are Saah (1994), a two part study that combines the study of certain construction types like interrogative sentences, focus marking and relative clauses with the acquisition and processing of such structures by Akan-speaking children of a mean age of 5 years, four months. A second work of Saah (2002) shows that ‘by the age of 5/6, University of Ghana http://ugspace.ug.edu.gh 3 Akan-speaking children have acquired near-adult competence in the use of wh- in-situ questions and that they are also aware of the animacy restriction that regulate the use of overt and null 3rd person object pronouns in the language’ (Saah, 2002:230) For Ewe, three works have been identified in the literature. These are Adjei (2005), Adjei (in press) and Noyau (2002). Adjei (2005) explored Ewe-speaking children’s knowledge of Ewe colour terms to find out whether they can make syntactic and semantic distinctions between the basic colour terms. Adjei (in press) is concerned with how children use two modal constructions – an epistemic certainty construction and an Undergoer Voice construction. Also found in the literature for the acquisition of Ewe is Noyau’s (2002) work on the development of narrative strategy in bilingual Ewe and French children. This shows that the terrain on the acquisition of Ewe in general lies fallow. Against this background, this thesis examines Ewe-speakers’ linguistic organization of motion events based on these proposals. It carries out the discussion by looking at Slobin’s (1987) “thinking-for-speaking hypothesis (which calls to mind the Sapir-Whorf hypothesis) in which descriptions of events are coloured at least partly by the relative richness or paucity of the language components available to the speaker. One of the approaches suggested by Slobin and which suits my work is the stages at which children talk about experience in ways that appear specifically shaped by the linguistic system they are acquiring. The thesis finds out how such language patterns develop as Ewe-speaking children grow older. Another suggestion, that children are prepared from the beginning to accept linguistic guidance as to which distinction they should rely University of Ghana http://ugspace.ug.edu.gh 4 on in organizing particular domains of meaning (Bowerman, 1985) and how specific semantic patterns influence the way speakers of a language talk about motion, was also examined. 1.1 Aims of the Study This thesis has two goals. First, it examines adults’ use of motion expressions in order to determine the typological classification of Ewe. If the adult Ewe speakers’ speech show a mixture of characteristics typically associated with both verb-framed languages ( in such languages, the preferred means of expressing Path is a verb, with Manner expressed in a subordinate constituent) and satellite-framed languages, (those in which the preferred means of expressing Path (the core component of a motion event) is a non-verbal element associated with a verb) then it gives supporting evidence that Ewe is indeed an equipollently-framed language. Second, it examines the developmental path of Ewe-speaking children’s use of motion expressions, and uses it to explore the debate between the “language –specific” and “cognitive-hypothesis” (section 1.5.3) as well as Slobin’s thinking-for-speaking hypothesis. If the characteristics of language use by the youngest age group show similar patterns as those of Ewe-speaking adults, it will provide supporting evidence for the hypothesis or claims made above. 1.2 Research Questions To serve the above research goals, the following research questions were addressed: University of Ghana http://ugspace.ug.edu.gh 5 1. How are motion events expressed in Ewe? This question can be divided into four sub-questions: (i) how is path of motion expressed in a narrative? (ii) how is manner of motion expressed? (iii) how is a combination of manner and path of motion expressed?, and (iv) how are complex paths expressed? 2. How does the typological category of Ewe correlate with (i) the saliency of manner (ii) the packaging of multiple path and ground elements, and (iii) the elaboration of path in narrations? 3. Do the features that children acquire first in Ewe resemble those of satellite-framed languages, verb-framed languages, or equipollently-framed languages. 4. Do Ewe-speaking children engage in the activity of thinking-for- speaking (Slobin, 1996a) as they select and organize information in discourse? 1.3 Methodology 1.3.1 Participants Participants for the children’s data were recruited from Akrofu and Sokode in the Volta Region of Ghana through friends and acquaintances. University of Ghana http://ugspace.ug.edu.gh 6 Participants included 60 Ewe-speaking1 children at age 3, 4, 5, 7 and 9 with 10 participants in each group along with 14 adult native speakers (age range 29-48). However, for the children, only 50 (ten participants each from each year group) were selected for the analysis. The adults’ selection was based on dialect background to ensure that the major dialect areas were covered. Subjects for the child data were selected based on the following predetermined criteria: 1. Parental willingness for the child to take part in the study. 2. At least one parent must be Ewe-speaking and Ewe is one of the dominant languages spoken at home. 3. All the children were acquiring Ewe as their first language and parents report that none of them had any intellectual or hearing impairment, nor any history of a speech or language disorder. (See Appendix C for a copy of the Information and Consent Form for Parents, child assent letter and researcher’s introductory letter which were translated into Ewe and read to the parents and 5-, 7- and 9-year olds). There were equal numbers of males and females in all groups. Participants included both monolinguals and educated bilinguals (Ewe and English) because it is scarce to find monolinguals these days. Most parents send their children to preschools by age 4. (See appendix A for information on participants). 1 Ewe-speaking children are children who satisfy all the predetermined criteria listed below. University of Ghana http://ugspace.ug.edu.gh 7 Suggestions of Eisenbeiss (2005:117) were followed because the study sites are communities where large families are the norm. Eisenbeiss suggests that it is advisable not to record a comparatively large group of individual children and to obtain input samples for each of these children from their primary caretakers. Rather one should try to make recordings in a small number of families who come from different social backgrounds. In this way, less time is required for traveling between recording sites and setting up equipment and one needs to record fewer adults for an analysis of children’s input. In addition, the recording of all siblings provides a better basis for input analysis as they provide crucial information which is often neglected in studies where only the primary caretaker’s conversations with the target child of the study are recorded and other sources of input are ignored. Moreover, factors of social group are easier to control if the sample involves several groups of children who share the same social and family background. 1.3.2 Data Collection Tools I elicited data by using three elicitation tools developed for research into motion expressions. - 1) Klo, a fi ka nèle? “Tortoise where are you? This wordless picture book invites a rich collection of motion event descriptions and presents an excellent source for cross-linguistic study of motion event descriptions in connected discourse. The same set of events can be narrated by speakers of different languages, and consistent differences, if found, cannot be attributed to the stimulus. For example, if certain University of Ghana http://ugspace.ug.edu.gh 8 aspects of events shown in the picture are proved likely to be expressed by one language group but not by others, that means that those aspects are given special attention by speakers of one language in the course of verbalization while they are ignored by speakers of other languages. - 2) Slobin Manner Clips (Slobin 2002). This is a set of video clips which show people moving in different manners of motion. I showed these to four adult consultants and four children from four groups. I asked them to describe what happened in the videos. Data for the three year olds were discarded because they could not give any meaningful description. - 3) Run Walk Crawl/Climb – This is a set of video clips designed by researchers involved in motion encoding in language project at NTNU (Norwegian University of Scienece and Technology). They show moving people and animals. I used these clips in elicitation six times; three times with two adults and three times with a group of four children. 1.3.3 Transcription Format and Data Analysis The transcription conventions excerpted from Berman and Slobin (1994), was adopted. All the speech of the participants were orthographically transcribed by the researcher. Reliability was ensured by giving the recording to two graduate assistants with Linguistics background to transcribe. The various data analysis procedures were discussed in chapters four and five. University of Ghana http://ugspace.ug.edu.gh 9 1.3.4 Tape Identification The following information was stated on the tapes for the children before beginning the recording session: Code for village, Child’s name, Sex, Age (year; month) and date of birth. For the adults, the participant’s name, sex, age, hometown and dialect spoken were stated. (See Appendix A). 1.3.5 Reference to the Data In this thesis, the source of every Ewe example is indicated in brackets, following the English example translation. Sometimes the source is followed by an indication of the specific elicitation tool used. The following abbreviation are used. SM = Slobin Manner Clips RW = Run Walk Crawl Climb video Clips Klo - Based on a picture book, “Tortoise where are you?” 1.4 Theoretical Considerations 1.4.1 Talmy’s Typology The research on motion expression has widely been influenced by the typology of motion events proposed by Talmy (1985, 1991, 2000). He asserts that languages can be grouped together on the basis of how they encode the core information of a specific semantic domain onto syntactical and lexical structures. There are two distinct groups: those that allocate information of a specific semantic domain in the verb and those that do so in some other elements called satellites. The two basic types of languages recognized by Talmy are satellite- University of Ghana http://ugspace.ug.edu.gh 10 framed (where path is lexicalized as a “satellite” to a verb, as in ‘1a’) and verb- framed languages (where path is lexicalized as a semantic component of a motion verb) illustrated by ‘1b’, the Spanish equivalent of ‘1a’, (see figure 1 also). 1a. The bottle floated out b. La botella salio (flotando) the bottle exited (floating) ‘the bottle floated out’ Figure 1: Talmy’s (1985) typology of path encoding Verb framed pattern Verbal conflation e.g. salir ‘exit’ Figure Predication Manner Path Ground Verbal conflation satellite eg. crawl eg. Out of Satellite-framed pattern (Adapted from Levinson and Wilkins 2006:18) However, as more languages are studied, issues have arisen to challenge Talmy’s dichotomous typology. Brown (2004) reports that Tzeltan, a Mayan language, can express path of motion in both main verbs and directional satellites, and consequently, Tzeltal could be classified as either language type. Zlatev and University of Ghana http://ugspace.ug.edu.gh 11 Yangklang (2004) report that Thai, a serial-verb language, allows juxtaposition of two main verbs with equal grammatical status in one clause, with one verb expressing manner. To add to the problem, serial-verb languages are not a rare phenomenon but rather are represented by a wide range of language families, such as Niger-Congo, Hmong-Mien, Mon-Khmer, Austronesian, Tai-Kadai, and Sino-Tibetan (Zlatev and Yangklang, 2004). Slobin (2004) classified such languages as a third language type – the equipollently-framed languages – because path and manner are expressed by two linguistic forms that have roughly equal morphosyntactic status (see also Ameka and Essegbey, 2001, Lambert- Bretie re, 2009). In addition to Talmy’s typology, other theoretical considerations that are important to the thesis include Slobin’s Thinking for-Speaking Hypothesis and the Cognitive and Language Specific Hypotheses. These are discussed below. 1.4.2 Thinking-for-Speaking Hypothesis Slobin’s dialectical approach to language has led him to develop one of his current major theoretical positions concerning the psychology of language, the hypothesis of Thinking-for-Speaking, which he formulates as follows (Slobin, 1996a) : [T]he expression of experience in linguistic terms constitutes thinking-for-speaking – a special form of thought that is mobilized for communication. In the evanescent time frame of constructing utterances in discourse one fits one’s thought into available linguistic frames.” Thinking for speaking involves picking those characteristics of objects and events that (a) fit some conceptualization of the event, and (b) are readily encodable in the language. (p.76) University of Ghana http://ugspace.ug.edu.gh 12 According to this hypothesis, in the course of acquiring the grammar of a particular language, a person has to adopt a particular framework for schematizing experience Slobin (1991:71). Such language specific ways of the schematizing experience constitute what he calls thinking for speaking. Slobin’s hypothesis is that when we present events or experiences in language, we have to take a grammaticalized point of view in order to fit them to the structure of the language. Since each language has its own particular grammatical structure, people who speak different languages must take different grammaticalized points of view. Speakers are required to pay active attention to conceptual distinctions made by the grammar of the particular language they speak. For example, if you were a speaker of Turkish and were asked to describe a past event, you would have to specify whether the event was directly witnessed or not, since such a distinction forms a part of the grammar of Turkish. However, it is highly unlikely that an English speaker would even try to make such a distinction in describing the same event. In this thesis, I examined the rhetorical style (Berman and Slobin 1994, Slobin, 1996a, 2004) of the CONTROL GROUP’S (Ewe-speaking adults) use of motion expressions as a path or window to measure how children learn and use motion expressions. 1.4.3 Cognitive Hypothesis versus the Language-specific Hypothesis One of the key debates in child language development is between the Cognitive and the Language-specific hypothesis. The cognitive hypothesis (for general discussion of this position see Cromer, 1976; Bowerman, 1976, Clark, University of Ghana http://ugspace.ug.edu.gh 13 1977) claims that children come to the task of language learning with a pre- existent cognitive representation of the world. In the initial stages of language learning, children learn how to map the structures of the cognitive system onto the linguistic system. In contrast, the language – specific hypothesis (for general discussion of this position, see Brown, 1958, Bowerman, 1985, Gentner, 1982) claims that the language learning process is often under the “immediate influence of … the semantic structure of the input language” (Bowerman, 1985:1305), and such influence begins, as Bowerman argues, from the very beginning. She states: I argue that children are prepared from the beginning to accept linguistic guidance as to which distinction … they should rely on in organizing particular domains of meaning. (Bowerman 1985:1284). The central issue in this debate is not whether the cognitive or linguistic structures do or do not play any role in child language development, since people on both sides would agree that both ultimately play a role. The crucial question in this debate is when and how each of the two factors exerts its influence on language development. The research on motion events has opened out new research perspectives showing how these three theoretical approaches influence various activities including speaking, writing and translation, etc and how speakers of typologically distinct languages attend to different dimensions of motion event. There are many studies on language use and on the acquisition of lexical packaging, most of them based on Talmy’s typology of motion event encoding. These three theoretical approaches were examined into detail in Chapter 2 when revising the literature on motion expression. University of Ghana http://ugspace.ug.edu.gh 14 Other factors were also taken into consideration. Studies on grammatical and lexical development tend to focus on children below the age of six; and studies on phonological development often even start with the very earliest sound productions of children. In contrast, studies on narrative development typically involve older children. Thus, data from a broad age range is required to allow for different types of analysis. 1.5 The Outline of the Thesis The thesis is further organized into five chapters. Chapter 2 presents a detailed account of the typological frames in which Talmy describes motion events and summarizes the whole gamut of research influenced by the Talmian theory of lexicalization patterns for motion events. Such research will be grouped into two blocks: Slobin’s “Thinking-for-speaking” research (Section 2.8.2) and Linguistic relativity research. Chapter 3 provides a detailed description of the linguistic resources available for the formal encoding of motion events in Ewe and discusses the place of Ewe in the motion event typology. Chapter 4 examines the patterns of motion event descriptions by adult Ewe speakers in oral narratives elicited from Ewe speakers from different age groups using the wordless picture book Klo2 afi ka nele? ‘Tortoise where are you’. 2 The tortoise was selected in place of the frog because first, nobody in the African sense will keep a frog as a pet. Second, the frog carries negative connotations in the children’s University of Ghana http://ugspace.ug.edu.gh 15 Chapter 5 investigates Ewe children’s development of motion event expressions through the examination of motion event descriptions in tortoise stories produced by learners/children at ages 3, 4, 5, 7 and 9 and Chapter 6 summarizes the main findings of this dissertation, addresses possible directions for future research and suggests some applications of the work. environment. Any child who wets the bed is rolled up in his/her mat. A frog is tied at the edge of the mat and the culprit is carried amid singing and clapping and dumped in a stream at dawn. The frog is believed to heal the child of that “illness”. The psychological effect this has on the child, coupled with the shame, makes them to hopefully stop wetting the bed. Second, in the performance of certain purification rites, a frog and a chick are tied to a palm frond and dragged through people’s homes, footpaths and streets to purify the town/village and ward off evil spirits. The sight of such suffering creatures frighten children a lot and make them fear the frog. University of Ghana http://ugspace.ug.edu.gh 16 CHAPTER TWO LITERATURE REVIEW 2.0 Introduction This chapter reviews the typological frames of Talmy in the description of motion events as well as cross-linguistic studies on motion events. The chapter continues to highlight contributions that Slobin and others have also made to research on Motion events. 