SITUATION TYPES IN THAI

CASSIE WALLACE

Presented in Partial Fulfillment of the Requirements for the Degree of MASTER OF ARTS IN LINGUISTICS

Payap University January 2020

Title: Situation Types in Researcher: Cassie Wallace Degree: Master of Arts in Linguistics Advisor: Assistant Professor Audra Phillips, Ph.D. Approval Date: 22 January 2020 Institution: Payap University, Chiang Mai,

The members of the thesis examination committee:

1. ______Committee Chair (Professor Christian Rathmann, Ph.D.)

2. ______Committee Member (Assistant Professor Audra Phillips, Ph.D.)

3. ______Committee Member (Elissa Ikeda, Ph.D.)

Copyright © Cassie Wallace Payap University 2020

ACKNOWLEDGEMENTS

Many people have shared their time, knowledge, and expertise with me throughout this study. I am deeply grateful for their contributions. I am especially grateful to the members of the Thai Deaf community who helped me in this project: Mr. Kittichai Piumsuwansiri, Mr. Narakon Chaona, Mr. Sumet, Mr. Suphakit Piumsuwansiri, Ms. Chadaporn Sasomsub, Ms. Chadapon Boonpakob, Ms. Ninlawan Nanthataiphob, Mr. Phurithut Khongbangpor, Mr. Anupup Chaowchat, Ms. Orathai Chompusai, Ms. Rattanaporn Topinjai, Mr. Phuriphong Saenchaiban, Mr. Apichat Pokeaw, and Mr. Jarin Tienmai. Your insight into ThSL and patience in explaining various nuances of meaning were invaluable. Thank you for your friendship and encouragement.

I would also like to thank Dr. Audra Phillips for the expertise and guidance she has provided. Thank you for walking me through such a complex and “fragile” topic. Your enthusiasm was often the incentive I needed to tackle the next step. I would also like to thank Dr. Tyler Heston for his help and encouragement during the initial stages of data collection and analysis. Thank you, too, to Dr. Christian Rathmann for your insightful questions and encouragement, and to Dr. Elissa Ikeda for your thorough reading and review of the thesis.

The encouragement of friends, family members, and classmates along the way have made the journey a pleasant one. Thank you for your kindness and prayers.

Finally, I would like to thank our good and gracious Father. He is the rock that is higher than I, the God who sees and provides. His has manifested His countless mercies, steadfast love, and perfect goodness throughout these years of study. He is worthy of all praise.

Cassie Wallace

i

Title: Situation Types in Thai Sign Language Researcher: Cassie Wallace

Degree: Master of Arts in Linguistics Advisor: Assistant Professor Audra Phillips, Ph.D. Approval Date: 22 January 2020 Institution: Payap University, Chiang Mai, Thailand Number of Pages: 128 Keywords: Thai Sign Language, situation types, event structure, aktionsart, Event Visibility Hypothesis

ABSTRACT

The dynamic, durative, and telic values of a predicate determine its situation type. These values are inherent features that can be uncovered systematically using test frames. However, the applicability of tests varies across languages, making it necessary to begin by identifying which tests work well in the language under study. Several researchers have developed tests successfully for spoken languages. However, only one such investigation has been conducted in a sign language.

This study begins to address this disparity by systematically identifying situation types in Thai Sign Language (ThSL) using test frames. These test frames were first applied to a set of 30 ThSL predicates and the clauses tested over several sessions with a group of Thai Deaf. Extended testing with two new groups was then done with 12 of the predicates. The results indicate that ThSL covertly lexicalizes at least six situation types: individual-level states, stage-level states, activities, accomplishments, achievements, and semelfactives.

The study also shows that the proposed morphological markings for telicity and duration (the Event Visibility Hypothesis) in sign languages do not account for the data found in ThSL. Furthermore, it shows that the bases for these proposed morphemes are shaky. Instead, the results of the study suggest that the phonological form of a predicate reflects various subevent properties of the encoded event.

ii

This is especially true of durative events since they involve multiple stages. For example, walk can be broken down into individual steps, each step representing a stage. The of the ThSL sign WALK reflects these stages. In contrast to clear, cyclic stages, certain events involve gradient change. For example, melt involves a gradient transition from solid to liquid. This change is reflected in the single, gradient movement of the ThSL sign MELT. The connection between the surface form of a predicate and its internal structure supports Kuhn’s (2017) iconic analysis of phonological movement rather than the EVH.

iii

ชื่อเรื่อง: รูปแบบสถานการณ์ในภาษามือไทย ผู้วิจัย: แคสซี่ วอลเลซ ปริญญา: ศิลปศาสตรมหาบัณฑิต (ภาษาศาสตร์) อาจารย์ที่ปรึกษาวิทยานิพนธ์หลัก: ผศ. ดร. ออดร้า ฟิลลิปส์ วันที่อนุมัติผลงาน: 22 มกราคม 2563 สถาบันการศึกษา: มหาวิทยาลัยพายัพ จังหวัดเชียงใหม่ ประเทศไทย จ านวนหน้า: 128 ค าส าคัญ: ภาษามือไทย, รูปแบบสถานการณ์, event structure, aktionsart, Event Visibility Hypothesis

บทคัดย่อ

ลักษณะทางพลวัตร ความต่อเนื่อง และหน้าที่ของภาคกริยาเป็นสิ่งที่ก าหนดรูปแบบสถานการณ์ คุณลักษณะเหล่านี้เป็นสิ่งที่สามารถเปิดเผยได้อย่างเป็นระบบโดยการใช้กรอบทดสอบ อย่างไรก็ตาม การใช้กรอบเหล่านี้แตกต่างกันไปในแต่ละภาษา ดังนั้น จึงเป็นสิ่งส าคัญที่ต้องเริ่มด้วยการระบุว่า กรอบทดสอบใดที่เหมาะสมกับภาษาที่ก าลังศึกษาอยู่ นักวิจัยจ านวนมากได้พัฒนากรอบทดสอบภาษา พูดจนประสบส าเร็จ อย่างไรก็ตาม มีการทดสอบดังกล่าวกับภาษามือเพียงครั้งเดียวเท่านั้น การศึกษาครั้งนี้เริ่มด้วยการกล่าวถึงความไม่เท่าเทียมกันนี้ โดยระบุถึงรูปแบบสถานการณ์ภายใต้การ ใช้ภาษามือไทย (ThSL) อย่างเป็นระบบโดยการใช้กรอบทดสอบ กรอบทดสอบเหล่านี้ได้ถูกน ามาใช้ เป็นครั้งแรกกับ 30 ประโยคในภาคกริยาและอนุประโยคของภาษามือไทย โดยผ่านการทดสอบหลาย ครั้งกับกลุ่มคนไทยหูหนวก นอกจากนี้ ได้มีการทดสอบเพิ่มเติมกับอีก 2 กลุ่มโดยใช้ประโยคภาคกริยา 12 ประโยค ผลปรากฏว่าภาษามือไทยบ่งบอกถึงลักษณะทางไวยากรณ์อย่างน้อย 6 รูปแบบ สถานการณ์ ได้แก่ individual-level states, stage-level states, activities, accomplishments, achievements, และ semelfactives การศึกษาครั้งนี้ยังแสดงให้เห็นว่า การบ่งชี้ของหน่วยค าที่เคยมีการน าเสนอเกี่ยวกับหน้าที่และ ระยะเวลา (the Event Visibility Hypothesis) ในภาษามือไม่สามารถใช้ได้กับข้อมูลที่พบในภาษา มือไทย และยังแสดงให้เห็นว่าฐานหน่วยค าเหล่านี้ไม่มั่นคง ผลของการศึกษากลับบ่งชี้ว่าลักษณะทาง เสียงของประโยคภาคกริยาสะท้อนให้เห็นถึงคุณลักษณะย่อยของเหตุการณ์หลายอย่างที่น าเสนอ โดยเฉพาะอย่างยิ่งในเหตุการณ์ที่มีความต่อเนื่อง เนื่องจากประกอบไปด้วยการแสดงหลายขั้น

iv

ยกตัวอย่างเช่น เดิน สามารถแบ่งเป็นขั้นตอนย่อย โดยแต่ละขั้นตอนแสดงถึงแต่ละระดับ การ เคลื่อนไหวของภาษามือไทยที่แสดงถึงการเดินสะท้อนให้เห็นถึงระดับขั้นเหล่านี้ ซึ่งตรงกันข้ามกับ เหตุการณ์ที่มีการเปลี่ยนแปลงทีละน้อย โดยจะแสดงระดับขั้นเป็นวงจรอย่างชัดเจน ยกตัวเล่น ละลาย ประกอบไปด้วยการเปลี่ยนถ่ายจากของแข็งเป็นของเหลว การเปลี่ยนแปลงนี้สะท้อนให้เป็น ผ่านการเคลื่อนไหวที่ค่อย ๆ เปลี่ยนแปลงครั้งเดียวของภาษามือไทยในค าว่า ละลาย ความเชื่อมโยง ระหว่างลักษณะภาคกริยาและโครงสร้างภายในนี้สนับสนุนการวิเคราะห์รูปสัญญะของ Kuhn (2017) เกี่ยวกับการเคลื่อนไหวเพื่อแสดงคุณลักษณะทางเสียงมากกว่า EVH

v

TABLE OF CONTENTS

Acknowledgements ...... i Abstract ...... ii บทคัดย่อ ...... iv List of Tables ...... ix List of Figures ...... x List of Abbreviations and Symbols...... xi Glossary ...... xii Chapter 1 Introduction ...... 1 1.1 Overview ...... 1 1.2 Research questions, reasons for the study and limitations ...... 1 1.3 Situation types ...... 2 1.3.1 Temporal features and basic situation types ...... 2 1.3.2 Viewpoint aspect and telic situations ...... 8 1.4 Thai Sign Language ...... 9 1.4.1 Previous studies in ThSL ...... 9 1.4.2 The visual-gestural modality ...... 10 1.4.3 Grammatical features and transcription conventions ...... 12 1.5 Situation types and sign languages ...... 17 1.6 Outline of the study ...... 20 Chapter 2 Methodology ...... 21 2.1 Overview ...... 21 2.2 Test selection ...... 21 2.2.1 Dynamism ...... 21 2.2.2 Duration ...... 25 2.2.3 Telicity ...... 28 2.2.4 Tests chosen for ThSL ...... 31 2.3 Predicate selection ...... 32 2.4 Data elicitation ...... 34 2.5 Language informants ...... 37 Chapter 3 Situation Types in Thai Sign Language ...... 38 3.1 Overview ...... 38

vi

3.2 Individual-level states ...... 38 3.2.1 Testing for dynamism – individual-level states ...... 38 3.2.2 Testing for duration – individual-level states ...... 39 3.2.3 Testing for state type with MANY.TIMES ...... 42 3.3 Stage-level states ...... 44 3.3.1 Testing for dynamism – stage-level states ...... 44 3.3.2 Testing for duration – stage-level states ...... 45 3.3.3 Testing for state type with MANY.TIMES ...... 47 3.4 Activities ...... 49 3.4.1 Testing for dynamism – activities ...... 49 3.4.2 Testing for duration – activities ...... 50 3.4.3 Testing for telicity – activities ...... 52 3.5 Accomplishments ...... 54 3.5.1 Testing for dynamism – accomplishments ...... 55 3.5.2 Testing for duration – accomplishments ...... 57 3.5.3 Testing for telicity – accomplishments ...... 58 3.6 Achievements ...... 62 3.6.1 Testing for dynamism – achievements ...... 63 3.6.2 Testing for duration – achievements ...... 65 3.6.3 Testing for telicity – achievements ...... 67 3.7 Semelfactives ...... 71 3.7.1 Testing for dynamism – semelfactives ...... 71 3.7.2 Testing for duration – semelfactives ...... 72 3.7.3 Testing for telicity – semelfactives ...... 74 3.8 Summary of results ...... 77 Chapter 4 Extended Testing ...... 82 4.1 Overview ...... 82 4.2 Individual-level states ...... 83 4.2.1 Testing featural values for individual-level states ...... 83 4.2.2 Comparison of results for individual-level states ...... 86 4.3 Stage-level states ...... 87 4.3.1 Testing featural values for stage-level states ...... 87 4.3.2 Comparison of results for stage-level states ...... 90 4.4 Activities ...... 91 4.4.1 Testing featural values for activities ...... 92 4.4.2 Comparison of results for activities ...... 95

vii

4.5 Accomplishments ...... 96 4.5.1 Testing featural values for accomplishments ...... 96 4.5.2 Comparison of results for accomplishments ...... 99 4.6 Achievements ...... 100 4.6.1 Testing featural values for achievements ...... 100 4.6.2 Comparison of results for achievements ...... 104 4.7 Semelfactives ...... 105 4.7.1 Testing featural values for semelfactives ...... 105 4.7.2 Comparison of results for semelfactives ...... 109 4.8 Summary of results ...... 110 Chapter 5 Conclusion ...... 111 5.1 Overview ...... 111 5.2 Summary of findings on ThSL situation types ...... 111 5.3 ThSL and the Event Visibility Hypothesis (EVH) ...... 113 5.4 Testing for situation types in other sign languages ...... 120 5.5 Directions for further research in ThSL ...... 125 5.5.1 Phonological movement and sub-event structure ...... 125 5.5.2 Additional topics for further research ...... 128 Bibliography ...... 129 Appendix A Predicates in Thai Sign Language ...... 134 Appendix B Informant Background Questionnaire ...... 143 Appendix C Informant Background Information ...... 144 Resume ...... 147

viii

LIST OF TABLES

Table 1 Featural values of situation types...... 4 Table 2 Tests chosen to identify the featural values of ThSL predicates ...... 31 Table 3 Predicates used to test for situation types in ThSL ...... 33 Table 4 Test results for individual-level states in ThSL ...... 43 Table 5 Test results for stage-level states in ThSL ...... 48 Table 6 Test results for activities in ThSL ...... 53 Table 7 Test results for accomplishments in ThSL ...... 61 Table 8 Test results for achievements in ThSL ...... 70 Table 9 Test results for semelfactives in ThSL ...... 76 Table 10 Compatibility of ThSL predicates with commands (Test 1)...... 77 Table 11 Compatibility of ThSL predicates with INTERRUPT (Test 2) ...... 78 Table 12 Compatibility of ThSL predicates with pace adverbials (Test 3) ...... 78 Table 13 Compatibility of ThSL predicates with temporal extent (Test 4) ...... 79 Table 14 Compatibility of ThSL predicates with STILL.THERE (Test 5) ...... 80 Table 15 Compatibility of ThSL predicates with MANY.TIMES (Test 6) ...... 80 Table 16 Predicates used in extended testing of situation types in ThSL ...... 82 Table 17 Comparison of results for individual-level states in ThSL ...... 86 Table 18 Comparison of results for stage-level states in ThSL ...... 90 Table 19 Reduplication of stage-level states in multi-event clauses ...... 91 Table 20 Comparison of results for activities in ThSL ...... 95 Table 21 Comparison of results for accomplishments in ThSL ...... 99 Table 22 Comparison of results for achievements in ThSL ...... 104 Table 23 Comparison of results for semelfactives in ThSL ...... 109 Table 24 Binary approach to identifying situation types ...... 124

ix

LIST OF FIGURES

Figure 1 The manual parameters of a sign ...... 11 Figure 2 ThSL sign BEND.BAR (iconic sign) ...... 11 Figure 3 ThSL sign ANGRY (arbitrary sign) ...... 12 Figure 4 Nominalization by reduplication (Supalla & Newport, 1978, p. 102) ...... 14 Figure 5 Verb agreement in ThSL ...... 15 Figure 6 Verb agreement with a plural object in ThSL ...... 15 Figure 7 Object incorporation in ThSL predicate ...... 16 Figure 8 Manifestations of the EndState morpheme (Wilbur, 2008, p. 232) ...... 17 Figure 9 Manifestations of the Extent morpheme (Wilbur, 2008, p. 235) ...... 18 Figure 10 ASL sign DIE (Kuhn, 2017, p. 12) ...... 19 Figure 11 Manipulation of the ASL sign DIE (Kuhn, 2017, p. 12) ...... 19 Figure 12 ThSL signs CAN (Frame 1), PLEASE (Frame 2), and GO (Frame 3) ...... 23 Figure 13 ThSL sign INTERRUPT ...... 24 Figure 14 ThSL signs VERY.SLOW (left) and SLOW (right) ...... 26 Figure 15 ThSL signs DAY-THREE (left) and MINUTE-FIVE (right) ...... 27 Figure 16 Variations of the ThSL sign STILL.THERE ...... 29 Figure 17 ThSL sign MANY.TIMES ...... 30 Figure 18 ASL sign POSTPONE (left) and ThSL sign SHOUT (right) ...... 114 Figure 19 ASL sign HIT (left) and ThSL sign PUNCH.PERSON (right) ...... 115 Figure 20 Completed form of ThSL sign BEND.BAR ...... 116 Figure 21 Uncompleted form of ThSL sign BEND.BAR ...... 117 Figure 22 ASL signs SICK (left) and BECOME.SICK (right) ...... 118 Figure 23 ASL sign LOOK.AT (Wilbur, 2008, p. 244) ...... 119 Figure 24 ThSL sign ERASE ...... 120 Figure 25 Hierarchical approach to identifying situation types ...... 123 Figure 26 ThSL sign RIDE.BICYCLE ...... 126 Figure 27 ThSL sign DRIVE.MOTORCYCLE ...... 126

x

LIST OF ABBREVIATIONS AND SYMBOLS

* Unacceptable construction ? Marginally acceptable construction ASL EVH Event Visibility Hypothesis HKSL Sign Language JSL

IPFV Imperfective aspect Nod Head nod articulated with the sign (transcribed as a subscript after the verb - VERBnod)

PFV Perfective aspect ThSL Thai Sign Language

xi

GLOSSARY

Accomplishment A situation that is [+dynamic], [+durative], and [+telic]. Achievement A situation that is [+dynamic], [-durative], and [+telic]. Activity A situation that is [+dynamic], [+durative], and [-telic]. Arbitrary A signal (e.g., a word or sign) that does not have a clear form-meaning relationship. Atelic A situation that does not have an inherent endpoint involving a result state. Bounded A situation involving a final endpoint. The endpoint may be temporal, spatial, or scalar. A perfective portrayal of an event also includes a final endpoint, making the event bounded. Completed A telic event that has reached a result state. Contralateral A sign that is articulated on the opposite side of the body. Covert The inclusion of semantic components (e.g., telicity) as part of lexicalization the inherent meaning of a word. Deaf/deaf The term ‘Deaf’ refers to a linguistically and culturally distinct group; ‘deaf’ refers to a physical state. Dispositional A predicate indicating an ability or preference. predicate Durative A situation that unfolds or exists through time. Dynamic A situation that involves motion and energy. EndState A morpheme proposed in the EVH to mark [+telic] situations. morpheme Event A situation that has a [+dynamic] value (i.e., activities, accomplishments, achievements, and semelfactives). Extent morpheme A morpheme proposed in the EVH to mark [+durative] situations. External override When the featural values of a constituent override those of the predicate, thus altering the overall situation type of a clause. Featural values The temporal properties ([+/-dynamic], [+/-durative], [+/-telic]) that determine the situation type of a predicate. Grammatical See viewpoint aspect. aspect

xii

Iconic A signal (e.g., a word or sign) that has a clear form-meaning relationship. Imperfective A grammatically marked portrayal of a situation from within, without reference to initial or final endpoints. Inceptive Entry into an event. Individual-level A stative situation that typically lasts over the lifetime of the state participant (e.g., be tall, be humble, be Hmong). Instantaneous A situation that occurs in an idealized moment of time (i.e., a situation with a [-durative] value). Ipsilateral A sign that is articulated on the same side of the body. Manual The part of a sign articulated by the signer’s hand(s). component Non-manual Any part of a sign that is not articulated by the signer’s hands component (e.g., articulated by the head, face, or body of the signer). Orientation The direction of the palm during the articulation of a sign (e.g., right, left, up, down). Participant Any entity involved in an event. Perfective A grammatically marked portrayal of a situation as a whole, including initial and final endpoints. Punctual See instantaneous. Radial The thumb-side of the hand. Representation The English word(s) used to transcribe the meaning of a sign. In sign language literature, this is often called the gloss. Semelfactive A situation that is [+dynamic], [-durative], and [-telic]. Sign language A language expressed using a visual-gestural modality. Situation A state or event. Situation types Situation types refer to categories of states and events that are determined by the featural values [+/-dynamic], [+/-durative], [+/-telic]. Stage-level state A temporary state that may change over the course of the participant’s life (e.g., hungry, tired, glad). State A situation that is [-dynamic], lacking motion or energy. Telic An event that has an inherent endpoint involving a result state. Ulnar The pinkie-side of the hand. Unbounded A situation that lacks a final endpoint. Compare with bounded.

xiii

Uncompleted A telic event that has not reached a result state. Unrealized When entry into an event is not actualized/does not occur. inceptive Viewpoint aspect The grammatically marked portrayal of an event either from without (perfective) or from within (imperfective).

xiv

Chapter 1 Introduction

1.1 Overview The dynamic, durative, and telic values of a predicate determine its situation type. Each of these features represents a covert lexical category that can be uncovered using test frames. Multiple studies have used such tests to investigate the featural values of situation types in spoken languages. However, relatively little work has been done in sign languages. The current study helps address this disparity by providing a systematic investigation of situation types in Thai Sign Language (ThSL) using test frames.