2.1 Talmy’s Work 2.1.1 Talmy’s Theory of Lexicalization and Motion Events Talmy (2000) delves into the exploration of the semantic relations in language between meanings and overt linguistic forms – in other words, he delves into the process of lexicalisation. “Lexicalisation is involved where a particular meaning component is found to be in regular association with a particular morpheme” (Talmy, 2000b:24). His basic assumption is that we can isolate elements or components separately within the domain of meaning and within the domain of linguistic expression. Then, the next step (a semanticist has to take) is to examine which semantic elements are expressed by which linguistic elements. Talmy remarks that this relationship is not one-to-one; a combination of semantic elements may be expressed by a single linguistic element, and a single semantic element may be expressed by a combination of linguistic elements. Moreover, semantic elements may be expressed by the same type of surface element, and the same type of semantic elements may be expressed by several different surface elements. For example, an English motion University of Ghana http://ugspace.ug.edu.gh 17 verb surface element can encode distinct types of semantic information: Manner of motion (e.g. hop), Cause (e.g. kick) and Path (e.g., exit, enter). On the other hand, the Path element may be encoded in English by verbs and by prepositions (e.g., out, into), that is, by two different linguistic elements. According to Talmy, by looking at the relations between meaning and linguistic forms, a range of universal principles and typological patterns might emerge. Talmy’s approach to lexicalisation (adapted from Talmy 2000b: 22; 2007:67) can be summarized as follows: a. Determine various semantic entities in a language. b. Determine various surface entities in the language. c. Observe which (a) entities are expressed by which (b) entities – in what combination and with what relationship – noting any patterns. d. Compare (c) - type patterns across different languages, noting any metapatterns (universal principles). e. Compare (c) – type patterns across different stages of a single language through time, noting any shifts or nonshifts that might be guided by a given universal principle (or a (d) – type metapattern). f. Consider the cognitive processes and structures that might give rise to the phenomena observed in (a) through (e). University of Ghana http://ugspace.ug.edu.gh 18 Finally, two directions or perspectives can be adopted for exploring meaning. One direction is to hold constant a selected surface or linguistic entity and observe which semantic entities are variously expressed by it, and the other direction is to keep a particular semantic entity constant and observe the surface or linguistic entities in which it can appear. The two typologies proposed by Talmy, the three-way classification (Talmy 1972, 1985) and the two-way classification of languages (Talmy, 1991, 2000b; cf. Cifuentes Férez 2008:24) result from adopting either of these two directions. In sum, Talmy’s concern is to find out whether, for a particular semantic domain, languages exhibit a wide variety of patterns, a small number of patterns (i.e., a typology), or a single pattern (i.e., a universal pattern). The semantic domain of Motion, (though his findings are also generalized to other domains) is one of the domains favoured by his research. He has examined the conceptual structure of motion events as well as the typological patterns in which this conceptual structure is parceled out. In the sections that follow some of these issues were addressed. 2.1.2 The Motion Event Talmy (2000b) defines an event as a portion of reality which has been delimited or bounded by the human mind. As he explains, ‘(t)he human mind in perception or conception can extend a boundary around a portion of what would otherwise be a continuum, whether of space, time […] and ascribe to the excerpted contents within the boundary the property of being a single unit entity’ (Talmy, 2000b: 215). Moreover, an event can be conceptualized as having a University of Ghana http://ugspace.ug.edu.gh 19 particular type of internal structure and degree of complexity. Thus, there are complex events, which are made up of a main event or framing event and a subordinate event or co-event (both of which are ‘conceptualised as unitary events), together with the relation that the co-event bears to the framing event. The framing event provides the schematic structure for the motion event and can be analysed into four components; (1) a moving figure, (2) a physical ground (i.e., a landmark) with respect to which the figure moves (3) an activating process, namely motion, and (4) a path that relates the figural entity to the ground entity. The co-event may take one of several forms, with the two most common forms being the manner event, which encodes the manner in which the motion is carried out (e.g., floating, running), and the causation event, which encodes the event originating the motion (e.g., kicking, throwing). Talmy considers a situation containing motion and the continuation of a stationary location alike as a motion event. In his own words: The basic Motion event consists of one object (the Figure) moving or located with respect to another object (the reference object or Ground). It is analysed as having four components: beside Figure and Ground, there are Path and Motion. The Path (with capital P) is the path followed or site occupied by the Figure object with respect to the Ground object. The component of Motion (with capital M) refers to the presence per se of motion or locatedness in the event […]. In addition to these internal components, a Motion event can be associated with an external Co-event that most often bears the relation of Manner or of Cause to it (Talmy 2000b: 25, 2007: 70-71). Let us illustrate this with the following example: 1. Sika walked slowly down the stairs. University of Ghana http://ugspace.ug.edu.gh 20 Sika is the Figure, the stairs is the Ground and down is the Path. The verb, to walk expresses simultaneously the fact of Motion (framing event) and the Manner of motion (Co-event). There are two types of motion found in motion events: Translational motion and self-contained motion. In Talmy’s (2000b:35) words, ‘’[i]n translational motion, an object’s basic location shifts from one point to another in space. In self- contained Motion, an object keeps its position, or “average” location. Self- contained Motion generally consists of oscillation, rotation, dilation (expansion and contraction), wiggle, local wander, or rest’. Let us consider examples (2), (3) and (4) to illustrate these notions. 2. Adzo entered the room. = translational motion 3. The aeroplane hovered over the town. = self-contained motion 4. Joan slid through the hall in her socks. = self-contained + transitional motion. Example (2) depicts the Figure’s change of location from the outside to the inside of the room. In contrast, example (3) shows self-contained motion; the Figure hovers over the town by gliding its wings. Sometimes, it is difficult to tell transitional and self-contained motion apart, as (4) shows. The manner verb to slide includes a component of friction, or rubbing (i.e., self-contained motion) between the Figure (Joan) and the Ground (the hall). However, such friction can only exist in the course of the Figure’s translational motion (through the hall). Thus, in English, the activity of self-contained motion has often come to be University of Ghana http://ugspace.ug.edu.gh 21 anchored in the framework of translational motion (Talmy, 2000b: 228-229, 2007: 79-80). Self-contained motion is intrinsically linked to manner of motion, in other words, to the Manner relation between the co-event and the motion event. In the Manner relation, the co-event co-occurs with the motion event, as we have seen in example (4), and is conceptualized as an additional activity that the Figure of the motion event exhibits. This activity directly pertains to the motion event but is distinct from it. However, the relation between the co-event and the motion event need not be limited to that of Manner, rather it can bear a wide range of relations (Talmy, 2000b: 42-47): a. Causal relation: ‘the co-event can precede the main Motion event in the case of onset causation, or it can co-occur with the main Motion event in the case of extended causation’ (Talmy, 2000b: 44-45). 5. Our tent blew down into the gully from a gust of wind – Onset Causation. 6. I squeezed the gel out of the tube = Extended causation. b. Precursion relation: the Co-event precedes the main motion event but does not cause or assist its occurrence. In the example below, the splintering of the glass preceded but did not cause the motion of the glass onto the carpet. 7. The glass splintered onto the carpet. c. Enablement relation: the Co-event directly precedes the main motion event and enables the occurrence of an event that causes the Motion but does not itself cause this Motion: University of Ghana http://ugspace.ug.edu.gh 22 8. I scooped jelly beans up into her sack. 9. Could you reach/grab that bottle down off the shelf? d. Reverse enablement: the Co-event named by the verb is an event that has previously taken place and that now gets undone. This new event, in turn, enables the main motion event named by the verb particle ‘auf’ in the following example: 10. Ich habe den Sack aufgebunden (German) Lit: I have the sack open-tied = I untied the sack and opened it.) e. Concomitance relation: the co-event co-occurs with the main motion event and is an activity that the Figure of the motion event additionally exhibits. But this activity does not in itself pertain to the concurrent Motion, that is, it could just take place by itself: 11. She wore a green dress to the party. 12. I whistled past the arena. f. Subsequence relation: the co-event takes place directly after the main motion event, and is enabled by, caused by, or is the purpose of that motion event. 13. I’ll stop at your office (on my way out of the building). 14. They locked the prisoner into his cell. University of Ghana http://ugspace.ug.edu.gh 23 2.1.3 Diversity of Patterns in Motion Expression Having looked at the definition of a motion event and examined the semantic components which underpin the conceptualization of this event and its varying relations with a co-event, we move on to deal with the characteristic lexicalisation patterns proposed by Talmy in detail. As we briefly mentioned in the introduction, the three-way classification of language (Talmy, 1972, 1985) and two-way classification (Talmy, 1991, 2000b) represent a different perspective taken on the relations between the semantic level and the morphosyntactic (linguistic) level. The former perspective kept constant a morphosyntactic constituent, the verb root, and looked at which semantic components were characteristically placed in it by various languages. It was found that most languages characteristically express either the Path, the co-event (the Manner or the cause), or the Figure in addition to the fact of Motion in the verb. In the later perspective, the semantic component of Path was kept constant, focusing the examination on which morphosyntactic constituent it was characteristically placed in by various languages. It was observed that most languages characteristically place the Path component either in the verb root (in verb-framed languages) or in the satellite3 and/or preposition (in satellite-framed languages). 3 Talmy (2000:102) defines satellite as the grammatical category of any constituent other than a nominal complement that is in a sister relation to the verb root. It relates to the verb root as a dependent to a head. The satellite which can be either a bound affix or a free word, is thus intended to encompass all of the following forms, which traditionally have been largely treated independently of each other. English verb particle, German separable and inseparable verb prefixes, Latin or Russian verb prefixes, Chinese verb complements, Lahu non-head ‘versatile verbs’ – Caddo incorporated nouns, and Atsugewi polysnthetic affixes around the verb root. The set of forms that can function University of Ghana http://ugspace.ug.edu.gh 24 2.1.3.1 The Three-Way Typology: The Verb Root This typology results from looking at which semantic components are characteristically lexicalized in verb roots by several languages. Three typological principal lexicalisation types for verb roots are presented by virtue of the verb root expressing either the Co-event (Manner or Cause), the Path, or the Figure in addition to the fact of Motion. Other conflations or minor patterns may exist within a language, as we shall see in the course of our discussion, although languages are categoriesd according to the most characteristic lexicalisation pattern they exhibit. In most cases, a language uses only one of these types for the verb in its most characteristic expression of Motion. According to Talmy (2000:27), ‘characteristic means that (1) it is colloquial in style, rather than literary, stilted and so on; (2) it is frequent in occurrence in speech, rather than only occasional; (3) it is pervasive, rather than limited – that is, a wide range of semantic notions are expressed in the type. 2.1.3.2 Lexicalisation Pattern: Motion + Co-event In one group of languages, including Finno-Ugric, Chinese, Ojibwa, Warlpiri and all branches of Indo-European languages (except for Romance languages), the verb typically expresses at once the Motion and a Co-event, as satellites in a language often overlaps partially, but not wholly, with the set of forms in anoher grammatical category in that language, generally the category of prepositions, verbs, or nouns. University of Ghana http://ugspace.ug.edu.gh 25 usually either the Manner or the Cause of the Motion. English is the protoypical example of this group. Below are English expressions of Motion with conflated Manner or Cause extracted from Talmy (2000:28, 2007) 4. Move + Manner i) Non-Agentive5 15 a. The rock slid/rolled/bounced down the hill. 15 b. The gate swung/creaked shut on its rusty hinges. 15 c. The smoke swirled/squeezed through the opening. (ii) Agentive 15 d. I slid/rolled/bounced the keg into the storeroom. 15 e. I twisted/popped the cork out of the bottle. iii) Self-Agentive 4 Talmy (2007) is a revised and expanded version of Talmy (1985). A version that is still further revised and expanded than Talmy (2007) appears as chapters 1 and 2 of Talmy (2000b). And chapter 3 in that volume extends the framework in Talmy (2007) to additional semantic categories. 5 Non-agentive motion has to do with situations in which entities that are capable of motion on their own perform some motion. Agentive motion refers to a motion event whose Figure is moved by an agent; the agent causes the motion but the verb can express either its cause or the Manner in which the Figure moves. Self-agentive motion refers to events in which Figures are able to move by themselves. University of Ghana http://ugspace.ug.edu.gh 26 15 f. I ran/limped/jumped/stumbled/rushed/groped my way down the stairs. 15 g. She wore a green dress to the party. Move + Cause iv) Non-agentive 15 h. The napkin blew off the table. 15 i. The bone pulled loose from its socket. 15 j. The water boiled down to the midline of the pot. v) Agentive 15 k. I pushed/threw/kicked the keg into the storeroom. 15 l. I blew/flicked the ant off my plate. 15 m. I knocked/pounded/hammered the nail into the board with a mallet. 2.1.3.3 Lexicalisation Pattern: Motion + Path In the second typological pattern for the expression of motion, the verb conflates both the fact of Motion and Path. Semitic, Polynesian, Romance, Korean, Turkish, Tamil, Nez Perce, and Caddo belong to this pattern. Spanish motion verbs are perfect examples of this type: Spanish expressions of Motion with conflation of Path taken from Talmy (2000:49-51). Non-Agentive University of Ghana http://ugspace.ug.edu.gh 27 16a. La botella entró a la cueva (flotando) The bottle MOVED-in6 to the cave (floating) ‘The bottle floated into the cave.’. b. La botella salió de de cueva (flotando) The bottle MOVED-out from the cave (floating) ‘The bottled floated out of the cave?’ c. Metí el barril a la bodega rodándo lo I MOVED-in the keg to the storeroom rolling - it ‘I rolled the keg into the storeroom.’ d. Saqué el corcho de la botella retorciéndo lo I –MOVED-out the cork from the bottle twisting-it ‘I twisted the cork out of the bottle.’ As examples (16a-d) illustrate, if the Co-event (either Manner or Cause) is expressed in Spanish, it tends to be in an independent element, usually as adverbial or gerundive. In many languages, besides Spanish, the expression of Manner and/or Cause by adverbials and gerundives is stylistically awkward. That is why information about Manner or Cause is often omitted, especially when the Manner of motion is a default or expected Manner of motion of the Figure, or it has been previously established in the surrounding discourse. In contrast, English verb roots readily conflate the Co-event but not Path. This lexicalisation pattern is not characteristic, though English also has verbs that 6 Following the glossing system of Cifuentes-Férez (2008) Spanish motion verbs are glossed as MOVED-in/out (for non-agentive motion and MOVED-in/out MOVED- in/out (for agentive motion), to show the conflation of motion and Path. University of Ghana http://ugspace.ug.edu.gh 28 incorporate Path such as arrive, approach, circle, cross, descend, enter, exit, follow, join, pass, rise, return, separate, etc. but most of them are historic borrowings from Romance languages. Furthermore, Talmy (2000b: 62, 139) notes that this conflation pattern, i.e., Motion + Path in verb roots, rarely extends to conflation of location with site, such as ‘BE-on’ ‘BE-under’, etc., though in English, there are some incidental cases of such conflation; eg. surround (be around), top (be atop), flank (be beside) adjoin (be next to), span (be from one side to the other), line (be in line) and fill. 2.1.3.4 Lexicalisation Pattern: Motion + Figure In the third typological pattern, the verb root conflates Motion and Figure. Languages in this type are Navajo and Hokan languages (such as Atsugewi). English does have a few forms that conform to this pattern: English examples of conflation of Motion and Figure taken from Talmy (2000b: 57). Non-agentive 17a. It rained in through the bedroom window. Agentive 17b. I spat into the cuspidor. Talmy uses Atsugewi, a polysynthetic language of Northern California, as the prototypical example of Figure – type languages. In Atsugewi, verb roots tend to express movement of objects, body parts and garments: ‘Atsugewi has verb roots that refer to a particular garment moved or located for wear that takes University of Ghana http://ugspace.ug.edu.gh 29 affixes indicating whether the garment is on, or is put on or taken off oneself or someone else’ (Talmy, 2000b: 59). Atsugewi verb roots of Motion with conflated Figure taken from Talmy (2000b: 58). 