Section 1.2 discusses the reasons for the study, specific research questions, and the limitations of the study. Section 1.3 then lays the foundation for the study by an overview of situation types and their featural values. Section 1.4 introduces Thai Sign Language and the visual-gestural modality. Section 1.5 discusses earlier studies and theories related to situation types in sign languages. Section 1.6 gives an overview of the following chapters.

1.2 Research questions, reasons for the study and limitations The current study seeks to answer three questions concerning situation types in Thai Sign Language. These are the following:

1. Which tests effectively identify the featural values of ThSL predicates? 2. Which situation types does Thai Sign Language covertly lexicalize? 3. How consistently are the situation types of predicates classified according to these tests by different groups of signers?

To answer the first question, various tests were attempted with a small set of ThSL predicates to determine which ones effectively identified featural values in ThSL (see Chapter 2). The tests were then used to answer the second question by identifying the situation types of 30 ThSL predicates (see Chapter 3). The final question was investigated by testing a subset of predicates with two additional groups of ThSL signers (see Chapter 4).

1

The investigation of these questions produced several useful findings. First, the identification of practical tests for ThSL benefits those undertaking similar studies in other sign languages. Second, the identification of the situation types of specific predicates in ThSL adds to our knowledge of how situation types are encoded cross- linguistically and cross-modally. Using the tests with multiple groups of signers determines their extensibility and provides insight into the consistency of a predicate’s classification. The results also have repercussions for previously proposed morphological markings of telicity and duration (see Sections 1.5 and 5.3). Finally, the findings of the study prepare the way for further research in ThSL, especially the interaction between situation types, viewpoint aspect, and tense.

Several parameters limit this study. First, the study treats ThSL as a single variety and does not intentionally investigate variation among the signing of Deaf from different regions or schools. Second, it does not look at the role (if any) that situation types play in determining default viewpoint aspects, nor the of a situation on a timeline (i.e., tense). The interaction between viewpoint aspect and situation types, or any overt marking of viewpoint aspect, is beyond the scope of this study. Finally, the study focuses on the situation aspect of predicates rather than the situation aspect of larger syntactic units (see Xiao & McEnery, 2004, Chapter 3).

1.3 Situation types The dynamic, durative, and telic values of a predicate determine its situation type (Smith, 1997). Section 1.3.1 illustrates these features and the basic situation types. Section 1.3.2 discusses the relationship between telic situation types and viewpoint aspect (portraying an event either from within, without reference to initial or final endpoints, or from without as a single unit).

1.3.1 Temporal features and basic situation types The lexical meaning of a predicate includes temporal features. This factor has led some researchers to use the term lexical aspect to refer to situation types. Researchers have also used the term aktionsart, but with some ambiguity (see discussion in Comrie, 1976, pp. 6–7). Although predicates possess these features inherently, other components of the clause can override them and shift the overall situation type of the clause. This phenomenon leads Smith (1997, p. 2) to focus on the situation type of the clause (‘constellation’ in her terms) as a whole. However, the inherent temporal features of the predicate can be uncovered by testing it in “maximally

2

simple sentences” (Smith, 1997, p. 54). These maximally simple sentences are either intransitive or, if transitive, have a quantized direct object (i.e., a count noun). To identify the featural values of ThSL predicates, sentences that fall within these parameters have been used.

The first temporal feature of a predicate is its dynamic value. States are [-dynamic], meaning they do not require an ongoing input of energy to be maintained. Examples include be sick, own a car, and live in Chicago. These states exist in undifferentiated form through time. Thus, a state such as the baby is happy is true at any moment during which the state holds.

States consist of permanent, individual-level states such as be Hmong or be tall and transitory, stage-level states such as be sick or be happy. Due to their enduring nature, individual-level states generally sound odd with adverbs of temporal extent (e.g., one week, all morning). Stage-level states are compatible with such expressions. The distinction explains why a clause like the teacher was Hmong all morning is semantically ill-formed, but the teacher was happy all morning is acceptable. Such co-occurrence patterns can be used to distinguish stage-level from individual-level states.

In contrast to states, all events are [+dynamic] and require an ongoing input of energy, either internal or external, to be maintained. Typical events include walk, bend, and explode. These dynamic situations have a fundamental premise of motion and often involve agency, activity, and change (Smith, 1997, pp. 28–29). Unlike states, events are comprised of stages. For example, swim can be broken down into individual strokes with each stroke representing a stage of the event. Alternatively, the event may consist of a single stage, as in explode, where the object transitions from intact to in pieces.

Second, a situation may be either [+durative] or [-durative]. All states are [+durative]. Events that are [+durative] contain successive stages that unfold over a period of time. Situations such as walk, bend, and build a boat are, thus, durative. The stages of these events may be homogenous (repetitions of identical cycles), such as each step in walk, or heterogeneous, such as each stage in build a boat. In contrast, [-durative] events are instantaneous and include events such as die and pop. These events consist of a single stage. Although the instantaneous nature of the situations is idealized (Smith, 1997, p. 29), a distinction between durative and instantaneous events is still reflected in many languages (Comrie, 1976, p. 43; Smith, 1997, p. 30).

3

Third, situations may be [+telic] or [-telic]. Events that are [+telic] have inherent endpoints and include situations like melt, straighten, and die. When these events arrive at their endpoint, a result state ensues. For example, the result state of die is be dead, and the result state of meet a friend is be with a friend. Because a result state is present, the event cannot repeat unless the state is reversed or a new participant is selected to enter the state. For instance, the clause I met my friend yesterday and again today implies that the participants parted company for a time (reversed the result state) and then met again (repeated the event). Alternatively, a new participant can be involved. Thus, I popped the balloon and then popped it again is odd, but I popped the red balloon, and then the blue balloon is acceptable.

In contrast, events that are [-telic] do not have a result state. Although the event can terminate (i.e., temporally bounded), the point of cessation is arbitrary (Smith, 1997, p. 29). Atelic events include walk, paint, and knock. Instantaneous atelic events, such as knock, appear to be telic since they have a predictable terminus. However, the end of the event is due to its instantaneous nature, not the participant’s entry into a result state. The repeatability of these events is evidence of their atelic nature. For example, the clause the girl knocked and knocked and knocked is acceptable while the window broke and broke and broke is not.

These three binary features group predicates into five situation types. Table 1, which closely matches the table presented by Smith (1997, p. 20), lists these types.

Table 1 Featural values of situation types

Dynamic Durative Telic English examples

State - + n/a be sick, be tall, be in limbo

Activity + + - walk, meditate, sing, work,

Accomplishment + + + bend, straighten, dry, untangle

Achievement + - + explode, vanish, die, find, pop

Semelfactive + - - knock, clap, flash, blink, click

In Table 1, the first column gives the name of the situation type. The second, third, and fourth columns indicate the dynamic, durative, and telic values of each situation type. The final column illustrates the situation types with English predicates.

In addition to the situation types in Table 1, Smith (1997) and Croft (2012) review several additional types. In some instances, these are non-central examples such as

4

wait or stand, which have both stative and dynamic properties. Others have additional semantic features that divide the situation types into multiple sub-types. These are discussed further in Chapter 3 in relation to the ThSL data.

Xiao and McEnery (2004) propose two additional features of situation types, [+/-bounded] and [+/-result]. These proposed additions appear to stem from several factors. First, the authors treat the final endpoint of semelfactives as an additional feature [+bounded]) (Xiao & McEnery, 2004, p. 49). However, the boundedness of semelfactives is due to their being [-durative]. Thus, an additional feature is not needed to account for their final endpoint. Next, the authors treat incremental theme verbs as [+telic] (Xiao & McEnery, 2004, pp. 48, 56). These verbs generally involve creation (e.g., draw a picture), destruction (e.g., demolish a building) or consumption (e.g., eat an apple). However, the telicity of clauses with incremental theme verbs is a product of the arguments, not the predicates (see Rappaport Hovav, 2006). Thus, no additional features are needed to account for their behavior. Finally, the authors do not consider scale types (see Hay, Kennedy, & Levin, 1999; C. Kennedy & McNally, 2005) in their discussion of telicity. This omission leads them to unnecessarily tease apart the telic nature of achievements and accomplishments. In contrast, the current study treats incremental theme verbs as [-telic] and understands telicity as a product of scale type and pragmatic implication (see Section 3.5 for discussion). This analysis makes additional features unnecessary. Hence, only the three featural values discussed by Smith (1997) are used in this study.

As noted, the inherent featural values of a predicate can be uncovered using test frames. The test frames only occur felicitously with predicates that have a specific binary value (e.g., [+dynamic]). Predicates with the opposite value produce ill- formed clauses. For example, the phrase what happened can test the dynamism of English predicates. Dynamic predicates can be used to answer the question; stative predicates cannot. The examples in (1) and (2) illustrate the test.

(1) What happened? The boat sank.

(2) What happened? *Maggie is tall.

5

In (1), the answer the boat sank is felicitous with the question what happened? This felicity indicates that sank is [+dynamic]. In contrast, as (2) shows, Maggie is tall cannot be used to answer the question, indicating that is tall is [-dynamic].

Test frames can also identify the durativity of predicates. For example, an adverbial of temporal extent, such as all morning, is generally compatible with durative events but not with instantaneous ones. The clauses in (3) and (4) demonstrate the test.

(3) Rosie played all morning.

(4) *The car exploded all morning.

In (3), the predicate play is compatible with all morning, indicating it is [+durative]. Conversely, exploded in (4) is incompatible with the adverbial, indicating it is [-durative].

Finally, co-occurrence with temporal extent adverbials can determine the telicity of a situation. For instance, temporal extent adverbials using in such as in two minutes are compatible with telic events and generally unacceptable with atelic events. The clauses in (5) and (6) illustrate this tendency.

(5) The ice melted in twenty minutes.

(6) ?Kimber walked in twenty minutes.

In (5), the adverbial in twenty minutes occurs felicitously with melted, indicating it is [+telic]. In contrast, the same adverbial is incompatible with walked in (6), indicating it is [-telic].

Using a series of tests like those just demonstrated allows the featural values – and hence the situation type – of a predicate to be identified. However, other constituents of a clause can override the featural values of the predicate and cause a shift in the situation type. The clauses in (7) and (8) provide examples of shifted situation types.

(7) a. Abby ran. b. Abby ran to the park.

(8) a. Lucy knocked. b. Lucy knocked for five minutes.

6

In (7a), ran is an activity (dynamic, durative, atelic). Although Abby may stop running, the point at which she stops is arbitrary. However, the phrase to the park in the second line (7b) adds an endpoint to the situation. This addition changes the telic value of the clause to that of an accomplishment (dynamic, durative, telic). Likewise, although knock in (8a) is a semelfactive (dynamic, instantaneous, atelic), adding for five minutes in (8b) creates an iterative interpretation of the event. The iterative interpretation coerces the situation type of the clause into an activity (dynamic, durative, atelic).