18a. -lup- ‘for a small shiny spherical object (e.g., a round candy, an eyeball, a hailstone) to move/be located. b. –caq- ‘for a slimy lumpish object (e.g., a toad, a cow dropping) to move/be located.’ c. –staq- ‘for runny icky material (e.g., mud, manure, rotten tomatoes guts, chewed gum) to move/be – located. 19. Atsugewi expressions of motion with conflated Figure (Talmy, 2000b:59). a. /’- w – uh – staq – ik - / Locative suffix: -ik ‘on the ground’ Instrumental prefix: uh- ‘from gravity (an object’s own weight) acting on it’. Inflectional affix set: ‘-w-’ 3rd person subject (factual mood) Literal: Runny icky material is located on the ground from its own weight acting on it. Instantiated: ‘Guts are lying on the ground’ b. /’-w-ca-staq-ict-a/ University of Ghana http://ugspace.ug.edu.gh 30 Directional suffix: -ict ‘into liquid’ Instrumental prefix: ca- ‘from’ Inflectional affix set: ‘-w- - a ‘3rd, person subject (factual mood) Literal: ‘Runny icky material moved into liquid from the wind blowing on it Instantiated: ‘The guts blew into the creek’ c. /s-‘ –w-cu-staw-cis-a/ Directional suffix: -cis ‘into fire’ Instrumental prefix: cu- ‘from a linear object, moving axially, acting on the figure’ Inflectional affix set: ‘s-‘ –w- - a ‘I subject, 3rd person object (factual mood) Literal: ‘I caused it that runny icky material move into fire by acting on it with a linear object moving axially’ Instantiated: ‘I prodded the guts into the fire with a stick’ Languages can sometimes conflate the same kind of semantic distinctions but in very distinctive ways. For example, Southwest Pomo conflates Motion with Figure, not as Atsugewi does, but rather with the numerosity of the Figure: ‘for one/two or three/several together … to move’ (Cifuentes-Férez 2008:34). University of Ghana http://ugspace.ug.edu.gh 31 2.1.3.5 Other Minor Patterns Although these three conflating systems for Motion verbs are apparently the main ones found across languages, Talmy (2000b:62) notes other lexicalisation patterns occur as well. He identified some minor systems of agentive verbs conflating two semantic components: 20. Ground + Path + Motion, such as box and shelve in the following examples: a. I boxed the apples = cause to move into a box b. I shelved the book = cause to move onto a shelf. 21 Figure + Path + Motion, such as powder and scale in: a. She powdered her nose = cause facial powder to move onto her nose b. I scaled the fish = cause the scale to move off These multi-component conflation patterns do not seem to form a language’s characteristic lexicalisation pattern for expressing Motion. Such combinations would require an enormous verb lexicon with fine-grained semantic combinations (Talmy, 2000b: 62). On the other hand, there are some conflations which are dispreferred. One particular Motion-event component, the Ground, does not by itself conflate with the Motion verb to form any language’s main system or pattern for expressing Motion. Unlike other motion event components, it is not clear why the Ground University of Ghana http://ugspace.ug.edu.gh 32 component is so disfavoured. Talmy (2000b: 61) thinks that the explanation might be sought in a concept hierarchy: The different conflation types seem to be ranked in their prevalence among the world’s languages, with conflation of Path as the most extensively represented, of Manner/Cause next, and of Figure least so. It may therefore be the case that Ground conflation is also a possibility, but one so unlikely that it has not yet been instantiated in any language. Nonetheless, this does not really explain why the lexicalisation of Ground is dispreferred by languages. The prevalence of the three lexicalisation patterns among the world’s languages may be interpreted as if languages are much more concerned with expressing the trajectory of motion, the manner in which the Figure moves and even the entity that moves than with conveying the reference entity. It might be argued that Ground is the least prominent element of a motion scene, thus, it is disfavoured in lexicalisation; however, we have not come across any psycholinguistic research supporting this idea (Cf. Cifuentes-Férez 2008: 35). 2.1.3.6 Split and Parallel Systems of Conflation Previously, it has been discussed that a language apparently has a characteristic conflation type. However, a given language can characteristically (a) use a different conflation system for different kinds of Motion, that is a language may have a split system of conflation, or (b) use different conflation types with the same type of motion event, that is, a language may have a parallel system of conflation. University of Ghana http://ugspace.ug.edu.gh 33 To illustrate split systems of conflation, Talmy focused on Spanish, Emai and Tzeltal. As documented by Aske (1989) and Slobin and Hoiting (1994), Spanish uses the Path-conflating type for motion events whose paths are conceptualized as crossing a boundary, and the manner – conflating type when there is no boundary-crossing involved.7 A different split pattern is found in Emai (a language spoken in Nigeria, Schaefer, 1986). Emai has a great number of path verbs but generally uses them only for self-agentive motion. For non-agentive motion, it uses verbs conflating the Co-event. Tzeltal exhibits another split system. In fact, it uses each of the three lexicalisation types for separate types of motion events: a. Figure – conflating verbs when the Figure is or ends up supported at some location, Tzeltal verb roots ‘largely distinguish Figure objects in terms of their disposition: their form, orientation, and arrangement relative to other objects’ (Talmy 2000b: 65). b. Path-conflating verbs for autonomous motion of a the Figure, that is, ‘(for a Figure) to MOVE along X Path’ and for controlled agentive motion, that is, ‘(For an Agent) to MOVE (The Figure) along X Path while holding (it)’ (Talmy, 2000b: 65). 7 Slobin and Hoiting (1994) find that the boundary-crossing constraint (present in events in which the Figure crosses a boundary, such as ‘exit’ and ‘enter’) is at work in French, Japanese, Korean and Turkish as well. University of Ghana http://ugspace.ug.edu.gh 34 c. Co-event-conflating verbs in constructions with the ‘directional form of the Path verbs (which, according to Talmy (2000b: 65), function like Path satellites). Finally, for a parallel system of conflation, Talmy (2000b:66) comments on modern Greek, which uses the path- and the co-event conflating types to represent most events of self-agentive motion ‘with comparable colloquiality’: 21. a. Etreksa mesa (s- to split) I-ran in (to- the house [ACC]) ‘I ran in (-to the house)’. b. bika (trekhondas) (s-ito spiti) I-entered (running) to-the house ‘I entered (the house (running)’ [ACC] 2.1.4 The Two-way Typology: Path of Motion Talmy’s binary typology hinges on a single crucial criterion: whether a given language preferentially expresses path (i.e. change of location) in the main verb (as in V-languages) or in satellite elements associated with the main verb as in S-languages as Fig. 2 and 3 illustrate. Figure 2: Satellite-framed languages MANNER PATH ↓ ↓ verb satellite University of Ghana http://ugspace.ug.edu.gh 35 Figure 3: Verb-framed Languages PATH MANNER ↓ ↓ adjunct verb Adapted from Levinson and Wilkins (2006:18) Thus, Path accordingly to Talmy (1991, 2000), is the defining conceptual element, or core schema of motion, while Manner constitutes a subordinate, or supporting piece of information i.e., a co-event. He writes: Since the figural entity of any particular framing event is generally set by context and since the activating process [the motion] generally has either of only two values, the portion of the framing event that most determines its particular character and distinguishes it from other framing events is the schematic pattern of association with selected ground elements into which the figural entity enters. Accordingly, either the relating function alone or this together with the particular selection of involved ground elements can be considered the schematic core of the framing event … the relating function that associates the figural entity with the ground elements among which the transition takes place constitutes path. The core schema here will then be either the path alone or the path together with its ground locations. (Talmy 1991: 483) These typological differences are illustrated in (22) and (23) to show the preferential lexicalisation of Path and Manner of motion in these two types of language. 22. Satellite-framed pattern, e.g. Subject[Figure] Verb[Manner] Satellite[Path] Object[Ground] JulieF ranM acrossP the streetG 23. Verb-framed pattern, e.g. French Subject[Figure] Verb[Path] Object[Ground] Gerund[Manner] University of Ghana http://ugspace.ug.edu.gh 36 JulieF traversaP la rueG en courantM ‘James ran across the street.’ (Adapted from Kopecka, and Pourcel (2008:8- 9) Slobin (2004) provides a detailed list of verb- and satellite-framed languages:  Satellite-framed languages: o Germanic: Danish, Dutch, English, German, Icelandic, Swedish, Yiddish o Slavic: Czech, Polish, Russian, Serbo-Croatian, Ukrainian o Finno-Ugric: Finnish, Hungarian o Chinese: Mandarin o Australian: Warlpiri  Verb-framed languages: o Romance: Catalan, French, Galician, Italian, Portuguese, Spanish o Semitic: Moroccan Arabic, Hebrew o Turkic: Turkish o Basque o Japanese o Korean University of Ghana http://ugspace.ug.edu.gh 37 o Sign Languages: American Sign Language, Sign Language of the Netherlands Talmy (1991, 2000b) also suggests that the constituent where Path is characteristically expressed to a great extent is where aspect, change of state, action correlation, and realization are characteristically encoded as well. In other words, the binary typology for motion events extends to at least the other four types of events. Thus, verb-framed languages characteristically lexicalise the trajectory of motion, aspect, change of state, action correlation in the main verb, whereas it is expressed by verb particles or satellites in satellite-framed languages. Let us illustrate this with examples from English and Spanish taken from Talmy (2000b: 240-260). 24. Change of State I blew out the candle. Apagué la vela de un ‘I extinguished the candle [by] blowing air on it/with a blow’ 25. a. English: I burned him to death b. Spanish: Lo mataron con fuego quemandolo ‘They killed him with fire [by] burning him’. 26. Action Correlation a. English: I played the melody along with him. b. Spanish: (i) Yo lo accompane cuando tocamos la melodia ‘I accompanied him when we played the melody’ (both he and I played). University of Ghana http://ugspace.ug.edu.gh 38 (ii) Yo lo acompane tocando la melodia ‘I accompanied him [by] playing the melody (only I played)’. c. German: Ich ha be mit ihm die melodie mitgesplelt ‘I played the melody along with him.’ d. English (b) I outplayed him. e. Spanish: Yo le gane tocando la melodi ‘I surpassed him playing the melody.’ In all these examples, changes of state as well as action correlation are expressed in English with the particle or satellite, while the main verb encodes the Co-event (Manner or Cause). In contrast, in Spanish, the main verb expresses the transition to a new state in (25), and coactivity in (26), while the adjunct, either a prepositional phrase (de un soplido, con fuego), or a gerund (solplándola, quemándolo, tocando), expresses the Co-event. 2.1.5 Typological Shifts The shift of a language from one preferred lexicalisation pattern to another, or its maintenance through the course of time is still an unexplored research area. The factors that may have tilted one language towards maintaining its typological pattern category and another language toward shifting to another must yet be discerned. In a speculative fashion, Talmy (2000b: 118-119) comments that Latin, Classical Greek and Proto-Germanic all exhibited the presumably Indo-European pattern of using co-event-conflating verb roots together with path satellite that University of Ghana http://ugspace.ug.edu.gh 39 formed prefixes on the verb roots. Possibly because of phonological changes that made the path prefixes less distinct from each other and from the verb roots, all three languages apparently became unable to maintain their inherited pattern. Both Germanic Greek8 proceeded to develop a new set of path satellites that largely supplanted the prior set. This permitted the maintenance of the inherited pattern for representing motion events with co-event verb conflation. On the other hand, languages arising from Latin developed a new system of path- conflating verbs, rather than re-establishing the path satellite system. In this process, Talmy (2000b: 119) remarks that each of the daughter languages formed its set of path verbs in its own way by variously coining new verbs so as to fill out the basic directional grid of the new path verb system. For example, the Latin verb salio (jump) which combined with path prefixes such as up and out developed into Italian salire (ascend) and Spanish salir (‘go out’) (cf. Walchi, ms. cited in Cifuentes-Férez 2008: 40). At the same time, these languages may have undergone the complementary change of advancing their gerundive construction for the expression of Manner and Cause. Kopecka (2004, 2006a, 2009, in press), examines French’s typological shift from a satellite-framed pattern to a verb- framed pattern. Drawing on diachronic data, she provides diachronic evidence of this shift and concludes that this shift can be mainly attributed to the weakening of productivity of verbal prefixes. 9 Although old French prefixes lost their productivity, many remnants are found in modern French. On one hand, there 8 Papagragou et al. (2002, 2006) categorized modern Greek as a verb-framed language. 9 Other closely related factors contribute to this typological shift as well, namely, the loss of prefixed verbs by other lexical items. University of Ghana http://ugspace.ug.edu.gh 40 are verbs with quite transparent prefixes, such as s’en-voler ‘fly away’ and sur- voler ‘fly over’ which conform to the satellite-framed pattern. On the other hand, there are more opaque verbs, whose composite nature is no longer discernable, such as arrive r ‘arrive’ and éloigner ‘move away’ which is consistent with the verb-framed pattern. 2.1.6 Typology Revisited (A Third Category) Talmy’s typological framework offers an economical way of chracterising motion lexicalisation patterns across a wide range of languages, and has been widely discussed. Indeed, an important gap in the typology seems apparent when considering serializing languages, such as Thai (Zlatev and Yanklang 2004) or Ewe and Akan (Ameka and Essegbey (2001). The specificity of these languages is to express both Path and Manner equipollently in a single verb clause containing two or more obligatory verbs, i.e., one verb expressing Path and one verb expressing Manner. In other words, Path and Manner receive equal semantic emphasis within the same verb complex, e.g. 27. Thai Chan klap khaw paj/maa naj hɔ ɔ I return enter go/come inside room ‘I came back into the room’. (Zlatev and Yangklang, 2004: 164, ex. 8). Ewe 28. Ɖeví – á tá yi xɔ - a me. child- DEF crawl go room-DEF containing region ‘The child crawled into the room.’ University of Ghana http://ugspace.ug.edu.gh 41 (Ameka and Essegbey, 2001:3) Zlatev and Yangklang (2004) have argued that serializing languages share characteristics of both satellite- and verb-framing languages, and therefore do not clearly correspond to either of the patterns offered by Talmy’s typology. Moreover, Talmy’s theory of lexicalisation patterns does not seem to take into account the fact that languages within the same typological group show a differing degree of manner of elaborations. For example, Spanish and Basque are both verb-framed languages but their elaborations of Manner and Path are quite different, as has been shown by Ibarretxe-Antunano (2004a, 2004c, 2006a, 2006b). Basque describes Manner much more often than Spanish; thanks to its sound symbolic expressions or movement initiatives, and thus Basque is closer to satellite-framed languages. On the other hand, in terms of Path, Basque generally offers more detailed paths or trajectories than Spanish. As a way of resolving this issue, it has been proposed that ranking languages on clines or continuums of manner (Slobin, 2004, 2006a, 2006b) and path salience (Ibarretxe-Antunano, 2004a, 2004c, 2004b, 2008) would be more useful than placing them into typological categories. Slobin (2004, 2006a) and Ibarretxe- Antunano (e.g., 2004a, 2008) identify a number of factors which contribute to a language’s degree of path and manner salience respectively. This proposal of clines of semantic component saliency were examined after looking at the third category proposed. University of Ghana http://ugspace.ug.edu.gh 42 2.2 The works of Slobin and others 2.2.1 Equipollently-framed Languages Slobin (2004:249) proposed a tripartite typology of motion-event constructions: verb-framed, satellite-framed and equipollently-framed languages. Equipollently-framed languages are those languages in which both Path and Manner have roughly equal morphosyntactic status; in other words, Path and Manner are expressed by equivalent grammatical forms as stated earlier in Section 2.1.6. There are at least three subtypes of equipollently-framed languages according to Slobin (2006a: 64): 1. Serial-verb languages: In serial-verb language, it is not always clear which verb in a series, if any, is the main verb. The Niger-Congo, Hmong-Mien, Sino-Tibetan, Tai-Kadai, and Mon-Khmer families and some Austronesian languages belong o this group. 