The use of test frames can also create coerced interpretations of situation types. For example, although instantaneous events should be incompatible with temporal extent adverbials, the adverbial can create a coerced (but still acceptable) interpretation of the event. Since these clauses are acceptable, they are not marked in the examples. However, the coercion is evident from the free translation. Examples showing acceptable, uncoerced clauses are also unmarked. A question mark (‘?’) precedes questionable clauses. An asterisk (‘*’) precedes unacceptable constructions. The examples in (9) illustrate the annotation system.

(9) a. Tommie is hungry. b. *Tommie, be hungry! c. ?The jewel shattered again and again (but magically restored itself each time). d. Tommie coughed for five minutes.

In (9a), the clause Tommy is hungry is unmarked, indicating it is acceptable. The command in (9b) Tommy, be hungry! is infelicitous and so marked by an asterisk. The clause in (9c) the jewel shattered again and again is generally infelicitous, although, with enough context, it is acceptable. So, it is marked with a question mark. Finally, although the clause in (9d) is acceptable, coughed no longer receives an instantaneous interpretation (a single cough), but is coerced into a durative event involving a series of coughs. Clauses involving coerced situation types are discussed in prose following the examples.

When the featural value of a predicate conflicts with the value of another clause constituent, Smith (1997, p. 53) posits the “principle of external override.” This principle states that the value of the additional constituent overrides that of the predicate. Both (7b) and (8b) illustrate this effect. First, in (7b), the [-telic] value of ran was overridden by the [+telic] value of to the park. Then, in (8b), the [-durative] value of knocked was overridden by the [+durative] value of for five minutes. The same principle is seen in ThSL clauses in Chapter 3 and Chapter 4.

7

In addition to the inherent temporal features of a situation, the speaker or signer can choose to portray the event from within using viewpoint aspect. Such a portrayal does not include the initial and final endpoints of the event. Although this thesis does not explore viewpoint aspect, understanding the interaction between situation types and viewpoint aspects is germane to the study. Section 1.3.2 provides a relevant introduction.

1.3.2 Viewpoint aspect and telic situations Viewpoint aspects are indicated grammatically. The basic viewpoints are perfective and imperfective. First, a perfective viewpoint presents events as a whole, including the initial and final endpoints. Because the portrayal includes these endpoints, it presents the event as bounded. Imperfective viewpoints present the event from within, without reference to the initial and final endpoints. Languages such as Russian mark perfective and imperfective aspects morphologically. Examples (10) and (11), taken from Frawley (1992, p. 296), illustrate.

(10) ja napis al pismo I write-PFV past letter ‘I have written the letter.’

(11) ja pis al pismo I write-IPFV past letter ‘I was writing the letter.’

In both (10) and (11), the clauses are accomplishments. However, in (10), the form of the predicate napis ‘wrote’ is perfective, including the initial and final endpoints of the event. In contrast, in (11), the imperfective form pis ‘writing’ is used. The form portrays the event from within, without reference to either endpoint.

There is a salient interaction between the perfective aspect and telic events. Specifically, a perfective viewpoint includes the final endpoint of an event. For a telic event, the final endpoint involves a change of state. Thus, a perfective portrayal of a telic event indicates that a change of state has occurred. Once that change occurs, the event cannot be interrupted and cannot continue. The opposite is true for an imperfective portrayal of a telic event. This portrayal does not include the final endpoint. Thus, the event may be interrupted or continued (i.e., the process leading up to the change of state may be interrupted or continue).

8

The aspectual portrayal of an event, thus, has semantic implications for telic events. Modifying the form of ThSL accomplishment predicates can convey these same semantic implications. If the full (completed) movement of the predicate sign occurs, it indicates the final endpoint of the event has been reached. If a shortened (uncompleted) form is used, it indicates the final endpoint has not occurred (see discussion of Figure 10 and Figure 11).

Whether this modified movement represents dedicated, aspectual morphology or is an iconic modification of the sign (see Kuhn, 2017) is beyond the scope of this study. However, the semantic implications of the two forms are pertinent for the testing of ThSL predicates (see Sections 2.23.54.5). In order to discuss these semantic implications without making grammatical claims, the current study uses the term ‘completed’ to refer to accomplishments that have reached their final endpoint. ThSL expresses these by using the full movement of the predicate sign. The study will use the term ‘uncompleted’ to refer to accomplishments that have not reached this endpoint. These are expressed using an uncompleted form of the predicate sign. As noted, the different forms of the signs and their semantic implications interact with two of the tests used for the identifying situation types of ThSL predicates in the study. Section 1.4 presents additional information about ThSL.

1.4 Thai Sign Language Previous research on Thai Sign Language, the focus of this study, is reviewed in Section 1.4.1. The phonological parameters of a sign are introduced in Section 1.4.2. Relevant grammatical features and transcription conventions for ThSL are described in Section 1.4.3.

1.4.1 Previous studies in ThSL Thai Sign Language is the national language of the Deaf in Thailand (Ministry of Education, 1999). The population of profoundly deaf individuals in Thailand in 1997 was estimated at 51,000 (Fennig & Simons, 2018). However, a 2017 survey reported that the population to be approximately 300,000 (“NADT,” 2017).

Woodward (1996) conducted a lexicostatistical comparison between ThSL and several other sign languages. The comparison, based on a 90-word list, indicated a 52% lexical similarity between ThSL and American Sign Language (ASL). The similarity is due to the introduction of ASL into the Thai educational system in the

9

early 1950s (Woodward, 2003, p. 290). Several sign languages in Vietnam have, in turn, been influenced by (LSF), which belongs to the same as ASL. Because of this relatedness, Thai Sign Language also shows lexical similarity to three sign languages in Vietnam: Ha Noi Sign Language, Hai Phong Sign Language, and Ho Chi Minh Sign Language (Woodward, 2003, p. 291). However, this lexical similarity reflects a superficial, rather than genetic, relatedness between the languages.

Over the last several decades, several studies have been published on linguistic features of ThSL. These include studies on spatial predicates (Collins-Ahlgren, 1990; Eberle, 2013; Wallace, 2019), word-order (Lumtien, 1997), negation (Pradapwattanangune, 1998), yes-no questions (Wudthayagorn, 1998), the formation of entity-denoting signs (Kullavanijaya & Thepkanjana, 2001), and ThSL syllable and word structure (Tumtavitikul & Niwatapant, 2008). Additional studies have also contrasted ThSL and Thai (Niwatapant, 2006; Niwatapant & Tumtavitikul, 2005; Tumtavitikul, Niwatapant, & Dill, 2009). Finally, Danthanavanich’s (2008) dissertation on ThSL grammar covers a wide variety of topics. These include a sociolinguistic introduction to sign languages in general and to ThSL specifically. Thai Sign Language clause structure, word formation, use of signing space, the role of non-manual features, and the relevance of gesture are also touched on. The first study of situation types in ThSL was Wallace (2020). The tests developed by Wallace are applied in the current study to an extended set of predicates (see Chapter 3) and then tested with a wider audience (see Chapter 4).

1.4.2 The visual-gestural modality1 The hands, face, and body of a signer are used to articulate signs. The hands are further identified as the dominant hand (the right hand for right-handed signers) and the non-dominant hand (the left hand for right-handed signers). The dominance of the hands is reversed for a left-handed signer.

1 The discussion here has been kept as simple as possible; only the features most relevant to the topic of situation types in ThSL are covered, and even these are covered in a simplified fashion. It is hoped that readers unfamiliar with sign languages are able to grasp the pertinent information, and those with a background in sign languages will already be aware of the finer details (e.g., dual, trial and plural agreement marking; backward verbs) and differing views (e.g., analysis of pronominal reference, person agreement and deixis), which do not directly pertain to the present study.

10

The signer’s hands constitute the manual component of a sign. The parameters of the manual component are (e.g., fingers spread or held together), location (e.g., the space in front of the signer or on the signer’s body), orientation (e.g., palms facing up or down) and movement (e.g., up, down, arching). The sign in Figure 1 illustrates these parameters.

Figure 1 The manual parameters of a sign

Figure 1 shows the sign PLANT in ThSL. The signer’s hands have a “flat O” handshape. The palms face downward (orientation), and the sign is articulated in front of the signer at chest-height (location). The signer’s dominant (right) hand moves forward in small arcs (movement), reminiscent of planting seed after seed.

Most of the non-manual components in Figure 1 are in neutral positions. These components include the aperture of the eyes (e.g., wide, narrow), eyebrows (e.g., raised or lowered), mouth (e.g., puckered or flat), cheeks (e.g., puffed or pulled-in), the position and movement of the head (e.g., tilted or nodding), and the signer’s posture (e.g., leaning forward or to the side).

Individual signs vary in their degree of iconicity. A highly iconic sign has a form that closely reflects its meaning. Figure 2 illustrates such a sign, BEND.BAR.

Figure 2 ThSL sign BEND.BAR (iconic sign)

Figure 2 shows the highly iconic sign BEND.BAR. The signer holds two closed fists (as if grasping a bar) at shoulder-height. The dominant hand then moves downward

11

with a movement that originates at the shoulder. The handshape and movement used in the sign mimic those used to bend a bar, presenting a clear form-meaning relationship.

In contrast, the sign ANGRY in ThSL is arbitrary (i.e., the form of the sign does not clearly relate to its meaning). Figure 3 shows this sign.

Figure 3 ThSL sign ANGRY (arbitrary sign)

As seen in Figure 3, for the sign ANGRY the signer places the fingertips of a claw- shaped hand above one eye. The hand is then drawn back sharply. The relationship between the form of the sign and its meaning is opaque.

Iconic predicates like BEND.BAR are often called depicting verbs. Although some studies of situation types in sign languages omit these verbs (Wilbur, 2008), the present study include them. Including them provides a more comprehensive study of ThSL predicates and allows the behavior of depicting and non-depicting predicates to be compared.

Several standard conventions are adopted here to transcribe ThSL with simple English text. Section 1.4.3 describes these conventions.

1.4.3 Grammatical features and transcription conventions The complex visual-gestural expression of sign languages makes transcribing them two-dimensionally challenging. Since the current study focuses on semantic features rather than phonetic forms, however, a simplified transcription using English words to represent signs is used. Rather than capturing the phonetic form of the signs, this solution merely captures the meaning and relevant grammatical information encoded in each sign, as (12) illustrates.

(12) MOTHER CHILD LOVE ‘The mother loves the child.’

12

The example in (12) illustrates the clause ‘the mother loves the child’. It is composed of three lexical signs – MOTHER, CHILD, and LOVE – as captured by the English representations. The representations themselves do not capture the non-manual components that identify the subject and object of the clause. Instead, they are made evident by the free translation.