2. Bipartite verb languages: the Hokan and Perutian languages described by Delancey (1989; 1996) are languages in which the verb consists of two morphemes of equal status, one expressing Manner and the other Path. Talmy (2000b:113), drawing from Aoki (1970), provides a similar description of Nez Perce manner prefixes, such as quqú- láhsa ‘gallop-ascend’. Rhodes, (cited in Cifuentes-Férez, 2006: 42) in a personal communication with Slobin (in 2003) reports that such constructions are typical of Algonquian, Athabaskan, Hokan; and Klamath-Takelman Huang and Tanangkingsing (2004) report that at least one Austronesian language, Tsou, has apparently developed bipartite manner – path verbs from serial-verb constructions. University of Ghana http://ugspace.ug.edu.gh 43 3. Generic verb languages: the Australian language, Jaminjung (Schultze-Berndt, 2000) has a very small lexicon of about 24 function verbs. For encoding motion events, one of five verbs is used, expressing a deictic or aspectual function: go come, fall, hit, do. These verbs are combined with satellite-like elements or covers, which encode both Path and Manner in the same fashion. In response to Slobin’s proposal of a third group of language, Talmy (in Ibarretxe-Antunano, 2005: 331) agrees that the equipollently-framed category is a good way to view such linguistic variation, but he strongly argues that the criteria used for judging main verb status in those languages have been too few, and that authors need to apply an expanded set of criteria to elucidate which constituent is privileged with main verb status. These criteria or factors range from phonological to morphosyntactic to semantic factors. Ameka and Essegbey (2001/2013) for instance, drawing on data from Akan and Ewe, Kwa languages, and Sranan, a Caribbean Creole with Gbe substrate spoken in Surinam, investigate the expression of complex translational motion events in serializing languages in the light of Talmy’s typology. They argue that serializing languages pose a challenge to this typology as they express Path as well as the co-event of manner infinite verbs that together function as a single predicate in translational motion clause. They concluded that these languages constitute a type because in terms of discourse properties, verb serializing University of Ghana http://ugspace.ug.edu.gh 44 languages behave like verb-framed languages with respect to some properties and like Satellite-framed languages in terms of others: When the properties are tallied, we find that serializing languages share more properties with S-languages than the V-languages to which they are supposed to belong while still possessing a unique property. What this shows is that they cannot be said to belong to either type. Instead, they appear to belong to a class of their own. Ameka and Essegbey 2001/ 2013:36). This study also contributed to the revision of the typology hence Slobin’s (2004, 2006) proposal that a third type, equipollently-framed languages be added to include serial-verb languages and other types of languages in which both manner and path are expressed by equipollent elements that is, elements that are equal in formal linguistic terms and appear to be equal in force or significance. The role of verb serialization in the expression of spatial notions and strctures that encode motion events is further explored in Ameka and Essegbey (2006) for Ewe. These works that discuss motion provide a starting point for the exploration of the acquisition and use of motion expressions as constrained by typology in the current thesis. Slobin’s proposal of the equipollently-framed type has been challenged by Talmy (2009) who insists that constructions used in the serializing languages (exemplified mainly with Chinese) for the translational-motion events are satellite-framing. Croft et al (2010), who also proposed a revision of Talmy’s typology, suggested that the framing type should not be used to characterize University of Ghana http://ugspace.ug.edu.gh 45 whole languages and deconstruct the various framing types into construction types. Lambert-Br ti re’s (2009) paper also challenges the typological classification of serializing languages by proposing that a serializing language like Fon, a Kwa language mainly spoken in South Benin, is better analyzed as a satellite-framed language, lexicalizing the core-schema of motion (path) in a verb satellite rather than as a verb-framed or equipollently-framed language. In this paper, semantic and syntactic arguments are presented, leading to a new definition of verbal satellite in functional terms. An important point that this paper makes is that there is no need for a specific typological classification of serializing languages like Fon. Instead, a representation of Talmy’s motion typology as a cline between the two lexicalization patterns, verb-framed and satellite-framed, accounts for languages sharing characteristics of both framing types. A combination of the structural typology by Talmy (1985, 1991, 2000, 2008) and of the typological continuum by Slobin (2004) and Ibarretxe- Antu ano (2004) is appropriate for categorizing languages of the serializing type like Fon. On the other hand, van Putten (2009, in press) examining narrative discourse on motion in Avatime, a Ghana-Togo Mountain Language of the Kwa family, concludes that Avatime is equipollently-framed like Thai and Chinese, as well as Kwa languages like Ewe and Fon. There is thus diversity among verb serializing language types that are equipollently-framed. University of Ghana http://ugspace.ug.edu.gh 46 Schultze-Berndt (2006) presents a case study of the dialectic interaction between linguistic theory and in-depth language documentation in Jaminjung, a Non-Poma-Nyungan language of Mirndi family. Findings show that the lexicalisation and discourse uses of motion expressions in Jaminjung neither fit the characteristics of an equipollently-framed language, as predicted by Slobin (2004) (and previously, Schultze-Berndt, 2000) nor do they reveal satellite- framed characteristics as claimed for the structurally similar language, Warlpiri. A language like Jaminjung does, however, point to the need for a careful definition of ‘verbs’ and ‘satellites’ in the typology in distinguishing between “main verb status’ and ‘open class membership’ as a defining criteria. Given the close-class nature of ‘verbs’ in Jaminjung, the language is of particular interest to theories of lexicalisation patterns because ‘path information is in the Inflecting Verbs, but not all types of ‘path’ information are treated equally. While deixis and motion with respect to a reference point is expressed by Inflecting Verbs (IVs), path shape (direction), information about the region at the end point, and boundary crossing information only get expressed by Uninflected Verbs, at par with manner. She however suggests that there is the need to carry out a typological comparison of languages with close-class verbs that might reveal an implicational hierarchy for the lexicalisation of these subtypes of path. 2.2.2 Cline of Manner Salience Besides allocating languages into three typological categories, Slobin (2004, 2006, 2006) argues for a cline or continuum of manner salience. Independently from the lexicalisation pattern they belong to, languages differ significantly in the amount of manner information that is given, the frequency of University of Ghana http://ugspace.ug.edu.gh 47 mention of manner details, and the sort of fine grained manner distinctions that they encode. Thus there are high-manner-salient languages and low-manner- salient languages. In Slobin’s (2004: 250) own words: In High-manner-salient languages, speakers regularly and easily provide information about manner when describing motion events, whereas in Low-manner-salient languages manner information is only provided when manner is foregrounded for some reason. Furthermore, Slobin points out that high-manner-salient languages possess a rich expressive lexicon of manner verbs encoding fine-grained distinctions, and that their speakers are thought to pay attention to Manner. In contrast, low-manner- salient languages have a less extensive verb lexicon expressing Manner, and are thought to attend less to the Manner component of motion. Slobin has observed that there are a number of factors that interact with lexicalization patterns in influencing manner salience: lexical and morphemic availability, semantic constraints and processing load. One of the factors which plays a role in the amount of manner information given by a language is lexical and morphemic availability. The more accessible and codable the semantic component of Manner is in a language, the more manner information this language is likely to express. According to Slobin (2004), Manner of Motion is more codable in a language when it is expressed (1) by a finite rather than non- finite verb, (2) by a single word rather than a phrase or clause, and (3) by a high frequency rather than low frequency lexical item. In high-manner-salient languages there is an accessible slot for manner made available in a number of ways: University of Ghana http://ugspace.ug.edu.gh 48 i. A main verb in satellite-framed languages. ii. A manner verb in serial-verb languages. iii. A manner morpheme in bipartite verbs iv. A manner preverb in Jaminjungang languages v. As ideophones or expressive forms (e.g., Basque (e.g., Ibarretxe- Antu ano, 2006a), Japanese (e.g., ita, 1997; Ohara, 1995, 2003; Sugiyama, 2000, 2005). Apart from lexical availability and codability of manner, it is important to explore other factors, such as semantic constraints and processing load, which greatly influence the expression of manner of motion across languages. Lexical availability and ease of codability are not sufficient to understand why some languages possessing the means to express manner of motion fail to do so. One answer lies in what Slobin and Hoiting (1994) have called the boundary-cross constraint, drawn from Aske’s (1989) observations on the role of telicity in the expression of motion. It seems that verb-framed languages (such as Spanish) only license the use of manner verbs when the motion event depicted is atelic, that is, when the event is a motion activity with duration and no boundary-crossing is predicated. Consider the following example from Aske, (1989 cited in Cifuentes-Férez 2008: 44). 29. Juan bailó en cículos/ de un lado para otro / John danced in circles/ from one place to another/ hacia la puerta/ hasta la puerta University of Ghana http://ugspace.ug.edu.gh 49 towards the door / to the door ‘John danced in circles from one place to another/towards the door/to the door.’ For telic events where the end location of the Figure is stated, verb- framed languages cannot use a manner verb, as in (30), but necessarily mark this change of location with a path verb (entrar ‘enter’, salir ‘exit’, cruzar ‘cross’, etc.) or neutral verb (ir ‘go’, venir ‘come’, etc.) In order to add manner, some sort of subordinate construction is required (gerundive forms, prepositional phrase, etc.) as in 31. 30. *Nadaron adentro de la Nadaron ade ntro de la cueva They-swam to-inside of the cave ‘They swam into the cave.’ 31. Entraron a la cueva nadando They-enter to the cave swimming ‘They swam into the cave.’ A further specification on the semantic constraints in the use of Manner verbs in Spanish is made by Naigles et al (1998). In their study, they found that Spanish speakers predominantly used path verbs for boundary-crossing events when the boundary traversed was horinzontal. In contrast, for punctual vertical boundary crossing10 situations (e.g. tirarse a la piscine ‘throw oneself/plunge 10 Boundary crossing refers to a path that crosses the limits of a bounded extent (hence the event is telic, whereas non-boundary crossing refers to a path that unfolds) within an unbounded extent, the motion being thus atelic. (Slobin and Hoiting 1994; Aske 1989) University of Ghana http://ugspace.ug.edu.gh 50 into the pool’), Spanish licenses the use of manner verbs. It appears that a sudden boundary crossing allows the use of manner verbs in Spanish and maybe in other verb-framed languages. Besides semantic constraints, another factor which comes into play apart from low-manner salience in verb-framed languages is processing effort. Although verb-framed languages have the option of using adjuncts or subordinate constructions to express Manner of motion, this alternative is avoided most of the time. As Slobin (2006: 67) points out, the explanation for this general avoidance of Manner in verb-framed languages may be sought in the unnecessary foregrounding of Manner, which results in heavy processing of these constructions in terms of speaker’s production and hearer’s comprehension. When Manner of motion is expressed outside the verb root, it ‘forms part of the semantic background where it attracts little direct attention’ (Talmy, 2000b: 128). Talmy observes that a semantic concept which is backgrounded is more readily expressed in a language, and ‘its informational content can be included in a sentence with apparently low cognitive costs-specifically, without much additional speaker effort or hearer attention’ (Talmy, 2000b: 129). Thus, in verb- framed languages, the expression of Manner outside the verb root is often neglected as it would direct one’s attention to it causing an extra processing load. To summarise, languages differ considerably in the amount of manner information they express (even languages within the same typological group). The main factor is the preferred lexicalization pattern (verb-, satellite-, or equipollently-framed languages)/ though other factors come into play interacting with it and influencing the degree of manner of salience in a given language. University of Ghana http://ugspace.ug.edu.gh 51 2.2.3 Cline of Path Salience In line with Slobin’s cline of Manner salience, Ibarretxe-Antu ano (2004a, 2004c, 2008) proposes a cline of Path salience. Languages vary in the degree of detailed description of Path of motion independently from the lexicalization pattern they belong to. Her proposal classifies languages along a continuum between two ends: high-path-salient languages and low-path-salient languages. The former offers detailed and frequent descriptions of Path, whereas the latter, coarse descriptions usually limited to the use of path verbs with or without a Ground. Although Path is an obligatory constituent of a motion, languages differ in the degree to which they present Path and Grounds as well as complex trajectories. For example, Spanish favours the use of path verbs with no Ground over path verbs with explicit mention of the Ground, whereas Basque seems to favour the opposite. Moreover, when those two languages describe complex paths, Basque generally expresses complex paths with more than two Grounds, whereas Spanish typically uses one. Also, in Avatime, a Ghana Togo Mountain Language, complex paths can be expressed in serial verb constructions, but only two ground elements can be combined in this way and only one ground argument is ever used per clause in spontaneous speech. Not more than three path verbs can be combined in serial verb constructions and SVCs cannot contain more than one of the same kind of verb (departure, passing, arrival). Consecutive constructions can be used to combine more than three path verbs, more than two ground elements and more University of Ghana http://ugspace.ug.edu.gh 52 than one of the same kind of verb. The preferred way of expressing complex paths is by using a separate clause for each path element (van Putten 2009: 90). As already mentioned in the previous section, with respect to the expression of Manner, apart from typological factors, other factors have an effect on the elaboration of a semantic component. At this point, it is necessary to investigate what makes path of motion salient in a language. Ibarretxe-Antu ano (2004c, 2008) argues that some factors such as:  rich system for space and motion description  word order  VP gapping-ellipses  Dummy verbs  Cultural systems  Conceptionally-oral languages contribute to the degree of path and ground elaboration. Regarding the first factor, the available linguistic devices for space and motion, high-path-salient languages often have rich lexical and morphological resources such as locational cases, locative nouns, sets of directionals, etc. The factor of word order is related to the observation that languages whose verbs are in final positions tend to express details about Path and Ground before the verb. Thus, these complements have already provided all the University of Ghana http://ugspace.ug.edu.gh 53 information for path of motion. This factor is crucial for languages such as Basque, Chantyal, Japanese and Turkish (Ibarretxe-Antu ano, 2008). The third factor has to do with verb omissions. High-path-salient languages seem to be more likely to allow verb omissions than low-path-salient languages. The former readily omit the verb as the path complements present in the utterance are explicit enough, whereas the latter cannot omit the verb as it is one that conveys the necessary information for the description of the motion event. The fourth factor has to do with the concept of language oraltity. Ibarretxe-Antu ano (2008) states that och and Oesterreicher (1985) and Oesterreicher (2001) think that irrespective of the mode of expression (written, oral) languages can be classified as conceptually written and conceptually oral. Conceptually oral languages are languages characterized by redundancy, elliptic constructions, hyperbolic expressions, self-corrections, etc. Ibarretxe-Antu ano (2008) notes that conceptually oral languages such as Basque and Swiss dialect Muotathal, are more likely to be high-path-salient than conceptually written languages. Last but not the least, according to Ibarretxe-Antu ano (2008), cultural factors have a say in Path salience. Wilkins (2004) and Bavin (2004) argue that aboriginal communities show a great concern for detailed attention to motion paths and for orientation in space. As a consequence, their linguistic descriptions reflect this concern. University of Ghana http://ugspace.ug.edu.gh 54 Slobin (1987, 1991, 1996a, 1996b, 1997a, 2000) has taken this typology one step further and proposed a modified version of the classic Sapir-Whorf arguments on linguistic relativity which were much debated in the first half of the twentieth century (Sapir 1924, Whorf 1956): the thinking-for-speaking hypothesis. The next section discusses linguistic relativity and Slobin’s proposal to shift the subject from “Thought and Language” to “thinking and speaking.” 2.2.4 Linguistic Relativity and Slobin’s Proposal to Shift the Subject from “Thought and Language” to “Thinking and Speaking” The concept of thinking for speaking is by no means unrelated to linguistic determinism/relativity, or the often called Sapir-Whorf hypothesis. Slobin himself refers to the idea of thinking for speaking as “a modified form of the Spair-Whorf hypothesis (Berman and Slobin 1994: 612). The original idea of what has been called linguistic determinism /relativity, which is more or less attributed to Humboldt, Boas, Sapir, and Whorf, is summarized in Gumperz and Levinson (1996: 2): “the semantic structures of different languages might be fundamentally incommensurable, with consequences for the way in which speakers of specific languages might think and act”. Linguistic determinism/relativity claims that language and thought, and culture, are deeply interconnected, and that different languages lead to different world-view. The problem with such an argument, pointed out by many scholars/linguists, for example Lucy (1992 a, b), Pinker (1989), Slobin (1991), and Lakoff (1987) is that it is primarily concerned with the relationship between University of Ghana http://ugspace.ug.edu.gh 55 language and an abstract entity called “world-view” or “thought which can hardly be specified.” Combined with growing criticism against linguistic determinism/relativity for the difficulty to prove its claim of empirical methods, what further discredited linguistic determinism/relativity was the foundation and development of cognitive science in the 1960’s. Researchers of cognitive science emphasize the commonality of the genetic basis of human cognition from which they claim language derives (cf for the basic assumptions of cognitive science, see Garfield, 1990). More recently, researchers attempting to test the validity and causal link between language and thought try to specify what is meant by “thought” or “world-view”. While some scientists limit their research to some specific areas of language and cognition (for example, Steiner, 1975; Lee 1996; Lucy 1992a, b) Slobin (1991) insists on discarding the notion of language and thought itself. His proposal is to shift the focus of the discussion from static, abstract entities like language, thought, or world-view to the relationship between two dynamic processes of thinking and speaking. Slobin explains what he calls thinking for speaking as something involved only in the process of speaking. It means in Slobin’s words, “a special sort of thinking [which] comes into play, on-line, in the process of speaking in a particular language” (cf 1.5.2). Slobin’s approach toward linguistic determinism/relativity is more similar to that of Sapir than Whorf. Sapir believes that people tend to pay special attention to what he called, ‘contents of experience [that are] capable of University of Ghana http://ugspace.ug.edu.gh 56 expression in linguistic terms’ (Mandelbaum 1958, cited in van Kruistum 2006). Whorf tended to claim more radically that language and thought are identical. Slobin is at this point skeptical about any further effects language has on the mind beyond the moment of speaking. ‘We will probably never succeed in demonstrating the effects of grammar on world-view or nonlinguistic behavior. But there is a special kind of thinking that is intimately tied to language – namely, the thinking that is carried out, on-line, in the process of speaking’ (Slobin 1991: 11). Slobin outlined three approaches to demonstrate linguistic relativity in his thinking-for-speaking sense. One is to find the stages at which children talk about experience in ways that appear specifically shaped by the linguistic system they are acquiring; another is to identify the difficulties that second language learners have in adapting their thinking to the new language; the third is to look at languages historically – the elements most resistant to change being possibly those most deeply ingrained in thought; In each of these approaches, spoken language is the only source of information, and none of them breaks into the logical circle of language representing itself. Child speakers of different languages acquire expressive habits that mirror the semantic-structural differences among their languages (Choi and Bowerman, 1991). According to Levelt (1989, cited in Riemier (2010: 412) it is during first language acquisition that the effects of thinking-for-speaking are most noticeable. In learning their native language, the child gradually learns what kind of conceptual distinctions are relevant in framing messages: University of Ghana http://ugspace.ug.edu.gh 57 In learning the language, the speaker (the child) must surely have realized that the language requires him to attend to certain perceptual or conceptual features when he encodes a message. And … the child makes characteristic errors that reveal his successive hypotheses about the conceptual properties required for the assignment of his language’s morphology. Each native language in other words, ‘has trained its speakers to pay different kinds of attention to events and experiences when talking about them’ (Slobin 1996: 89). Languages without an explicit perfective/imperfective contrast, for example, do not require speakers to attend to this dimension of an event, whereas languages with an explicitly coded definite/indefinite contrast will require speakers to determine the definiteness of values of the NPs they mention. (see also Gentner and Boroditsky, 2001, for the extent to which language exerts an influence on conceptualization in child learning). Papagraguo et al (2002) discuss the question of linguistic relativity as it has resurfaced in the current psychological and anthropological literature, using motion verb representation and expression as the test bed. The overall findings suggest a good deal of independence between conceptual and linguistic representation. Such findings come as little surprise to recent proponents of the linguistic relativity position for they – as well as Whorf and Sapir themselves, except in their most enthusiastic moment – have no quarrel with the notion that there are certain immutable core concepts shared across the species. Rather, these investigators have tried to show that in addition to these shared concepts there are different ones too, literally caused and imposed by the language differences themselves. Sometimes this view is called ‘weak Whorfianism.’ Based on these findings, the writers have urged quite a different perspective. They assert: University of Ghana http://ugspace.ug.edu.gh 58 Many apparent effects of language on thought are more appropriately interpreted as effects of language on language. In tasks that explicitly or implicitly call for knowledge of language, subjects can in fact draw upon that knowledge. Because the languages differ, speakers of the different languages will differ accordingly. But this proves no more (and no less!) than that, English speakers speak English and Greek speakers speak Greek. The linguistic relativity question interpreted nonvacuously, is whether by having learned these languages speakers differ in the very basis of their inductions, whether they are representing, categorizing, remembering, and reasoning in terms of a (partially) different set of experiential categories. (Papafraguo et al 2002:190). In their view, the answer to these questions tends to be no. They argued the case by pointing to the following generalization: the more language-like the subjects’ task, the more speakers of different languages can be shown to vary in their performance; the more language is removed from the task situation, the more subjects exhibit their human conceptual communalities. Slobin (2003:160) extends thinking-for-speaking to all forms of linguistic production (speaking, writing, translating, signing) and reception (listening, reading, viewing) as well as a range of mental processes (understanding, imagining, remembering) in an attempt to account for the far-reaching effects that language use has on mental processes tied to language. Throughout his work, he argues that serious study of language in use points to pervasive effects of language on selective attention for particular motion event characteristics. Path is the core of the motion event descriptions, though by different linguistic elements (e.g. Talmy, 2000b). However, Manner of motion is an external component optional in verb-framed languages but readily encoded in satellite- framed languages. This fact suggests that speakers of satellite-framed languages may pay more attention to Manner of motion than speakers of verb-framed University of Ghana http://ugspace.ug.edu.gh 59 languages. In brief, Slobin’s proposal is that habitual on-line attention to Manner has made it especially salient in S-language speaker’s conceptualization of motion events (Slobin 2003: 164). Slobin himself explains the core motivation for this hypothesis which has been stated under Section 1.5.2. This means experience cannot be verbalized without having taken a specific perspective, influenced, if not determined, by the typological characteristics and lexicalization patterns of a given language. What we experience/ perceive might be the same event but the way we choose to talk about it seems to be different across languages. This is why Slobin (1996a) says any event (in our case, a motion event) can be described in terms of two different cognitive frames. On the one hand, that which refers to the actual event or experience that we want to describe (the translational motion from one place to another), and on the other hand, the tools provided and constraints imposed on speakers in expressing that event in a particular language: what he calls a “discourse frame and a typological frame” respectively. Thus he hypothesizes a number of cognitive consequences of this differential encoding of Manner of motion. He further asserts that if a language provides fine-grained, habitual and economical expression of manner of motion, a. references to manner of motion will occur frequently, across genres and discourse contents. b. manner-of-motion verbs will be acquired early. c. the language will have continuing lexical innovation in this domain, including extended and metaphorical uses. University of Ghana http://ugspace.ug.edu.gh 60 d. speakers will have rich mental imagery of manner of motion. e. manner of motion will be salient in memory for events and in verbal accounts of events. (Slobin, 2003: 163-164) Review of speaking-for-speaking research would be based on the five points discussed above. To sum up, that languages differ in their thinking-for-speaking demands is a version of the linguistic relativity hypothesis, the proposition that language influences thought in different ways. Thinking-for-speaking differs from the so- called ‘strong’ Whorfian version of the linguistic relativity hypothesis, as we understand it: The later (Whorf, 1956; Lucy, 1987, 1992) refers in general to langue – wide patterns of ‘habitual thought’, patterns that according to the hypothesis, are embodied in the forms of the language and analogies among them. The thinking-for-speaking hypothesis, in contrast refers to how speakers organize their thinking to meet the demands of linguistic encoding on-line, during acts of speaking – what Saussure termed parole rather than langue (de Saussure 1959). The thinking-for-speaking version and the Whorfian version of the linguistic relativity hypothesis are not mutually exclusive, but neither are they identical. The distinction between them parallels the characterization of Whorf as ‘synchronic’ compared to Vygotsky (1987) as ‘diachronic’ that was offered by Lucy and Wertsch (1987, cited in Riemier 2010). University of Ghana http://ugspace.ug.edu.gh 61 2.3 Cross-linguistic studies on motion events 2.3.1 Introduction The domain of motion has received a good deal of research since Talmy’s seminal work (e.g., 1985, 1991, 2000a, 2000b). This section provides a review of the whole gamut of research inspired or influenced by the Talmian theory of lexicalization patterns for motion events. Such research could be divided into two parts: Slobin’s thinking-for-speaking research (Section 2.8.2) and linguistic relativity research (Section 2.9). These are examined in the sections that follow. 2.3.2 Thinking-for-Speaking Research To investigate thinking-for-speaking, Slobin concentrates on the domain of motion. His approach is comparative, initially investigating English and Spanish (e.g., Slobin 1996a, 2000), but later broadening the scope to include other languages (e.g., Slobin, 2003, 2006). As noted previously, Slobin (2003:160) extends thinking-for-speaking to all forms of linguistic production (speaking. writing, translating, signing) and reception (listening, reading, viewing) as well as a range of mental processes (understanding, imaging, remembering) in the attention for particular motion event characteristics. Path is the core of the motion event and, as such, it is always encoded in motion event descriptions, though by different linguistic elements (e.g. Talmy, 2000). However, Manner of motion is an external component, optional in verb-framed languages but readily encoded in satellite- framed languages. This fact suggests that speakers of satellite-framed languages may pay more attention to Manner of motion than speakers of verb-framed University of Ghana http://ugspace.ug.edu.gh 62 languages. In brief, Slobin’s proposal is that ‘habitual, online attention to Manner has made it especially salient in S-language speaker’s conceptualisations of motion events’ (Slobin, 2003:164). Thus he hypothesizes a number of cognitive consequences of this differential encoding of Manner of motion 11 which have been mentioned earlier in the previous chapter. Although these five consequences are called cognitive, it should be noted that the first three consequences refer to linguistic cognition, whereas the last two are cognitive hypotheses about speakers’ conceptualization of motion (i.e. non- linguistic cognition). Our view of thinking-for-speaking research in the following section will thus focus on the first three, leaving the final two for Section 2.9. 2.3.3 Narrative Style A great bulk of research on motion events across a wide range of languages comes from elicited spoken narratives using a wordless picture book ‘Frog where are you’ (Mayer, 1969). This book is about a boy who goes in search of his run away frog. The book depicts several motion scenes, involving multiple types of figures (human and animals), Paths, and Manners of motion. In the first volume of the frog story studies, Berman and Slobin 1994:198-199) summed up the typological contrasts found between the narrative of three verb-framed languages (Hebrew, Turskish and Spanish) and two satellite-framed languages (English and German): 11 Refer to Chapter 2 (section 2.2.3) for Slobin’s proposal. University of Ghana http://ugspace.ug.edu.gh 63 Satellite-framed languages allow for detailed description of paths within a clause, because their syntax makes it possible to accumulate path satellites to a single verb, along with prepositional phrases that add further specification (e.g., the deer threw them off over the cliff into the water) […] The satellite-frames languages in our sample also tend towards greater specification of manner, probably because the lexicon provides a large collection of verbs that conflate manner with change of location (crawl, swoop, tumble, etc.) often conflating cause as well- dump, hurl, shove etc.). In verb-frames languages, such elaboration is more of a “luxury”, since path and manner are elaborated in separate expressions which are generally optional, and which are less compact in form. As a consequence of these differences, it seems - at least in our data – that English and German narrations are characterized by a great deal of dynamic path and manner description, while Spanish, Hebrew and Turkish narrations are less elaborated in this regard, but are often more elaborated in description of locations of protagonists and objects of endstates of motion. (Berman and Slobin, 1994:118-119). Commenting on this quotation, Cifuentes-Férez (2008:60) notes that narrative styles in verb- and satellite-framed languages are quite distinct in terms of dynamism of the events being depicted. Satellite-framed languages presented more dynamic and lively descriptions of motion events both with respect to manner of motion and paths, while verb-framed languages tended to focus on static descriptions of the setting and the protagonist’s endpoint of location. Applying Ikegami’s (1991) terminology of become-language (i.e., a language which focuses on the change from one state into another, eg., Japanese) versus do-language (i.e, a language such as English which focuses on the activity of an individual.), verb-framed languages might be described as become-languages since they devote more narrative attention to static descriptions of the setting and tend to express more often the figure’s endpoint of location, leaving the figure’s ways of moving through space and the trajectories it follows to be inferred. These ways of moving through space and trajectories, in contrast, are favoured University of Ghana http://ugspace.ug.edu.gh 64 by satellite-framed languages or do-languages. These differences in narrative style were unexpected consequences of the two lexicalization patterns posited by Talmy. Importantly, the differing ways of expressing Manner and Path in verb- and satellite-framed languages have an impact on narrative style which would not have been observed unless studying language in use. Berman and Slobin (1994) and Slobin (1996, 1997a, 2004) compare the way motion is described in the two types of languages and conclude that their different lexicalization patterns have resulted in different narration styles. The main differences are that manner of motion is more salient in S-languages than in V-languages, and S-languages tend to elaborate more on the description of the path. For instance, regarding Spanish and English, Berman and Slobin (1994) and Slobin (1996a, 2004) found that expressing Manner, English elicited narratives displayed a greater token and type frequency of manner verbs than Spanish elicited narrative which in turn contained a higher number of path verbs. Furthermore, in Spanish narratives, manner tended to be subordinated, i.e., expressed in optional constituents such as adverbs, gerundives, subordinated clauses, etc., since the Spanish main verb typically encoded Path of motion. In terms of the expression of Path of motion, Spanish speakers tended to use bare motion verbs, i.e., ‘verbs with no elaboration of path beyond the inherent directionality of the verb itself (Slobin, 1996a, 2000) more often than English speakers, who frequently added locatives and directionals to their motion verbs, as (32) illustrates University of Ghana http://ugspace.ug.edu.gh 65 32. Se cayó. 3SG fall (Lit.) ‘S/he fell’ In addition, in Spanish narratives, just one piece of information about the ground was usually given, either the source (e.g., from X) or the medium (e.g., to, towards X). English speakers, in contrast, mentioned more ground elements per clause. In general, Spanish speakers tended not to express complex paths with a single verb; when a complex path was described together with one motion verb in Spanish, the trajectory usually involved the motion from a source to a goal as in (33). In contrast, in English, it is morphosyntactically possible to attach several path segments to a single verb. Thus, English speakers readily compacted several trajectories (expressed by satellites) with a single verb as in (34). 33. Se cayó de la ventana a la calle 3SG fall from the window to the street (Lit.) S/he fell from the window to the street 34. He threw him over a cliff into a pond. Finally, when Spanish speakers described complex paths, they typically did so by using several verbs and breaking the event into several segments. 