Clause transcription becomes complicated when pronominal arguments, person agreement, reduplication, and classifier-incorporation are involved. These phenomena are discussed below, followed by a description of how they are transcribed in the study.

First, pronominal reference can be indicated by pointing with the index finger in various directions – towards the signer for first person, towards the addressee for second person, and away from the signer (but not towards the addressee) for third person. Pronominal uses of these signs in ThSL are represented by INDEX followed by the person number (e.g., INDEX-1 indicates first person). For free translations, third- person pronominal uses of INDEX are glossed as ‘he’ or ‘she’ for simplicity’s sake, even though gender is not encoded.

Often, signers omit pronominal and nominal references altogether. This omission commonly occurs in the ThSL data since the examples were taken directly from discussions. Specifically, once a signer had identified a referent, it was frequently omitted afterward. For these examples, the implied referent occurs in parentheses in the free translation.

The sign INDEX can also be used as a locative meaning ‘here’ or ‘there’. When it is used this way, the sign is represented by INDEX, and the free translation uses ‘here’ and ‘there’ as appropriate.

Reduplication of a sign can serve multiple functions. First, reduplication of a predicate can express pluractionality (i.e., multiple instances of an event). For example, the predicate SICK in ThSL is often reduplicated when used in expressions such as ‘I am sick often’, as (13) and (14) illustrate.

(13) ?GRANDMA SICK MANY.TIMES ‘Grandma has been sick many times.’

(14) GRANDMA SICK SICK MANY.TIMES ‘Grandma has been sick many times.’

13

In (13), the predicate SICK is signed only once with the adjunct MANY.TIMES. This construction was often unacceptable to ThSL informants. In (14), the sign SICK is repeated, making it compatible with the implication of repeated occurrence and creating a well-formed clause. As in (14), repetition of the English representation (e.g., SICK SICK) indicates a repetition of the predicates.

Another function of reduplication is to change a verb into a noun. Figure 4 provides a prototypical example of this nominalizing function.

Figure 4 Nominalization by reduplication (Supalla & Newport, 1978, p. 102)

In Figure 4, the first picture shows the ASL verb SIT. The signer extends the middle and index fingers of his non-dominant (left) with his palm facing down. He also extends the middle and index finger of his right hand, bending them slightly. The signer brings his right hand down so that the extended fingers of the right hand rest on the extended fingers of the left hand.

The second frame of Figure 4 shows the ASL noun CHAIR. The shape of the hands and the location of the sign are the same as in the sign SIT. However, the articulation of the sign is repeated twice, and a tenser movement is used. Supalla and Newport (1978, pp. 101–102) note that this tense, repeated movement consistently distinguishes a class of nouns in ASL from related verbs. Nominalization through reduplication was also observed for ThSL by (Danthanavanich, 2008, pp. 114–115). In this study, nominalized signs are represented in English by the appropriate noun (e.g., TIN.CAN).

As seen earlier, signing space (the area in front of the signer) can be used to identify a pronominal referent. Signing space can also be used to encode person agreement. Verbs that use space in this way are called agreement verbs (see Padden, 1988). With most agreement verbs, the sign starts at a location in signing space assigned to the subject argument and moves towards the area representing the object. Figure 5 illustrates this use of space.

14

Figure 5 Verb agreement in ThSL

In Figure 5, the sign 1-BLOCK-3 is formed by holding the non-dominant (right) hand in an O-handshape in front and to the left of the signer’s body. The ulnar (pinkie) side of the dominant hand, which is open with the fingers held together, is brought forcefully against the radial (thumb) side of the non-dominant hand. The movement starts near the signer’s body, encoding first-person agreement with the subject. The movement away from the signer (but not towards the addressee) encodes third- person agreement with the object. The convention x-VERB-y is used to show agreement with singular objects: the number to the left of the verb indicates subject agreement, the number to the right indicates object agreement.

The verb agreement in Figure 5 is singular. However, the movement can be repeated in multiple locations in front of the signer to indicate agreement with a plural object (e.g., ‘I blocked many people’). Figure 6 illustrates this form.

Figure 6 Verb agreement with a plural object in ThSL

Figure 6 shows the stem 1-BLOCK-3 repeated multiple times, each time in a slightly different area of signing space. The movement encodes agreement with a plural object. Verbs that are repeated in multiple locations are transcribed by adding a different subscript after each repetition of the predicate (e.g., 1-BLOCK-3x 1-BLOCK-3y

1-BLOCK-3z).

15

In addition to agreeing verbs, sign languages also have plain verbs and spatial verbs. First, plain verbs include cognitive (e.g., THINK, DREAM, BELIEVE) or stative (e.g., HAPPY, SICK, TIRED) predicates. Most researchers do not analyze these predicates as marking agreement (but see Meir, Padden, Aronoff, & Sandler, 2007). They are, thus, transcribed in the current study without person agreement marking (e.g., THINK, HAPPY). Second, spatial verbs use signing space to indicate the movement or location of a participant in real space. For example, moving the person classifier from the left (point A) to the right (point B) indicates that a person moved from point A to point B. However, the addition of locative information can coerce the telic value of a clause. Thus, the study of ThSL did not include predicates from this class.

Finally, sign languages can incorporate a classifier for the object in a predicate sign. Figure 7 illustrates this feature with the sign ICE.MELT.

Figure 7 Object incorporation in ThSL predicate

Figure 7 shows the phrase ‘the ice melted’. The first frame is a still shot of the sign ICE. In the second frame, the signer bends the fingers of his right hand as if holding a small, three-dimensional object. This handshape is the classifier for an ice cube. In the last two frames, the fingers relax and spread open, allowing the right palm to come into contact with the left palm and slide forward, imitating the melting of a three-dimensional ice cube. If the ice referred to were a sheet of ice, instead of a cube, the sign would be modified to incorporate a classifier for a flat, two- dimensional object instead.

Although the predicate does not explicitly encode the actual object (e.g., ‘ice’) represented by the classifier, it is still identified in the English representation (e.g., ICE.MELT instead of THREE.DIMENSIONAL.OBJECT.MELT) for this study. This convention simplifies the transcription, makes the referent transparent, and avoids creating a transcription system for different classifiers.

16

Several earlier studies have investigated the expression of situation types in sign languages. These are reviewed in Section 1.5.

1.5 Situation types and sign languages Rathmann (2005) presents the only prior study of situation types in a sign language that utilizes test frames. In this study, he uses these tests to demonstrate the dynamic, durative, and telic value of different situations types in ASL.2 By demonstrating the inherent presence of these values, Rathmann shows that ASL distinguishes states, activities, accomplishments, achievements, and semelfactives.

Other studies of sign languages have omitted test frames and instead looked at situation types through the lens of Wilbur’s (2008) Event Visibility Hypothesis (EVH). According to the EVH, morphological marking indicates the durative and telic values of sign language predicates. These markings are visible in the articulation of the sign due to the visual-gestural nature of sign languages. The EVH proposes that the ‘EndState’ morpheme marks telicity. This morpheme consists of a rapid deceleration to a stop that manifests as a change in handshape, orientation, or location. Figure 8 illustrates these surface forms.

Figure 8 Manifestations of the EndState morpheme (Wilbur, 2008, p. 232)

Figure 8 shows four predicates in ASL that purportedly include the EndState morpheme. The first, SEND, involves a change of handshape (from closed to open). The second, HAPPEN, involves a change of orientation (from palms facing up to facing down). The third and fourth signs involve a change in location. In POSTPONE, only the dominant hand moves forward in space. The sign HIT includes the forward movement of the dominant hand plus final contact with the non-dominant hand. The movement used for each of the signs in Figure 8 is a rapid deceleration to a stop – the EndState morpheme.

2 Following Smith (1997, p. 27), Rathmann (2005) looks at the featural values of the predicate and its arguments (the constellation) rather than the situation type of the predicate alone.

17

The EVH also proposes the Extent morpheme as an overt marker of duration. The Extent morpheme is manifested in the articulation of the predicate by path movement. Figure 9 presents several potential surface forms of the morpheme.

Figure 9 Manifestations of the Extent morpheme (Wilbur, 2008, p. 235)

In Figure 9, three ASL predicates are used to demonstrate the Extent morpheme. In the first sign, RUN, the index fingers are bent and straightened repeatedly at the second knuckle as the hands move forward in space. For the sign PLAY, the signer extends the pinkie and thumb of each hand, pinkies pointing downward, and then twists his wrists up and down rapidly. In the final sign, READ, the signer extends the middle and index finger of his dominant hand while the other fingers are closed. The non-dominant hand is held under the dominant hand with the palm facing up, representing the object being read. The signer then repeatedly bends the wrist of his dominant hand backward and forward, like two eyes moving up and down a page. According to Wilbur, each of these signs involves a path movement that morphologically encodes temporal duration. Signs that lack this type of movement are, by default, classified as instantaneous. When both the Extent morpheme and EndState morpheme are present in the articulation of a sign, the event is both durative and telic.

The EVH’s claim that telicity and duration are morphologically marked is unusual since these values, cross-linguistically, tend to be lexical properties of the predicate. In contrast, viewpoint aspect is typically encoded morphologically. Thus, the potential morphological marking of a value should be tested to determine whether it encodes viewpoint aspect rather than durative or telic values. So far, such tests have not been conducted to verify the claims of the EVH.

Furthermore, the omission of objective tests, as pointed out by Davidson et al. (2018), leads to a pattern of circular reasoning. For example, the presence of the Extent morphemes determines the durative value of the predicate. The durative value of the predicate then establishes the presence of the Extent morpheme. Because of these issues, the analysis of ThSL verbs does not use the proposals of the

18

EVH. However, Chapter 5 does review the implications for the hypothesis based on the results of the study.

Kuhn (2017) presents an alternative analysis of the expression of telicity and duration in sign language predicates. He proposes that there is an iconic correlation between telic predicates articulated with rapid deceleration to a stop and durative predicates articulated with path movement. Wilbur (2008) herself posits an iconic origin for these morphemes but stresses that they have been grammaticalized. As Kuhn (2017) points out, however, a morphological analysis cannot account for the many ways in which the articulation of a predicate can be meaningfully manipulated. He illustrates this with the ASL sign DIE. Figure 10 shows the basic form of the sign.

Figure 10 ASL sign DIE (Kuhn, 2017, p. 12)

As Figure 10 shows, the ASL sign DIE is articulated by placing both hands in front of the signer with the fingers extended and held together. The palm of the dominant hand faces down, and the palm of the non-dominant hand faces up. Both wrists are then twisted so that the direction in which the palms face is reversed. The sign consists of a single movement that ends in an abrupt stop – the EndState morpheme according to the EVH.

However, the sign can be manipulated in several ways. Figure 11 demonstrates three of these. Note that the frames only show the signer’s dominant hand.