35. El viervo le Ilevó hasta un, sitio, donde debajo había un río; Entonces el ciervo tiró al perro y al ni o al rio. Y despues cayeron. University of Ghana http://ugspace.ug.edu.gh 66 Lit: The deer took him to a place, where below there was a river. Then it (deer) threw the dog and the boy to the river. And after that they fell. The same patterns observed in oral elicited narratives are widely attested in novels (e.g., Slobin, 1996a, 1996b, 1997; Mora-Gutiérrez, 2001) and newspapers across a range of languages (e.g., Slobin 2003; Slobin 2006a. Writers of creative fiction and reporters, though free to make full and imaginative use of the language, conform to the same pattern found in the frog stories. In terms of Manner, writers using satellite-framed languages devoted much more attention to Manner of motion than writers using verb-framed languages. Özҫaliṣkan and Slobin (2003) provided interesting insights into the differing functions that manner expressions beyond the verb phrase seem to play in novels written in satellite-framed languages such as English and in verb- framed languages such as Turkish. Turkish writers tended to accompany their path verbs with adjuncts expressing Manner to compensate for the impossibility to encode Manner in the verb on many occasions (such as in boundary-crossing events, etc.). Thus, the high rate of such manner expressions beyond the main verb in Turkish suggests a compensatory function of these expressions. In contrasts, manner verbs in English were very often accompanied by manner adjuncts, which further contributed to and enriched the manner details of the motion scene expressed by the main manner verb. Thus, such manner expressions serve an augmentative function in English. In terms of Path, novels written using satellite-framed languages provided their readers with more elaborate descriptions of path or trajectories of movement than novels within in verb-framed languages. In contrast, writers using verb-framed languages offered University of Ghana http://ugspace.ug.edu.gh 67 more information about the physical setting and the protagonists psychological state, which allowed their readers to draw inferences about the manner in which the protagonist moves and the trajectories he or she follows. Oh (2003) found that manner verbs were more heavily used in novels written in English (an S- language) than those written in Korean (a V-language). The Frog stories considered in volume II of Strömqvist and Vernhoeven (2004) offer a more complex picture of the crosslinguistic differences in the expression of motion. In the light of the data from a large number of languages, the rigid dichotomy of verb- and satellite-framed languages has been questioned (e.g., Ibarretxe-Antunãno 2004a, 2004c; Slobin 2004; Zlatev and Yangklang, 2004). A third typological group was thus proposed, equipollently-framed languages, (see Section 2.1.6) for those languages whose morphosyntactic constituents expressing Path and Manner have equal status. On the other hand, it was argued that it could also be possible to rank languages in clines or continuums of manner and path salience, as languages within the same typological group display different degrees of elaboration of manner and path. Takashashi (2009) discusses Thai motion event expressions in relation to verb-serialisation. He notes that the entire structure of motion event expressions in a verb-serializing language is difficult to formulate because of the expressions’ diversity and complexity. Comparing three studies (Thepkanjana, 1986/2006; Muansuwan, 2002, and Kessakul, 2005), he concludes that there is evidence that the syntax and the semantics of Thai motion event expressions cannot be straight University of Ghana http://ugspace.ug.edu.gh 68 forwardly analysed with the principles of the linguistic theories that have been developed based primarily on analysis of Indo-European languages. He cautions: “Linguists working on Thai can draw on the theories only after carefully examining whether the preconditions of the theories fit the nature of Thai grammar with caution.” Takashashi (2009:42). Another important outcome from Strömqvist and Verhoeven’s (2004) volume has been the proposal or identification of a number of linguistic and non- linguistic factors which interact with lexicalization patterns. Slobin (2004) pointed out that the Talmian typology alone cannot account for language use, and suggested that there are a number of additional linguistic factors playing a part in the expression of motion. A language provides its speakers with a range of lexical and morphological means for describing motion events. It has also been shown that sound symbolic expressions contribute as well to motion event descriptions in some languages, such as Basque (e.g., Ibarretxe-Antunano, 2004b), Japanese (e.g. Hamano, 1998, Kita, 1997), Ewe (Ameka and Essegbey, 2001/2013, 2006). Examples of sound symbolic expressions in Zulu, Emai and Ewe, (taken from Slobin 2004:233) are: 36. a. gulukudu ‘rush in headlong’ (Zulu) b. kítíkítí ‘at a stomp’ (Emai) c. minyɛminyɛ ‘stealthily’ (Ewe) University of Ghana http://ugspace.ug.edu.gh 69 In addition, a variety of non-linguistic factors, including cognitive and cultural factors have to be taken into consideration to fully account for patterns of narrative style. Processing load (as stated in Section 2.2.2) is a cognitive factor that might account for the tendency to omit manner information in verb-framed languages. Although these languages have the option of using adjuncts or subordinate constructions to express Manner of motion, this alternative is avoided most of the time. The reason seems to lie in the heavy processing of these constructions in terms of speaker’s production and hearer’s comprehension (Talmy, 2006:128). With regard to cultural factors, as Bavin (2004:17) suggested: ‘[c]ultural values will influence what a speaker determines as important when telling a story’. Wilkin’s (2004) and Bavin (2004) pointed out that culture seems to play a role in the great attention Arrente and Warlpiri speakers give to path details. Australian aboriginal culture highlights the importance of journeys as well as locations, and this is without doubt reflected in narrative style. Thus, in their frog stories, Arrente and Warlpiri speakers overall described scenes with many more path fragments than speakers of other languages. Moreover, Bavin reported cultural preferences for repetition of old information together with new information. Information is repeated with some new added […] telling the frog story, a speaker might give the information that someone fell to the water, then someone fell down to the water and then specify that the child and dog fell. (Bavin, 2004:20) University of Ghana http://ugspace.ug.edu.gh 70 In Akan, a Kwa language, some research has been carried out on motion events. Drawing data from naturally spoken and written texts, Afreh (2011) explores the lexical and syntactic resources Akan uses for the expression of Motion (in the sense of Talmy 2000) and the semantics of motion verbs in Akan. In addition, the thesis also considers metaphorical motion in Akan by examining the types of metaphorical mappings that are structured by spatial motion. The study reveals that the components of Motion, Manner and Path are mainly expressed through verbs. Path is also expressed through postpositions. It is shown that the Path verbs in Akan encode different components of a single trajectory. Another finding of the study is the persistent use of serial constructions to convey information about Motion, Manner, Cause and Path. It has been established that speakers of Akan use posture verbs not only to describe the location and specify the spatial configuration of humans and animals, but also of inanimate entities, in which case the central meanings are extended for the purpose. An examination of some of the metaphorical motion events has revealed that the target domains in such expressions can be classified into four domains; those that relate to time, those that relate to emotional states, those that relate to events, and miscellaneous ones. It was also found out that Akan has the Moving Time, the Moving Ego and Sequence as Position on a Path metaphors. In sum, Talmy’s dichotomy of satellite- and verb-framed languages provides valuable insights into lexicalization patterns of motion in language. It has been shown however, that lexicalization patterns alone cannot account for how language is used in narrative discourse. Verb- and satellite-framed rhetorical styles could not have been understood, except through a close examination of University of Ghana http://ugspace.ug.edu.gh 71 language in use. Furthermore linguistic and non-linguistic factors should be taken into consideration so as to offer a more complete picture of how different languages talk about motion. 2.3.4 First Language Acquisition Event description is common in our lives. Our daily conversation usually involves motion event descriptions (Johnson 1987; cited in Chen, 2005). Besides, motion verb and verb complement are two pivotal elements in motion event description, indicating how the object moves and its movement trajectory respectively (Slobin, 1996). Studies by Clark (1993), and Pulverman, Hirsh- Pasek, Golinkoff, Pruden and Salkind (2008) found that motion verbs like fall and jump play a crucial role in the acquisition of children’s early verb lexicon. It was reported that English-speaking children started to use path satellite (e.g. up) and manner verbs (e.g run) in single-word utterances around 14-16 months and 17-18 months respectively (Choi and Bowerman, 1991). This suggests that children start talking about movement and motion in space rather early in their language development. Other studies also found that Chinese-speaking children have more verbs than nouns in their early vocabularies and verbs tend to appear in the salient position in the sentence (Tardiff, 1996; Choi and Gopnik 1995). Therefore, investigation of the acquisition of motion events is important as it could provide information on early verb development in children. It has been attested that cross-linguistic differences can be observed in how children talk about motion in elicited oral narratives (e.g. Berman and Slobin, 1994; cited in Cifuentes-Férez 2008:74; verbal descriptions of video clips University of Ghana http://ugspace.ug.edu.gh 72 (e.g., Hickamn, 2006; Oh, 2003) and spontaneous conversations, (e.g., Semilis and Katis, 2003). Research using the frog story picture book as an elicitation tool has shown that children acquiring their mother tongue are guided by the set of distinctions in their language to attend to specific features of events while speaking (Berman and Slobin, 1994; Ozçaliskan and Slobin, 1999; Slobin, 1996a, Slobin, 2001). Generally speaking, children speaking satellite-framed languages use a higher percentage of tokens and types of manner verbs than children speaking verb-framed languages, who tend to use path verbs instead. As Slobin (2003) hypothesized, manner-of-motion verbs are acquired very early by children learning satellite-framed languages. Furthermore, children speaking satellite-framed languages use motion verbs with some path elaboration, whereas such descriptions are hardly used by children speaking verb-framed languages. As age increases, they are capable of providing more detailed descriptions of the motion event in line with the adult system. Verb-framed-language speaking children are more concerned with setting the scene and asserting changes of location (results) than satellite-framed-language speaking children, who assert trajectories and attend to Manner of motion (actions). Oh (2003) elicited verbal descriptions from 3-year-old Korean-speaking and English-speaking children using clips depicting everyday motion by humans. In line with the previous findings, she also found that children are influenced very early by the typological properties of their native language. Korean- speaking children produced manner verbs less frequently and path verbs more frequently than English-speaking children, who produced verbal descriptions encoding both Manner and Path. On the other hand, some common University of Ghana http://ugspace.ug.edu.gh 73 characteristics were noted between Korean-speaking and English-speaking children. At an early age, both groups rarely expressed Path and manner together in their descriptions, though English children did so more frequently. Oh speculated that at an early age children might be too limited cognitively to attend to various components of a motion event at the same time, though another possibility might be that they have not yet mastered the skill to pack both manner and path information into a single description. In addition, it was found that children from both language groups seemed to be more biased toward Manner than toward Path, and that they tended to produce coarse manner of motion verbs (such as walk, or run). Hickman (2003, 2006) used animated cartoon on a computer screen to elicit verbal descriptions. English and French as satellite-framed and verb- framed languages respectively strongly influence how children (three- and five- year-olds) talked about motion. In line with previous findings, she noted that English- and French-speaking children used more complex descriptions of both Path and Manner together with increasing age. In terms of spontaneous or naturalistic conversations, Selimis and Katis (2003) showed that Greek- and English-speaking children’s descriptions followed the typological patterns of their mother tongue in preferring verbs which lexicalize either Path (for Greek) or Manner (for English). Moreover, it was found that these differences appeared even earlier than two years of age, but then diminished with age as the manner verb lexicon in Greek children got richer in later development. University of Ghana http://ugspace.ug.edu.gh 74 Cross-linguistic research have also unveiled variations in how children acquire spatial language (e.g., Bowerman, 1996; Bowerman and Choi, 2001; Casasola, 2005; Choi and Bowerman, 1991; Choi et al, 1999 all cited in Cifuentes-Férez 2008). Such variability across languages casts doubts on the existence of universals in the acquisition of spatial language, and suggests that language-specific patterns have an impact on linguistic and cognitive development. At an early stage, children are guided by their language and seem to construct a spatial language which closely fits the adult system. Bavin (2004) also reports on the uses of locative markers in Walpiri children’s and adult’s narratives. The findings show that narrative length, the frequency and type of locative elements differ across age groups – older speakers produce longer narratives and refer to more specific locations. For instance, in addition to mentioning where the main characters were going, the older storytellers thought it was important to say where the boy and his dog were located and where they were looking. Brown (2004) examines position and motion in stories by Tzeltal- speaking children and adults. The study shows that even 3-5 year old children are already attuned to language-specific properties of Tzeltal narration, such as integration of position and motion/path in one clause. Engberg-Pedersen and Trondhjem (2004:62) shifting the focus to action in motion description in West-Greenlandic, also examined narratives by children and adults. Their findings show that West-Greenlandic differs from other verb- framed languages, such as Spanish; “the rhetorical style of the West-Greenlandic University of Ghana http://ugspace.ug.edu.gh 75 frog stories seems to be influenced not so much by what is ‘readily encodable as by what is coded semantically richly in the verb’ Regnarsdottir and Strömqvist (1997) compare time, space, and manner expressions in narratives elicited from children and adults in two closely related satellite-framed languages, Swedish and Icelandic. The findings of the study confirm the qualitative change in the discourse and structure of stories elicited from children between the ages of 5 and 9. In comparing the two languages, the authors have observed an increase in cross-linguistic similarity in grammatical development and an opposite tendency, that is a decrease in cross-linguistic similarity in lexical development. Wilkins (2004) also compares the discussion of motion in narratives told by children and adults in English and in Arrente, a language spoken in the deserts of Central Australia. This comparison shows that Arrente-speaking children differ from their English-speaking counterparts in the channeling of attention and complexity of the motion-event description. In Zlatev and Yangklang (2004), motion events are also central. The writers examine Thai narratives collected from children and adults. The authors argue that despite the fact that Thai expresses path through verbs and not satellites, it is not necessarily a verb-framed language, since on several dimensions it behaves like satellite-framed languages. Aksu –Koç and Tekdemir (2004) study narratives collected from Turkish- and English-speaking children and show that the two groups do not differ in terms of mindreading but do differ in terms of preferred narrative strategies – University of Ghana http://ugspace.ug.edu.gh 76 English-speakers favour topic shifting and Turkish-speakers topic maintenance, continuing the narration from the point of view of the boy. Küntay and Nakamura (2004) compare Turkish and Japanese narratives elicited from children and adults and suggest that these data do not show any age-related differences in the number of evaluative devices. The two groups did differ however in the types of devices used; for instance, Turkish narratives displayed evaluative remarks, while Japanese narrators did not use such remarks. Strömqvist, Nordqvist, and Wengelin (2004) expand the discussion of frog stories to include written narratives and compare thinking-for-speaking, with thinking-for-writing. Their study shows that 9 year olds are still influenced by thinking-for-speaking when thinking-for-writing, while 15 year olds behave more strategically when composing their written narrations. Overall, the volume represents a major advance in the study of narrative development and language typology. The contributors extend narrative inquiry to languages in which little narrative research has been conducted before, including but not limited to Arrernte, Basque, Icelandic, Thai, Tzeltal, Walpiri, and West-Greendlandic. They also represent an array of important findings in terms of development of written and spoken narratives across age groups, contexts and modalities. The key contribution of the volume is the argument, backed by evidence from several languages, that Talmy’s (1985) dichotomous view of verb-framed and satellite-framed languages is oversimplified and should be replaced by a scale or a continuum, where languages can show a combination of properties of verb- and satellite-framed languages. University of Ghana http://ugspace.ug.edu.gh 77 Chen (2005) investigates Chinese children’s first language acquisition and adult’s description of motion events embedded in the general theoretical framework of Talmy’s (2000) linguistic typology of motion events. Fifty-nine subjects in five different age groups of children (age 3, 4, 5, 9) and adults (between age 18 and 22) were involved. He found that Mandarin-speaking children did not seem to show any significant developmental difference in the use of different types of motion verbs. In his study, a motion event containing two motion elements (ie. manner and path) was defined as “manner expression whereas it was defined as manner + path expression in this thesis. He reported that the percentage of “manner expressions” was always higher than that of path expressions across ages. Similarly, Chau (2006) studied motion event expressions in Cantonese narratives. A total of sixty children aged 3, 4 and 5 and twenty adults participated in telling the frog story. She found that the adult group produced significantly more types of manner and path verbs than the children’s groups but with no statistically significant differences among the children’s groups. Whether a developmental group trend exists in the use of motion event expression is thus still a controversy amongst linguists and further investigation is required, she added. (Chau 2006, cited in Cheung Lai Yee 2008:8). Chen and Guo’s (2010) paper reports results from the analyses of three different types of data; one from the adult speakers’ narratives, one from elicited children’s spoken narratives from age 3 to 9 and one from written narratives in nine contemporary novels. The three types of data reported in this paper show a converging pattern of language use regarding motion expressions. Regarding University of Ghana http://ugspace.ug.edu.gh 78 diversity of Manner verbs, the three sets of data show consistently that Mandarin speakers have very diverse and rich lexicon for Manner verb, very similar to the patterns found among S-language speakers. Whereas the original binary typology distinguishes between verb-framed and S-framed languages, Chen and Guo’s cross linguistic comparison of the structural and discourse patterns of motion event descriptions treat Chinese as an equipollently-framed language exhibiting hybrid pattern associated with both English and satellite-framed languages. Section (2.3.4) has concentrated on child language acquisition research. A number of studies discussed in this section have concluded that children are influenced very early by the typological properties of their languages. Verb- framed-language speaking children are more concerned with setting the scene and asserting changes of location (results) than satellite-framed language speaking children, who assert trajectories and attend to manners of motion (actions). Irrespective of language, it has been shown that older children are closer to the adult system than younger children. At an early age, children rarely express Path and Manner together in their descriptions, though English children do so more frequently than verb-framed language speaking children. With increased age, children are capable of providing more detailed descriptions of motion events in line with the adult system. It might therefore be concluded that children are trained to think-for-speaking in terms of their mother tongue, acquiring some language-specific properties earlier than others. All in all, these studies provide linguistic data suggesting that children adopt language-specific characteristics from a very early age. However, some characteristics seem to be mastered early, while others develop more slowly. University of Ghana http://ugspace.ug.edu.gh 79 Findings from the studies above go to support Bowerman and Choi’s (2001:505) suggestion that ‘children are sensitive to language-specific categorization principles from their earliest productive uses of spatial forms, and at least in some cases in comprehension even before production begins’ and Hickman’s (2006: 281) study also suggests that ‘[L]language properties influence how children select or organize spatial information during the course of development’. 