Figure 11 Manipulation of the ASL sign DIE (Kuhn, 2017, p. 12)

19

Figure 11 shows variations of the sign DIE made by manipulating the movement of the sign. In the first frame, the movement stops before the wrists are fully twisted. This form has the meaning ‘almost died’. In the second frame, the movement is paused slightly and then completed, conveying the meaning ‘died after a struggle’. Finally, in the far-right frame, the movement is paused several times before reaching its final point, meaning ‘gradual death’. Kuhn (2017) argues that the many ways in which the movement of a sign can be manipulated are best explained by an iconic analysis rather than a morphological one. Specifically, a morphological approach under-generates the number of possible forms (e.g., it does not account for all the forms found in Figure 11). An iconic approach correctly predicts both the forms and their meanings.

Although an iconic analysis of these manipulations is more elegant than a morphological one, it pertains primarily to manipulated rather than basic forms of predicates. Because of this, the plausibility of an iconic analysis is beyond the scope of the current study. Section 1.6 provides an outline of what will be covered.

1.6 Outline of the study The methodology used to investigate situation types in ThSL is presented in Chapter 2. The review includes a discussion of predicate and test selection. In Chapter 3, the tests are applied to ThSL predicates. In Chapter 4, the results of testing a subset of predicates with additional informants are described and compared with the initial results. Chapter 5 summarizes the key findings of the study and presents additional methodological considerations. It also discusses the implications of the study for the EVH before concluding with suggestions for areas of further research. Appendix A provides still shots of the ThSL predicates that were used in the study. Appendix B shows the questionnaire used to gather background information about informants. The tables in Appendix C summarize the information.

20

Chapter 2 Methodology

2.1 Overview The goal of this chapter is two-fold. First, it explains the rationale behind the selection of the tests (Section 2.2) and predicates (Section 2.3) for the study. Second, it reviews the process of data collection (Section 2.4), including information about language informants (Section 2.5).

2.2 Test selection Two tests were sought to identify each of the featural values of predicates: two tests for dynamism, two for duration, and two for telicity. Tests from studies of other languages were used whenever possible to facilitate cross-linguistic comparison. Section 2.2.1 discusses effective and ineffective tests for dynamism in ThSL, Section 2.2.2 discusses tests for duration, and Section 2.2.3 discusses tests for telicity. Some of the tests identified in these sections also distinguish stage-level from individual- level states. The table in Section 2.2.4 shows the tests, their bases, and which features they identify.

2.2.1 Dynamism Many of the tests used in studies of other languages did not adapt well to ThSL. Among these were several tests designed to identify the dynamic value of predicates. For instance, earlier works have observed that testing a predicate with manner adverbials can identify its dynamic value (Lakoff, 1965; Dowty, 1979, p. 55; Frawley, 1992, p. 152; Van Valin, 2005, p. 33). These adverbials, such as carefully, violently, strongly, or deliberately can often occur with dynamic predicates, but are incompatible with states. For example, the tree shook violently is acceptable, indicating that shook is [+dynamic]. In contrast, the tree was green violently is semantically ill-formed, showing the predicate is [-dynamic]. However, in ThSL, manner is most often communicated by modifying the form of the verb instead of adding an independent adverbial sign (Danthanavanich, 2008, p. 85). The

21

independent adverbial STRONG in ThSL was still attempted to represent this test. However, it proved to be ineffective due to limited collocational distribution.

Another test for dynamism is compatibility with the question what happened? (Smith, 1997, pp. 29, 243). In English, events such as the woman fainted can felicitously answer this question. A stative answer like the woman is kind sounds odd. Two signs for ‘happen’ were tried in an attempt to replicate this test. However, the first sign is a loan sign from ASL that is not widely known or used. The second sign had the more specific meaning ‘happen for the first time’, which made it infelicitous with ordinary events. This test, then, was unadaptable for ThSL.

Rathmann (2005, p. 67), following Parsons (1990), identifies another test for dynamism. Namely, he states that individual-level states cannot embed under verbs of perception. For example, Henry saw the tree shaking is acceptable, but Henry saw his dad knowing history is not. There are two issues with this test. First, the test is only designed to identify individual-level states (know history, be kind). Stage-level states can embed freely (e.g., Henry saw that his dad was tired/sick/mad) and so pattern with dynamic events. Because of this, Rathmann (2005) used the test specifically for distinguishing stage-level and individual-level states. However, the ThSL sign PERCEIVE.VISUALLY was used to replicate the test, states of both types could embed freely, making it unusable. Second, based on Rathmann’s presentation of the test, its effectiveness seems to hinge on the progressive form of the predicate more than its stative value. Thus, Henry’s dad is knowing history is odd, even without the verb of perception. In contrast, a non-progressive form can embed felicitously, as in Henry saw his dad knew history. These factors make the test ineffective for distinguishing states from events, especially if a progressive aspect has not been identified in the language. Because of these issues, the current study did not use this test.

An alternative test based on the volitional nature of many events (Dowty, 1979, p. 184; Frawley, 1992, p. 151; Smith, 1997, p. 42) was adaptable to ThSL. Volition allows dynamic situations to embed under verbs of force or persuasion, such as go buy milk or please sit down. In contrast, a state (specifically, entry into a state) is non- volitional. Hence, go be sick and please be tall sound strange. Rathmann (2005) utilized this feature in his study of situation types in (ASL) using the signs GO.AHEAD and DO.YOU.MIND. There is no direct equivalent to these signs in ThSL. Instead, requests use CAN or PLEASE, and commands use GO or a firm, final head nod at the end of the clause. Figure 12 illustrates the manual signs.

22

Figure 12 ThSL signs CAN (Frame 1), PLEASE (Frame 2), and GO (Frame 3)

In Figure 12, the first frame shows the sign CAN. It is formed by a closed fist with the thumb extended and pointing upward (“thumbs up”). The non-manual features of the sign include a forward lean, raised eyebrows, and lowered chin. These non- manual components mark the sign as a question. The sign may occur in a clause such as ‘can you help me?’ However, subtle variations in the movement and facial expression change the meaning to either ‘you can help me or ‘it is good for you to help me’. During initial testing, the sign would be subconsciously shifted between these forms, leading to mixed results. Because of this, it was discarded.

The second frame of Figure 12 shows the sign PLEASE. The signer holds an open hand palm up and moves it horizontally across the area in front of him. This sign was challenging to work with since it was too polite for most situations, so it was also discarded.

The final frame of Figure 12 shows the sign GO. The signer points his index finger outward and swings it sharply across the area in front of him. The sign is generally placed at the end of a clause to indicate a command. Alternatively, a single firm head nod can be co-articulated with the verb to indicate a command. Both command forms – GO and the final head nod – worked well for testing dynamism and were adopted as the first test for this feature. The final head nod is transcribed here using the convention VERBnod, where the subscript ‘nod’ indicates the head nod.

A second test for identifying the dynamism of ThSL predicates was based on the incompatibility of states with interruption. This incompatibility prevents stative situations from occurring with the sign INTERRUPT, as Figure 13 shows.

23

Figure 13 ThSL sign INTERRUPT

In Figure 13, the signer holds his non-dominant hand upright with the palm facing in. He extends the index and middle fingers of the dominant hand to form a V and moves it sharply around the wrist of the non-dominant hand as if cutting it. The sign occurs at the end of the clause. The reason for the interruption can be supplied in either the preceding or following clause (e.g., ‘I was swimming when interrupted by a leg cramp’).

The meaning of INTERRUPT is similar to that of the unrealized inceptive marking found in ASL (Liddell, 1984). In ASL, the marking is formed by modifying the movement and location of the predicate, adding the meaning ‘just about to begin x’. The modified form does not occur with states. The same restriction applies to ThSL INTERRUPT, making it an effective test for dynamism.

Although INTERRUPT can occur with both durative and instantaneous events, its semantic effect on the two is different. With durative events, the sign has the meaning ‘interrupted while doing x’. With instantaneous events, it means ‘kept from doing x’. The result is expected since instantaneous events lack duration (the end occurs immediately after the beginning). Thus, INTERRUPT causes a pre-process to be associated with the event. When this process is interrupted, the event itself does not occur. Because of the different semantic effects of INTERRUPT on these situation types, it can also be used to test the durative versus the instantaneous value of an event.

The test of interruption also interacts with the portrayal of an event. If an accomplishment is portrayed as completed (i.e., including the final endpoint), it can no longer be interrupted. Thus, the test INTERRUPT must be applied to a portrayal of a situation that does not entail the completion of the event (i.e., one that does not include final endpoints). For example, Superman was bending Captain America’s vibranium shield when interrupted by Lois Lane is acceptable since the event ‘bend’ is portrayed as an ongoing event (i.e., was bending). In contrast, Superman had bent

24

Captain America’s vibranium shield when interrupted by Lois Lane is ill-formed since the event is portrayed as completed (i.e., bent) and can no longer be interrupted. Thus, the test was used with uncompleted portrayals of predicates (see discussion in Section 1.3.2).

The two tests identified to determine dynamism in ThSL were compatibility with command forms and with the sign INTERRUPT. Dynamic situations can occur with both; states cannot occur with either. Section 2.2.2 reviews the tests that were selected for identifying duration.

2.2.2 Duration Two tests were also sought to identify the durative value of a predicate. First, a test for duration that was successfully implemented by Rathmann (2005) was tried. The test works on the premise that durative events can unfold in a certain manner (e.g., carefully, carelessly); instantaneous events cannot. In sign languages, the signer can express manner by modifying manual or non-manual features of the verb, or both. The ASL manner morphemes Rathmann utilized were non-manual mouth morphemes. These are formed by holding the mouth in a certain way during the articulation of the predicate. The first mouth morpheme, which conveys the meaning ‘carelessly’, is formed by projecting the tongue out slightly between the lips. The second morpheme, which means ‘in a regular manner’, is formed by pressing the lips together. However, although ASL has at least two such mouth morphemes, no similar morpheme was identified in ThSL. In place of these, the independent adverbial sign NO.FEELING was attempted. The sign, however, had too many collocational restrictions to be effective.

Another test to identify duration is based on the incompatibility of instantaneous events with pace adverbs like slowly (Smith, 1997, p. 42). Since instantaneous events do not conceptually take time, they either sound odd with pace adverbials or undergo a coerced interpretation. For example, the balloon popped slowly is semantically odd, whereas the nurse knocked slowly leads to an iterative interpretation of the event. In ThSL, the sign SLOW could also coerce a slow-motion interpretation of a single iteration of the event.

ThSL has two signs that mean ‘slowly’. One of these has the connotation ‘inordinately slow’ or ‘protracted’ and is represented here by VERY.SLOW. The other has the more neutral meaning ‘slow’ and is represented by SLOW. Figure 14 shows both signs.

25

Figure 14 ThSL signs VERY.SLOW (left) and SLOW (right)

In Figure 14, the first two frames show a variation of the sign VERY.SLOW. The signer’s right hand comes up with the palm facing left, and only the index finger extended. The forearm is brought back and up so that the hand ends beside the head. The next two frames of Figure 14 show the sign SLOW. In the first frame, the signer places his right index finger on the inside of his left wrist. The finger is then pulled backward and stops near the inside of the elbow (frame 2).

Initially, using pace adverbials as a test was avoided since sign languages often express pace by modifying the speed at which the predicate is articulated instead of adding a separate adverbial. In ThSL, it is possible to also use the adverbial signs in Figure 14 with a predicate, although this generally causes a slower articulation of the predicate as well. Compatibility with the adverbials SLOW and VERY.SLOW was, thus, chosen as the first test for duration.

Pace adverbials like SLOW can also indicate dynamism in durative events. Specifically, since durative, dynamic situations involve multiple stages, the stages can occur quickly or slowly. The same is not true of states since they lack internal stages. Thus, pace adverbials like SLOW can identify the durative value of events. They can also identify the dynamic value of durative events (i.e., accomplishments and activities).

Another test used by Rathmann (2005, p. 76) to identify duration was compatibility with temporal extent adverbials such as all day or one hour. The test worked well in ThSL as long as the duration specified adhered to pragmatic norms. Because of this requirement, the current study used several different adverbials. Figure 15 shows two of these signs.

26

Figure 15 ThSL signs DAY-THREE (left) and MINUTE-FIVE (right)

In Figure 15, the first two frames show the sign DAY-THREE ‘three days’. It is formed by placing the signer’s thumb at his temple and brushing it forward. The hand then opens to show the number of days (in this case, ‘three’). The third and fourth frames show the sign MINUTE-FIVE ‘five minutes’. The non-dominant hand is held open with the fingers together. The dominant hand is held with the index finger extended. It is placed against the open non-dominant hand and brushed forward, opening to show the designated number of minutes (in this case, ‘five’). These adverbials, and others like them, were used as the second test for duration.

Co-occurrence with SLOW or VERY.SLOW and adverbials of temporal extent were chosen as tests for the durative value of ThSL predicates. Both tests are compatible with durative events. When used with instantaneous events, they are either incongruous or cause the situation to receive a coerced interpretation.

In addition to identifying duration, both tests were able to distinguish stage-level states from individual-level states. First, when used with these predicates, both SLOW and VERY.SLOW were interpreted to mean ‘for a long time’. This shift made the adverbials incompatible with individual-level states since they tend to continue over the lifetime of the participant. The same predicates were also incompatible – for the same reason – with temporal extent adverbials (e.g., ‘one day’, ‘five minutes’). Stage- level states, on the other hand, occurred felicitously with both test frames.

The selection of tests to identify telicity, the final featural value, is described in Section 2.2.3. One of these tests (MANY.TIMES) can also serve to identify individual versus stage-level states.

27

2.2.3 Telicity Two tests were also sought to identify the telic value of ThSL predicates. A typical test used in English is compatibility with for an hour/month/year versus in an hour/month/year (Smith, 1997, p. 54). Atelic events are compatible with the former (he walked for an hour), but odd with the latter (he walked in an hour). However, ThSL does not use prepositions with these adverbials. Instead, signs like ONE.HOUR occur with both telic and atelic events. Thus, the test could not be adapted for ThSL.

An alternative test for English predicates, used by Smith (1997, p. 43), is the co- occurrence of spend with atelic situations (I spent an hour walking) and take with telic situations (it took an hour to walk four miles). Rathmann (2005) was able to adapt this test to ASL. However, a similar pattern was not identified for ThSL, and so the test could not be used.

Another test used by Rathmann (2005) was compatibility with the sign STILL. The test is based on Loebner’s (1989) and Krifka’s (2000) observations that still indicates a sentence is true at the time of utterance, making it incompatible with events that have reached a result state. Smith (1997, p. 64) also notes this phenomenon, highlighting that if the event is both telic and portrayed with a perfective viewpoint (i.e., viewed as a whole), it is incompatible with ‘still’. By implication, then, the remaining combinations of telicity and aspectual viewpoint are compatible with the test. Examples (15) through (18) illustrate these combinations.

(15) a. This morning the ice melted. b. *Is it still melting?

(16) a. This morning the ice was melting. b. Is it still melting?

(17) a. This morning Caroline was playing with her dolls. b. Is she still playing?

(18) a. This morning Caroline played with her dolls. b. Is she still playing?

The telic event melt in (15) and (16) possesses a result state. The perfective portrayal (conveyed by the past tense form of the verb, melted) in (15) indicates that a result state has been reached (i.e., the ice has undergone a total change from solid to liquid). The portrayal makes the clause incompatible with the question of continuation in (15). In contrast, the question can be used felicitously with the

28

remaining clauses. Specifically, while melting in (16a) is still a telic event, the imperfective aspect (conveyed by the progressive form of the verb), indicates the result state has not been reached. In contrast to ‘melt’, the event ‘play’ in (17) and (18) is atelic. This feature gives it the potential to continue, even when a perfective aspect is used, which makes it compatible with questions of continuation.

Since ThSL does not have a direct equivalent to the ASL sign STILL, alternative signs were sought. First, the sign CONTINUE was tried. However, it proved to be a discourse level sign meaning ‘and from x continued to y’ rather than ‘continued doing x’. Next, the sign STILL.THERE, which has both durative and locative meaning, was attempted. Informants identified the sign STILL.THERE as a loan sign from (HKSL) meaning ‘have’. In ThSL, however, it has the sense of an enduring state, event, or location. Figure 16 shows three variations of this sign, which is equivalent to the phrase ‘still there doing x’.

Figure 16 Variations of the ThSL sign STILL.THERE

The first frame in Figure 16 shows a variation of STILL.THERE used when the referent is present. In it, the signer rounds his lips, and holds a bent index finger against his chin. The second frame shows a variation of the sign used when the referent is not present. In this version, the signer still holds a bent index finger against his chin, and his lips are still rounded. In addition, the signer’s non-dominant hand points in the (idealized) direction of the referent. Finally, as the third frame shows, the sign can also co-occur with a fragment of the sign immediately preceding it in the clause. In the third frame, the immediately preceding sign was SELL. The signer’s non- dominant is held in the final position for this sign, with the palm up and fingers held open and together. The signer then places the bent index finger of the dominant hand again his chin and rounds his lips. Each of the variations in Figure 16 was accepted in the data, with informants choosing which form was best suited to each clause.

29

For initial elicitation sessions, STILL.THERE was placed in a simple statement frame meaning ‘x is still there doing y’. It was believed that this frame would be simpler to work with than the statement-question frame used for ASL. However, during subsequent sessions, the original statement-question frame was attempted. This form was also found to work well, and so both variations were used to test for telicity.

The second test for telicity found to work in ThSL on the observation that semelfactives (atelic instantaneous events) can be repeated in close succession (Van Valin, 2005, p. 38). Achievements, which are telic instantaneous events, cannot. The contrastive behavior is due to the result state encoded by achievements that prevents repetition of the event unless the end state is first reversed. Two studies (Pelkey, 2004; Phillips & Thiengburanathum, 2007) have used this feature to determine whether an instantaneous event is telic or atelic. Both studies used the test frame ‘just now … three times’ (e.g., ‘just now the nurse knocked three times’). However, in ThSL, the sign for ‘three times’ is minimally distinguished from the sign ‘a third time’. Because of this, when used with ‘just now’, informants reinterpreted the sign for ‘three times’ to mean ‘a third time’ to make the clause semantically acceptable. For example, ‘the student just now passed the test three times’ was changed to mean ‘the student just now passed the test a third time’. The change in the sign was made subconsciously when the informant signed the clause, making it a confounding factor for eliciting consistent judgments. The sign MANY.TIMES, which cannot shift to an ordinal meaning, was used instead. shows this sign.

Figure 17 shows this sign.

Figure 17 ThSL sign MANY.TIMES

The sign in Figure 17 is formed by placing a closed fist near the side of the nose, then moving the hand forward as the fingers spread open. By using this sign and dropping ‘just now’, a test was found that distinguished the telicity of both durative and instantaneous events. Thus, MANY.TIMES was chosen alongside STILL.THERE to identify the telic value of predicates in ThSL.

30

Although telicity does not technically apply to states (Smith, 1997, p. 20), it was found that the test MANY.TIMES could distinguish individual from stage-level states in ThSL. Specifically, stage-level states are temporary and can occur multiple times over the lifetime of the participant. This factor makes them compatible with MANY.TIMES. In contrast, individual-level states tend to continue over the lifetime of a participant, making them infelicitous with the test. This behavior made MANY.TIMES a useful test for identifying stage-level versus individual-level states in ThSL. In contrast, both types of states occur felicitously with the second test for telicity, STILL.THERE. Since the test was not needed to test for telicity, and it did not distinguish types of states, the STILL.THERE test with states is not presented in Chapter 3 and Chapter 4.

2.2.4 Tests chosen for ThSL Table 2 provides a summary of the tests selected for identifying situation types in ThSL. The first column shows the test’s number. The second column identifies the test itself.

Table 2 Tests chosen to identify the featural values of ThSL predicates

level level -

Test

level level

-

Individual Stage Activities Accomplishments Achievements Semelfactives Identifies dynamic value 1 GO Premise: volition allows events to X X ✓ ✓ ✓ ✓ embed under imperatives Identifies dynamic and durative values 2 INTERRUPT X X ✓ ✓ ✓* ✓* Premise: volition and duration allow interruption Identifies dynamic and durative SLOW, values 3 X ✓ ✓ ✓ X* X* VERY.SLOW Premise: motion and duration allow pace modification Identifies durative value MINUTE-FIVE, 4 Premise: duration allows temporal X ✓ ✓ ✓ X* X* ALL.DAY, etc. extent to be specified Identifies telic value 5 STILL.THERE Premise: events cannot continue once - - ✓ X X ✓ the result state has been reached Identifies telic value Premise: events that involve a result 6 MANY.TIMES X ✓ ✓ X X ✓ state cannot continue until the state is reversed

31

The numbers shown in the first column of Table 2 are used in later sections to refer to the tests. For example, Test 1 is used to refer to compatibility with commands. The second column shows the ThSL sign or signs used for each test. In the third column, the upper cell for each test shows the featural value (or values) the test identifies. The lower cell identifies the premise for the test. In the remaining columns, an “X” indicates that the test is incompatible with the situation type listed in the heading; a checkmark indicates that a test is compatible. An asterisk indicates further considerations. For example, when achievements and semelfactives occur with INTERRUPT, the process leading up to the event is interrupted, not the event itself. Thus, the cells for these combinations have both a checkmark and an asterisk. On the other hand, clauses that have an X and an asterisk are those that should be infelicitous, but under certain conditions may be acceptable. For instance, both achievements and semelfactives should be incompatible with temporal extent adverbs (Test 4). However, if they receive an iterative interpretation, the clause may be acceptable. Tests 3, 4, and 6 also serve to distinguish an individual-level state from a stage-level state. Section 2.3 discusses the ThSL predicates that the study tested.