2.4 Linguistic Relativity Research12 The domain of motion has received special attention in linguistic relativity research. According to Pourcel (2002: 127, cited in Cifuentes-Férez 2008:82) motion events are an ideal arena for investigating Whorfian effects for a number of reasons. First, languages differ in how they carve up the domain of motion. Second, motion events are referred to in language use with a high level of frequency, and they are representative of actual language use. And third, unlike the domain of colour, in this domain there are no ‘biologically-determined concepts waiting to be labelled’ (Slobin, 2000:122). Research on linguistic relativity effects in the domain of motion addresses whether the differences in lexicalization of Path and Manner result in divergent conceptualizations of motion events. More concretely, these research focus on whether the selective lexicalization of Manner, optional in verb-framed languages but readily expressed in satellite-framed languages, ‘entails 12 Information on Linguistic Relativity research is drawn from Cifuentes-Férez (2008). University of Ghana http://ugspace.ug.edu.gh 80 differential levels of cognitive salience of that variable across verb- and satellite- framed language speakers, resulting in overall divergent conceptualization of motion events’ (Pourcel, 2005: 126). A number of studies have been carried out to find out whether motion description has any bearing on cognition. Section 2.4.1 offers a summary of research which provides positive evidence for the claim that cross-linguistic differences in the encoding of motion, have an impact on the way we think about motion, in particular on mental imagery, on memory for manner of motion and on how people form new concepts. In contrast, research on the negative side suggesting that language does not affect cognition is assessed in Section 2.4.2. Even though the same methods are approximately followed, studies in this arena have produced mixed results. In light of this conflicting evidence, Section 2.4 reviews research by Pourcel (2004a, 2004b, 2005) and Kopecka and Pourcel (2005) which suggest that the intrinsic nature of the stimuli employed may be responsible for a significant portion of the divergent results. 2.4.1 Positive Evidence As we saw in the introduction, Slobin (2003) hypothesised two cognitive consequences concerning linguistic effect on imagery and memory: speakers of satellite-framed languages will have rich mental imagery of Manner of motion, and Manner of motion will be more salient in their memory. Slobin (2000) reported richer mental imagery of manner of motion in English speakers than in Spanish speakers. In a suggestive experiment, he gave passages from novels to be read to English and Spanish native speakers as well University of Ghana http://ugspace.ug.edu.gh 81 as English and Spanish bilinguals, asking them later to report mental imagery for the protagonist’s Manner of movement. The examples were extracted from Isabel Allende’s La casa de los espiritus, and did not contain manner verbs, but the author provided information about the setting and the protagonist’s inner state allowing inferences of manner of motion. His findings showed elaborated mental images for Manner in English reports and less vivid or poorly detailed mental images for manner in Spanish. Interestingly, bilinguals behaved according to the text’s language: if the text was in English, and therefore, they were asked to provide reports in English, their mental images for manner were significantly richer than when the text and the task were in Spanish. Kershten et al. (2003) showed that greater attention to Manner of motion by English speakers relative to Spanish speakers is revealed as well in learning tasks. English and Spanish monolingual participants viewed animated cartoons in which bug-like creatures moved along various non-nameable paths in various non-nameable manners. They were told that those creatures belonged to four different species and they had to guess which by pushing one of four buttons. After each choice they were told if they had guessed correctly or not. On the whole, no differences in how long they took to learn to distinguish the bug-like creatures in terms of Path was observed, but English speakers were significantly faster in learning to distinguish them in terms of Manner despite the fact that none of these manners of motion could be lexicalised in English manner verbs. Furthermore, they also tested bilinguals in Spanish and in English, and their results were similar to those found with monolingual speakers. They concluded that the observed differences between the two groups regarding learning Manner University of Ghana http://ugspace.ug.edu.gh 82 of motion concepts, were a result of the different languages they spoke. In sum, people learn to attend to the sorts of event details that are encoded in their language. Oh (2003) reported differences in recall of manner details between English and Korean monolingual speakers. She devised an experiment in which the initial task of the participants was to view and describe some video clips. The clips included several manner of motion (e.g., walking, striding, jogging, strolling, trudging, sprinting, etc.) as well as several paths (e.g., out of, up, down, around, along, etc.). Also, some filler clips which did not depict motion or displacement were included (cooking in a kitchen and working with a computer). As predicted for this task, English speakers provided more tokens and types of manner verbs and extended manner descriptions than Korean speakers. The two language groups differed according to their language-specific lexicalisation patterns. The difference across languages was more pronounced in the descriptions of Walking-and-boundary-crossing video clips. After participants completed this task, they were presented with an unexpected task. In this surprise task, they were told to compare their memory of each clip with a standard clip of the same actor walking at a normal rate and then answer a questionnaire on details of the events. The results showed that English speakers were more accurate in identifying the length of stride and degree of arm swing, in the original clips than Korean speakers, although those specific manner details were not present in their verbal descriptions from the initial task. Overall, English speakers performed better than Korean speakers on memory for manner information and answering manner questions about the videotapes. University of Ghana http://ugspace.ug.edu.gh 83 Pourcel (2005) compared English and French speakers’ performance on two recall tasks. English and French speakers were asked to view a short extract from a silent Charlie Chaplin film; City Lights, and they had to recall the scene verbally (free prose recall). 24 hours later, participants were presented with a questionnaire about the film in which they had to recall varying aspects of the film extract. Like Oh (2003), she used real-life motion, but her stimulus represented contextualized motion events whereas Oh’s stimuli were not fully contextualised. Error rates in immediate recall (after watching the short film) indicated significant cross-linguistic differences in line with the semantic dimensions highlighted in French and English. French showed better recall of agent details and path types (i.e., telic and atelic paths), but worse recall of manner types (i.e., default manners e.g., run, walk) were equally recalled by both groups. These findings suggest that English speakers pay closer attention to fine- grained manners than do French speakers. Pourcel (2005: 285) concluded that difference in memory across the two languages ‘appear to be in line with the conceptual dimensions highlighted in French and English prototypical linguistic encoding of motion, suggesting the possibility of relativistic effects of habitual language patterns on memory.’ 2.4.2 Conflicting Evidence Before reviewing those studies yielding contradictory results, it is necessary to provide a brief summary of the experimental paradigms adopted in order to better understand the findings and conclusions. Most of this research usually combine three tasks; naming, similarity judgments and memory tasks (recall and recognition). In naming tasks, speakers are told to either fully University of Ghana http://ugspace.ug.edu.gh 84 describe or just label with one word what is happening in visual motion video clips (e.g., Oh, 2003; Pourcel, 2005). In similarity judgments, the triad paradigm is preferred. Triads consist of three video clips: one is the target which displays both a Path and a Manner, and the other two video clips are the path- and manner- alternates. That is one of the video clips shares the same Manner as the target, and the other shares the same Path. The three videos in a triad usually display the same Figure (human or any other entity that moves) and Ground, so as not to bias participants responses. Other tasks examine participants’ memory, either through recall (i.e., memory for details of what they saw) or recognition tasks (i.e., just asks participants if the video clip they are watching is ‘new’ or ‘old’), Gennari et al. (2002) were interested in investigating whether different lexicalization patterns of motion events in English and Spanish had any effect on speakers’ performance in two non-linguistic tasks: recognition memory and similarity judgments. They also investigated Slobin’s thinking-for-speaking hypothesis asking some participants to verbally describe the motion events but not asking others. The stimuli consisted of 36 triad of videotaped human motion. Overall, no significant difference between English and Spanish speakers was obtained in their memory for Manner of motion. However, they did find a linguistic effect in the similarity task after verbal encoding. Spanish speakers in the verbal description condition were more likely than English speakers to select the clip with same path. Thus, the verbal coding task had an effect on Spanish speakers’ later performance in the similarity task. In contrast, English participants behaved in the same way whether they linguistically described the University of Ghana http://ugspace.ug.edu.gh 85 stimuli before or after the similarity judgments. Finally, when no linguistic encoding was required, performance across languages did not show any preference for one dimension of motion over the other. Their research suggests that linguistic and non-linguistic performance are dissociable, but language made available in the experimental context may mediate the speaker’s performance in other tasks. Papafragou et al (2002) conducted a study similar to Gennari et al. (2002), using English and Greek (a verb-framed language). Their participants were both children and adults. The tasks included (a) verbal descriptions and recognition memory of static pictures adapted from the frog story, and (b) similarity judgments using triads of static human motion pictures. They found that the two language groups differed in terms of their linguistic descriptions, but their performance in the two non-linguistic tasks was similar. Unlike Gernnari et al. (2002) they did not find a facilitating effect of verbal encoding in later non- linguistic tasks. Taken together, however, these two studies are consistent with the Universalist approach to language and cognition: cross-linguistic variability in the expression of motion does not reflect any substantive differences in the ways humans think about motion. Zlatev and David (2004) focused on French and Swedish (a satellite- framed language). They used sets of triads showing tomato man13 performing the motion scenes, which varied in terms of Manner (rolling spinning, sliding 13Tomato man is a virtual elicitation tool developed at the Max Planck Institute for Psycholinguistics (Nijmegen, NL). University of Ghana http://ugspace.ug.edu.gh 86 and jumping) and Path (e.g., from right to left). As in previous studies, participants saw triads and were asked to choose which alternate is most similar to the target. Their results showed an overall preference for Manner across languages. Unlike the research we have just reviewed, which concluded that there was not a preference for either motion dimension over the other, Zlatev and David’s results showed a significant preference for the dimension of Manner. In sum, this section reviewed research addressing the linguistic relativity hypothesis. This line of research addressees whether differences in the lexicalisation of Path and Manner in verb- and satellite-framed languages has an influence on speakers’ conceptualization of motion. This question has been extensively investigated, yielding mixed results for the influence of language on cognition. On the one hand, several studies suggest that speakers of satellite- framed languages seem to have richer mental imagery of Manner of motion, and that the Manner dimension of motion appears to be more salient in their memory for events than for speakers of verb-framed languages. On the other hand, a number of studies which suggest that language does not influence non-linguistic cognition have been reported. Although both sets of studies use similar methodologies, their results are contradictory. Research by Pourcel (2004a, 2004b, 2005) and Kopecka and Pourcel (2005) shed some light on this conflicting evidence by demonstrating that the stimuli employed in linguistic relativity research may be responsible for such divergent findings. These scholars show that the intrinsic nature of the stimuli, specifically, the types of Path, Manner and Figures, appear to have a significant effect on the salience of Manner and Path across typologically distinct languages. University of Ghana http://ugspace.ug.edu.gh 87 University of Ghana http://ugspace.ug.edu.gh 88 CHAPTER THREE FORMAL ENCODING OF MOTION EVENTS IN EWE 3.0 Introduction This chapter contains introductory material about the Ewe language and gives a description of the four basic semantic components of motion events as proposed by Talmy (1985) and Slobin (2000) and how these elements are encoded in the language. This will help in the classification of Ewe in the motion event typology. 3.1 The Language and Its Speakers Ewe14 also written as Eʋe [әße], Evhe, is a major dialect cluster of Gbe or Tadoid (Capo 1991, Duthie 1996) that belongs to the sub-group of the Niger-Congo family (Steward 1985, Williamson and Blench 2000 cited in Ameka 2005). It is spoken in the south-eastern part of the Volta Region of Ghana across to parts of southern Togo and across the Togo-Benin border by about three million people. Ewe dialects vary enormously. Groups of villages that are two or three kilometers apart use distinct varieties. Nevertheless, across the Ewe-speaking area, the dialects may be broadly grouped geographically into coastal or southern dialects, e.g., Aŋlɔ, Tɔŋú, Avenor, Dzodze, Watsyi and inland or northern 14 The information on the Ewe language is not different from the explications of other writers (Duthie 1996, Capo 1991, Ameka 1992, Ameka and Essegbey 2006, Essegbey 1999 etc.) University of Ghana http://ugspace.ug.edu.gh 89 dialects characterized indigenously as Ewedomegbe, e.g. Ho, Kpedzĕ, Hohoe, Peki, Kpando, Fódome, Danyi, Kpele etc. (Agbodeka 1997, Gavua 2000, Ansre 2000). Speakers from different localities, however, understand each other and can identify the peculiarities of the different areas. In spite of dialectal differences, motion events are expressed the same way. In addition, there is a written standard that was developed in the nineteenth century based on the regional variants of the various sub-dialects with a high degree of coastal content. With it, a standard colloquial variety has also emerged (spoken usually with local accent), used very widely in cross-dialectal contact sites such as schools, markets, and churches15 (Ameka and Essegbey 2006:359). 3.2 Phonology and Morphology Ewe is a tone language with high and non-high tonemes. Complex rising and falling tones also occur. It has a seven vowel system. Each of these has both an oral and a nasalized counterpart. There are 30 consonants including double articulated labial velar stops and bilabial fricatives written ƒ and ʋ that contrast with labio-dental fricatives /f/ and /v/. Similarly, there is a voiced apical post- alveolar stop /‘ɖ’/ which contrasts with a voiced dental stop /d/. 15 The ‘Klo afí ká nèle’ narrators used their various dialects. As such their language is the spoken form and hence does not necessarily conform to the expectations of someone familiar with he standard dialect; for example, the use of yè/yi/xe ‘when’ to introduce the adverbial clause instead of the standard written form esi (me) ‘when’. University of Ghana http://ugspace.ug.edu.gh 90 Tones may also fuse when morphemes combine in much the same way that the vowels may fuse. Tone is not customarily marked in the traditional orthography except on a few items with identical segmental forms to distinguish them. Thus, è second person singular pronoun (2SG) has a low tone compared with e third person singular pronoun which is not marked for tone. Similarly, lé ‘catch’ contrasts with le ‘be at’. A practice to mark all high tones in addition to the customarily marked low tones in the orthography introduced by Duthie (1996) is gaining currency in academic linguistic writings.16 Ewe syllable structure may be represented as (C1) (C2) VT (C3). Each syllable bears a tone. C1 may be any consonant except r, C2, a liquid as in ƒle ‘buy’, trà ‘to collapse’, ‘become watery’; or a palatal or a labial velar approximant as in fia [fya]‘show, teach, burn’ and sue [swe] ‘small’. The nucleus may be a vowel or the bilabial or velar nasal, in which case it carries tone. For example, ŋ di ‘morning’, ŋ keke ‘day’, ƒom ‘beat me’ ɖum ‘bit me’. The nucleus may also be filled by two vowels which are the same, yielding a long vowel or yielding a diphthong e.g. /ao/, /ai/. For example: dzaa ‘welcome’, kpáó ‘no’, yoo ‘Ok’. C3 is only filled by a nasal as in boŋ ‘rather’, sɔ ŋ ‘several’, kam = pé ‘scissors’. 16 Duthie’s (1996) approach to the marking of tones was adopted in the writing of this thesis. University of Ghana http://ugspace.ug.edu.gh 91 Morphologically, Ewe is isolating with agglutinative17 features. It makes use of compounding as well as reduplication, triplication and affixation processes in the formation of new words. Nouns have a non-high tone vocalic prefix à- or è- which are relics of Proto Niger Congo noun class markers. The è- prefix tends to be elided when the noun is not said in isolation e.g. a-me ‘prefix-person’, (e-) gli ‘prefix-folktale’. Some temporal nouns have a high tone prefix and they are never elided, égbe ‘today’, éto ‘to night’, and ázɔ ‘now’. Ewe has ideophones – a set of sound symbolic words with interesting phonological and syntactic properties, some of which code manner (or motion) concepts as we shall see in this section. Westermann (1930:107-109) gives forty ideophones that can be used in collocation with the general motion verb zɔ ‘move, travel,’ “according to the manner of going” Westermann (1930:107). 3.3 The Syntax of Ewe Syntactically, Ewe is an SVO language with alternative OSV and SOV orders being systematically linked to the basic one, and determined by semantic and pragmatic factors like topicalisation and focusing (cf. Ameka 1998. Essegbey 17 An isolating language is a language in which almost every word consists of a single morpheme. It is a linguistic typology category that defines a language with a low morpheme-per-word ratio. In the extreme case of isolating language or analytic language words are composed of single morphemes. On the other hand, an agglutinative language is a language where each affix typically represents one unit of meaning (such as “diminutives, ‘past tense’, ‘plural’ etc.), and bound morphemes are expressed by affixes (and not by internal changes of the root of the word, or changes in stress or tone). Additionally, and most importantly, in an agglutinative language affixes do not become fused with others, and do not change form conditionally by others. University of Ghana http://ugspace.ug.edu.gh 92 1999). In Ewe, who does what to whom is expressed by constituent order where the doer comes first followed by what is done (the verb) followed by one to whom it is done (the object). This can also be followed by the recipient action hence Ewe is an SVO language. The following examples illustrate this. 1. a. Danye to fufu Da – nye tó fufu Mother-POSS pound fufu ‘My mother pounded fufu.’ [SVO] b. Fufue danye to [OSV] fufu - e da – nye to Fufu - FOC mother-POSS pound ‘It was fufu my mother pounded’ c. Dzodoƒee danye toe le . [SOV] Dzodoƒe - e da - nye to – e le. Kitchen - FOC mother-POSS pound 3SG LOC ‘It was in the kitchen that my mother pounded it.’ Clausal negation is marked by a discontinuous negative morpheme: mé….o. Me occurs just before the VP and tends to be cliticised onto the first element in the VP while o occurs at the end of the clause but before sentence final utterance particles as in (2a) and (2b). 2. a. Yawa megayi aƒeme o. Yawa  me – ga - yi aƒeme o. Yawa NEG-REP-go home NEG ‘Yawa should not go home’ b. Megadzo o sea? Me -ga dzo o se - a? 3SG:NEG –REP leave NEG hear - UFP University of Ghana http://ugspace.ug.edu.gh 93 Don’t leave, you hear/understand? There are different kinds of nonclausal or constituent negation.18 One of these is the negative cleft construction in a clause, an NP or a predicate. Consider the following example. 3a. Abi makumaku aɖe le afɔ nɛɛ. Abi ma – ku - ma- ku aɖe le afɔ nɛ - ɛ. Wound NEG-die-NEG-die INDEF be LOC foot/leg DAT -3SG ‘S/He has some ulcer on the foot or leg.’ In this last example, the me part o the derivational negation is also marked in Ewe by the affix ma ‘un’, the privative marker. This affix is used in the derivation of adjectives and adverbials. It is usually prefixed to a verbal element and reduplicated together with it when necessary: mǎ–vɔ (NEG – finish) ‘everlasting’. Makumaku (ma-ku-ma-ku – NEG die-NEG-die ‘ulcer wound’/’will never heal’ Ma can occur with or without standard negation as in (4). 4. Menye numaɖumaɖue na wòdze dɔ o. Me - nye nu ma - ɖu - ma - ɖu - e na NEG-be COP thing NEG –eat NEG – eat –FOC cause wò - dze dɔ o. 3SG-fall sickness NEG ‘It is not his abstiencne from food that has made him fall sick.’ In example (4) above, this construction is used to emphatically negate a particular constituent in a clause, an NP or a predicate. The constituent that is 18 This information on negation is taken from Ameka (2001). University of Ghana http://ugspace.ug.edu.gh 94 thus negated is focus marked, either by the argument focus marker or the predicate focus marker. The language also has a logophoric pronoun ye (plural yewó) which is used in reportive contexts to designate the individual(s) (except for the first person) whose speech, thoughts, feelings and so on are reported or reflected in the linguistic context in addition to personal pronouns which are distinguished for subject and non-subject. The reflexive is formed by juxtaposing an independent pronoun (including the logophoric) to the noun ɖókui ‘self’ e.g. É-bé ye-sì ye ɖókui ‘she said that she (LOG) cut her (LOG) self’ as in (5a) ans (5b). Similarly, the reciprocal comprises a plural independent pronoun, and nɔe – wó ‘reciprocal –PL’. 5. a. Ebe ye sì ye ɖokui. E - be ye sì ye ɖo kui  3SG – COMP LOG cut LOG self ‘S/He said she/he cut herself/himself.’ b. Minya mia nɔewo Mi  -nya  mi a nɔe -wo 3PL-know 3PL self-PL ‘We know each other/one another.’ There are particles for framing discourse. A scene-setting expression may be preposed to the clause, and separated from the rest by a pause and/or marked by a particle lá or ɖé. (See examples (6) and (7). 6. Né ɖevíáwo kpã tsó suku lá, wóyia aƒéme. Né ɖevi  - a - wo kpã  tsó suku la wo  - yi - a aƒéme University of Ghana http://ugspace.ug.edu.gh 95 If child- DEF PL close from school TM 3PL go - HAB home ‘When the children close from school, they go home.’ 7. Nenye wòe ɖé avaa? Ne - nye wò - e ɖé à - vá a? If- be COP 3SG – FOC PART 2SG:POT – come Q ‘If it were you, will you come?’ If the preposed constituent is coreferential with a core argument of the clause, their relationship is indicated by an anaphoric pronoun in the clause. An argument of the clause may be front-shifted to the pre-core clause position for focus, that is before the subject slot but after the preposed constituent slot. The fronted element is marked by an argument focus marker. –(y) é. Typically, a gap is left in the slot within the clause structure where the fronted element would have occurred. 8. a. Kofí fi ga. Kofi fi ga Kofi steal money ‘ ofi stole money.’ b. Gae Kofi fi. Ga - e Kofi  fi Money – FOC Kofi steal ‘It is money ofi stole.’ Ewe has a speaker anchored two-basic demonstrative system with dialect variants: a proximal (sia, yi, ke) and a distal (kemá, má, kemí). The following are examples. 9. a. Ɖeví síá dím wóle. University of Ghana http://ugspace.ug.edu.gh 96 Ɖeví síá dí - m wó - le. Child – PROX search – PROG 3PL - PROG ‘It is this child they are looking/searching for.’ [standard dialect] b. Eyi dim mele. E  - yi - di  - m me - le 3SG – PROX - look/search – PROG 1SG - PROG ‘It is this one I am looking for.’ [coastal dialect] c. Eke di wóle. E - ke di  wó - le. 3SG- PROX look/search 3PL – be.at PRES ‘They are looking for this one. [inland dialect] d. Tsɔ ekema/ema/ekemi nam. Tsɔ ekema/ema/ekemi na - m Give/Take DIST/DIST/DIST DAT - 1SG ‘Give/Take this one for me.’ The distal forms can be argumented by the suffixation of -i ‘deictic’ to get kemi  ‘that yonder’ or by the adverbial particle ɖá ‘in the distance’: kemí ɖá ‘that in the distance’. The distal terms have truncated forms; kem, m which are always accompanied by a gesture either a manual, lip or head point. In a presentational construction, obligatorily accompanied by a gesture, the demonstrative terms occur by themselves as predicate as in Ama-é-má ‘That/there is Amá (Coastal) or Nye-é ké ‘Here I am/This is me’ (Inland)’. In the standard dialect, however, the focused NP and the demonstrative predicate are linked by the equative copula nyé ‘be’. In addition, the predicate has to be the pronominal form of the demonstrative as in Awó -e-nyé é -má ‘That/There is Awó’. Amuzu (p.c) however thinks that the Anlo dialect has a three demonstrative system: University of Ghana http://ugspace.ug.edu.gh 97 (esia) (ema) (emɛ a) Proximal distal extra distal 3.4 Ewe Verbal System Ewe is an aspect rather than a tense language like many other Kwa languages in West Africa19 because it does not have any overt marking for present and past time. Habitual aspect is the only category marked on the verb by a toneless suffix (n)a which inherits its tone from the preceding syllable. Preverbal markers express various modal and aspectual categories such as vá ‘ventive/eventually’ (see Ameka 2008, Essegbey 2008). For example, the verb in sentence (10) can be interpreted as denoting present or past time. 10. Ama yi xɔme. Ama yi xɔme Ama go room ‘Ama went into the room.’ A bare verb or the aorist form has a past semantics. For active verbs, this signals a past occurrence of the action. For incloative verbs, it indicates the prior occurrence of the change of state hence the state is current. The Potential can have future time interpretation in context. All these temporal interpretations can be reinforced by adverbials. The Potential and the Subjunctive are mutually exclusive with the Habitual (cf. Ameka and Dorvlo 2009). 19 See Ameka and Kropp Dakubu 2008 (eds.) for a detailed information on Aspect and Modality in Kwa languages. This volume explores the thesis that in the group of West African languages known as ‘ wa’, Aspect and Modality are far central to the grammar of the verb and the clause than Tense. University of Ghana http://ugspace.ug.edu.gh 98 Consider the following examples. 11. Kofi yia suku gbe sia gbe. Kofí yi - a sukú gbe siá gbe Kofi go – HAB school day every day ‘ ofi goes to school everyday.’ 12. a. Ele be wòavá kpɔm. E - le bé wò –a - vá kpɔ - m 3SG be.at PRES COMP 3SG-SUBJV come see - 1SG ‘S/He has to come and see me.’ b. Éle bé wóává kpɔ m É - le bé wó – á vá kpɔ - m 3SG be.at PRES COMP 3PL SUBJV come see – 1SG ‘They have to come and see me.’ 13. a. Wóàvá kpɔm. Wó - à - vá kpɔ - m 3PL – POT – come see 1SG ‘They will come and see me. b. Âvá kpɔm.  vá kpɔ - m 3SG:POT come see – 1SG ‘You will come and see me.’ In (12a) the underlying tone of the Subjunctive is high but it becomes low when preceded by a pronoun with a low tone. (See Essegbey (2008) for more details on the Habitual, Subjunctive and the Potential). The Potential differs from the Subjunctive both in its variant and in tone. Some dialects of Ewe have la- as the variant of the Potential, as opposed to the ná of University of Ghana http://ugspace.ug.edu.gh 99 the subjunctive. Secondly, the Potential has a basically non-high tone which is often realized as low (See examples (12a) and (13a) above). The potential captures the invariant meaning of the à morpheme, and the subjunctive, the (n)á segment which expresses wishes and command. Ewe also has the Progressive and Prospective aspect constructions. The prototypical function of the progressive construction is to signal that the states of affairs represented in the clause is on-going at the relevant reference time indicated by the verb. There are two verbs that can fill the verb slot in this construction: the suppletive locative verb set le  nɔ ‘be. at’ and a contact verb dze ‘contact’. The structure of the prospective construction is the same as that of the progressive except that the aspect marker that heads the aspectual phrase is the prospective marker ge ‘PROSP’ or its dialect variant gbe ‘PROSP’. The general meaning of the prospective construction is that the state of affairs characterized in the rest of the clause will happen after the reference time specified in the verb. (see Ameka and Dakubu 2008:215-289) for more details. Ewe has several multiverb constructions. A serial verb construction (SVC) is a sequence of two or more verb phrases (including any complements and adjuncts) without any marker of syntactic dependency used to express what constitutes one state of affairs. The VPs in the sequence occur within the same temporal frame, share the same mood (e.g. imperative) and can be formally marked for different aspect categories. The individual verbs can function as independent verbs in simple clauses (in the same form). All VPs in the series have one syntactic subject which is expressed only once on VP1. The VPs cannot be formally University of Ghana http://ugspace.ug.edu.gh 100 independently negated but they can be individually focused or questioned. There is no limit on the number of VPs that can constitute an SVC, except for restrictions on production and comprehension. For example, the sentence: 14. Đzoɖeke 20- e tsɔ ŋutsuví -é ƒú du vá de - e deer -DEF take boy DEF run come throw - 3SG ɖé tɔ - ɔ me ALL river DEF inside ‘The stag took the boy, ran and threw him into the river.’ is an SVC. The single subject is Đzoɖeke ‘stag’ The first verb is tsɔ ‘take’. The second verb is ƒu du ‘ran’, modified by the preverb vá ‘vententive’ and the third verb is da/de ‘throw’. 3.5 Ewe Adpositions Ewe has both preposition and postpositions (cf Ameka 2003, 2005). Prepositions in the language constitute a small closed class of less than ten elements, distinguished from verbs by the fact that they cannot occur with the habitual suffix –na. The prepositions provide the general orientation of a Figure (located object). It is demonstrated that spatial relations, such as those encapsulated in “the basic typological prepositions at, in and on in English (Herskovits 1986:9), are not encoded in single linguistic elements in Ewe but are distributed over members of different form classes in a systematic string. 20 We do not have “the deer” in Ghana. The animals that resemble the deer are ƒeɖekadzoɖekɛké (an animal that grows an extra horn every year), gbagbadzoɖeké (the unicorn), afiã or afidzoɖeké (the stag) and zi ‘the antelope’. These were names substituted for the ‘deer’ in the narratives and their English translations will be used as substitute for the deer. University of Ghana http://ugspace.ug.edu.gh 101 3.5.1 Prepositions They are distinguished from verbs by the fact that they cannot occur with the habitual suffix –na. Spatial prepositions are given in Table 1 with an indication of their verbal sources. Two other non-spatial prepositions, kple ‘with’ COMITATIVE/INSTRUMENTAL and ná ‘to/for’ DATIVE, also occur (see Ameka 1995, 2003b). Since the prepositions have evolved from verbs they have been referred to as a class of verbids (Ansre 1966a, 2000). Serial verb constructions (SVCs) are the channel for the general development of verbs into verbids (see Lord 1993, cited in Ameka and Essegbey 2006). Note that váséɖé ‘until’, for example, is the result of the compounding of the grammaticalized forms of three verbs, vá ‘come’, sé ‘stop’, ɖé ‘reach’. The combined semantics of these verbs is consistent with the meaning of the preposition. It is also important to note that there is no erosion or difference in form between the verbal source and the prepositional forms. From a semantic point of view, then, the grammaticalization of the verbs has resulted in the development of heterosemy of the forms (cf. Lichtenberk 1991 cited in Ameka and Essegbey 2006:368). University of Ghana http://ugspace.ug.edu.gh 102 Table 1: Ewe Spatial Prepositions Preposition Function Gloss Verbal Sources Source Gloss le ɖé ɖó* tsó tó va sé ɖe LOCATIVE ALLATIVE ABLATIVE PERLATIVE EXTENT ‘at’ ‘to, towards’ ‘from ‘through ‘up-to, until’ < le < ɖé < ɖó < tsó < tó < vá < sé < ɖe ‘be at’ ‘reach’ ‘arrive’ ‘originate from, come, arise’ ‘to pass (by)’ ‘come’ ‘stop’ ‘reach’ Adapted from Ameka and Essegbey (2006:368) *The form ɖé and ɖó and are alternants. ɖó occurs when the complement of the preposition is not adjacent to it. 3.5.2 Postpositions Postpositions constitute a closed class of about thirty members. They have evolved historically from nouns but now constitute a distinct form class which is not necessarily a sub-class of the nominal class. Postpositions have evolved mostly from body part terms. In the literature, only one, dzi ‘sky’ comes from a landmark term. Thus, Ewe exemplifies two of the sources, even if one only minimally, that have been noted in the literature for the development of adpositions from nouns (cf. Heine, Claudi and Hünnemeyer 1991, Heine 1998, Svorou 1994, all cited in Ameka and Essegbey 2006:369). University of Ghana http://ugspace.ug.edu.gh 103 Table 2 and 3 show that the sources of some postpositions are not entirely obvious. The postposition gbe is rather productive and occurs with several nouns to indicate the region where the figure can be found. It occurs in terms of different kinds of vegetation as in avegbe ‘forest area’, that is a place with forest; dzogbe ‘fire area’ grassland that is a place where grass is when one can set fire to; tógbe ‘mountain area’ (mountainous region). It is lexicalized with dzĭ ‘sky’ and anyí ‘down’ to form terms for upper region and lower region respectively, as in dzigbé ‘upper region’, ‘upper area’; and anyígbe ‘lower region, lower area’. University of Ghana http://ugspace.ug.edu.gh 104 Table 2: Simple Postpositions Post- positions Gloss Putative Source Gloss Dzĭ upper surface < dzĭ sky Ƒo flat horizontal surface < ƒo belly gbɔ place, vicinity unknown N/A Gbá Surface around something unknown N/A Gbé purpose unknown N/A Gbe area, region unknown N/A Xa beside < axa side (of body) Me containing region of unknown N/A Nŭ entrance, opening, end point < nŭ mouth ŋgɔ front < ŋgó forehead ŋú (tí) outer surface < ŋú (t)) skin, body Ta upper end, peak < ta head Té under, bottom unknown N/A Tó edge < tó ear Sí domain < sí hand Tables 2 is adapted from Ameka and Essegbey 2006 (pp. 368- 309). University of Ghana http://ugspace.ug.edu.gh 105 Table 3: Complex Postpositions Post-positions Gloss Putative Source Gloss Godó outside of, other/opposite side < go Side, direction, region, part, shore, beach, strand Godzí in the direction of < go+dzí bank + upper surface Góme part, region horizontal surface Tome hollow, interior