2.3 Predicate selection Five predicates were chosen to represent each of the situation types identified by Smith (1997) (see Table 1). A set of predicates for both individual-level and stage- level states were included, bringing the total number of predicates to 30. Several restrictions guided the selection of these signs.

First, only single-sign predicates were chosen. Predicates involving multi-verb constructions (e.g., HEART.STOP DIE ‘its heart stopped, and it died’) or other complex constructions (e.g., BORN HMONG DNA ABSORB ‘be Hmong’) were avoided. Since several tests depended on the agency of the subject, predicates that took animate, volitional subjects were preferred over ones that took non-volitional ones. Another consideration was whether the predicate was a central or borderline example of the situation type. For example, in several languages verbs of posture and position (e.g., ‘sit’, ‘stand’, ‘lean’, ‘perch’) form a unique class of predicates that reflect properties of both states and activities (Smith, 1997, pp. 39–40). Other predicates like ‘wait’ also reflect properties of more than one situation type. Borderline predicates like these were avoided, and more central examples were chosen. Table 3 shows the predicates that, based on these limitations, were ultimately selected.

32

Table 3 Predicates used to test for situation types in ThSL

Situation type ThSL Predicate

States – individual BEAUTIFUL, HONEST, SHORT, SMART, TALL States – stage AGITATED, ANGRY, BORED, SAD, SICK Activities DRIVE.MOTORCYCLE, PAINT, RIDE.BICYCLE, RUN, SWIM Accomplishments BEND.BAR, GROW.BEARD, ICE.MELT, SKY.DARKEN, STRAIGHTEN.BAR Achievements BLOCK, ERASE, EXPLODE, OPEN.CAN, TURN.OFF.LIGHT Semelfactives CLAP, KNOCK.ON.DOOR, PERSON.DUNK, PUNCH.PERSON, SHOUT

In Table 3, the first column identifies the situation type for each predicate. The predicates are listed in the second column in alphabetical order according to their English representation. The remainder of the study generally follows this order for consistency and ease of reference. Although each predicate is a single sign in ThSL, compound transcriptions are used in some cases (e.g., BEND.BAR, TURN.OFF.LIGHT) to more clearly express the meaning of a sign. Still shots showing each of the predicates can be found in Appendix A.

The sign BLOCK is used for blocking on social media. The sign ERASE can have several meanings, including ‘uninstall’ and ‘erase’. For this study, it was used in the context of removing the social media application Line from a cellphone or erasing a message.

The meaning and form of some of the predicates warrant further explanation. First, the sign SHOUT means to utter a single, loud cry. The subject referent may be a person or animal (e.g., ‘bark’). Next, the sign SKY.DARKEN often refers to nightfall. However, it can also refer to the sky darkening due to cloud cover. The signs PERSON.DUNK, PUNCH.PERSON, RUN, and SWIM each have two forms in ThSL – a depicting verb form and a non-depicting form. As far as possible, only examples that used the non-depicting form were used (see Appendix A). However, the two forms often occurred together in multi-verb constructions. These constructions were only used if examples using the single, non-depicting verbs were unavailable.

Next, the predicates BEND.BAR, ICE.MELT, STRAIGHTEN.BAR, OPEN.CAN, KNOCK.ON.DOOR, PERSON.DUNK, and PUNCH.PERSON involve classifier incorporation (see Section 1.4.3). As far as possible, the same form of the predicate was used for all examples. In rare cases, however, no example was available with the chosen classifier, and an alternative form (e.g., one using a classifier for small ice cubes in a cup versus a

33

large ice cube on a flat surface) was used. Section 2.4 discusses the process of identifying the featural values of these ThSL predicates.

2.4 Data elicitation The present study grew out of two smaller investigations. These earlier works identified practical tests, revealed considerations for predicate selection, and refined the progress for eliciting data. This section briefly describes each study and the methodological insights they revealed. The discussion is followed by general observations on eliciting and processing data.

The first study used two predicates to represent each situation type. In order to elicit a natural clause in which to use each predicate, the informants were shown pictures representing basic-level clauses and asked to sign them (e.g., a picture with a submerged boat was used to elicit the clause ‘the boat sank’). Once informants had provided these basic-level constructions, I then signed them in a series of test frames, asking informants if the resulting constructions were well-formed or not. Their responses were recorded in a Word document.

This initial study yielded several useful insights for methodology. First, it identified several tests that did not work well in ThSL, such as NO.FEELING and PLEASE. It also highlighted the importance of using prototypical predicates with volitional, agent- like arguments. Thus, for example, clauses like ‘the boat sank’ were omitted from the current study. Finally, typed notes were insufficient for capturing the data.

A second study (Wallace, 2020) was then conducted. Since many tests from the first study had proven ineffective, an initial session was held with informants to discuss alternative tests. Tests that did not work well in ThSL (e.g., CAN, CONTINUE, HAPPEN, STRONG) were discarded, and new ones tried until two tests had been identified for each featural value.

In order to apply these tests to predicates, two methodological tools used by Kuhn (2017, p. 3) were tried. First, video recordings were made of informants signing each predicate in each test frame. This step created a video questionnaire signed by native signers.

The format seemed preferable to signing the clauses myself since I am not a native signer. In order to make the video questionnaire, the informants were asked to sign each clause in each test frame without regard to its well-formedness. This (intentional) disregard for acceptability caused informants to sign on auto-pilot,

34

resulting in stilted and unnatural signing. This made them more distracting than helpful when used later to elicit acceptability judgments.

A second tool used by Kuhn (2017, p. 3) was a 7-point well-formedness scale along which informants could rate constructions. The well-formedness of some clauses is questionable rather than wholly unacceptable, which made a scalar system appealing. Since a 7-point scale seemed unnecessarily complicated, ThSL informants were asked to make acceptability judgments along a 5-point scale. When their ratings were later compared for analysis, however, no patterns could be identified. Scalar judgments were, thus, omitted from subsequent sessions, and group discussions were relied on to determine the acceptability and meaning of clauses.

Although these two tools (the video questionnaire and scalar judgments) did not work well, several other adjustments in methodology were beneficial. For example, in the second study, each clause was presented and then discussed by the informants. The discussion identified 1) if the clause was well-formed, 2) any reasons for unacceptability, and 3) instances of coercion. In addition, each session was filmed in its entirety. This step made the information readily available for later analysis.

For the current study, an expanded set of predicates were chosen and placed in the test frames identified in the second study. For data elicitation sessions, I would sign a clause (i.e., place one of the predicates in one of the test frames) and ask informants to discuss its meaning and acceptability. During the discussion, the informants often made subtle changes to the original clause. For instance, they would modify the form of the predicate (e.g., reduplicate it), or create a serial verb construction, or move the test frame to a new position in the clause. These subconscious changes often indicated a clash between the predicate and test frame and suggested which follow-up questions to ask. For example, since markers of temporal duration are incompatible with instantaneous events, the signers consistently shifted the adverbial (e.g., MINUTE-FIVE) to the beginning of the clause for semelfactives and achievements. This reordering changed the meaning from ‘x happened for y time’ to ‘in y time, x will happen’. When such alterations occurred, the form, meaning, and well-formedness of the new clause were noted. Afterward, the original clause was again presented for discussion.

The sessions took place in Bangkok and Chiang Mai, Thailand. In Bangkok, a Sony FDR AX33 was used to film the first sessions. Later sessions were filmed with a Panasonic HDC-TM900. The Panasonic was set to film at 960 x 540 resolution at

35

30/fps. This resolution produced sufficiently clear footage for analysis, although it was not crisp. The Sony was set to 1920 x 1080p at 25/fps and produced brighter and clearer footage. In Chiang Mai, a Canon EOS 80D was used. It was set to 1920 x 1080 at 25/fps and produced the best footage of the three cameras. In terms of frame rate, both 25 and 30/fps were generally sufficient to capture the information in elicitation sessions. However, for creating still shots (such as those used in Appendix A), the selection of frames was limited, and sometimes only blurred images were available.

As far as possible, the cameras were plugged into a power outlet while filming to keep them from turning off mid-session due to depleted batteries. The footage was recorded directly onto an SD card and copied onto a laptop during breaks. Once copied, the original files were deleted from the card to prevent the camera from shutting down mid-session due to a full memory card.

For analysis, the video files were first copied and then the original files were saved in a separate folder for safekeeping. The copies were next compressed using Handbrake (HandBrake, 2019) since smaller files worked more smoothly with ELAN (ELAN, 2019), a video annotation program. ELAN was used to annotate the videos for easy reference. The annotation made it easy to locate relevant portions of the discussion later for analysis. After annotating the videos, the results were compiled in an Excel spreadsheet for easy comparison. The spreadsheet also served to identify gaps in the data that needed to be filled in during subsequent elicitation sessions.

During the Bangkok sessions, the full set of 30 predicates was tested with a core group of Deaf. Two to six members of this group were present for each session. In general, it was found that having more participants made informants more confident in their well-formedness judgments and added energy to the sessions. However, larger groups also posed challenges. Because of this, extended testing was done with groups of four. This number proved to be ideal; participants were confident in their judgments, stayed engaged, and fit comfortably within the video frame. It also made it possible for me to see whenever someone started to sign, something that had been difficult with larger groups. Section 2.5 gives background information about the participants who helped in these sessions.

36

2.5 Language informants Three groups provided data for this study. The first group was composed of six native signers of ThSL who were residing in Bangkok. This group is referred to throughout as the Original Group (informants 01-06). The signers in this group ranged in age from 25 to 31. Among the six, four were men, and two were women. Five identified as Deaf/deaf while the last identified as hard of hearing. All informants reported learning ThSL when they began attending a Deaf school at age 8 or 9. Although the informants were living in Bangkok, they had attended schools in Nonthaburi, Nakhorn Pathom Prachin Buri, Chonburi, Suphan Buri, and Nakhon Ratchasima. Four of the six had completed university, one had done vocational training after 9th grade, and one had stopped school after the sixth grade.

After testing with the Original Group, twelve of the predicates were tested with two new groups of Deaf (informants 07-14). Each of these later groups had four members. This time, the sessions were held in Chiang Mai, Thailand. Together, the Chiang Mai groups included five men and three women ranging in age from 28 to 39. Seven were Deaf/deaf, and one was hard of hearing. One had learned ThSL from a Deaf older sibling before he began attending the local school for the Deaf. The others had learned at the local Deaf school when they were between 6-12 years old. All the informants were residing in Chiang Mai. They had attended schools in Chiang Mai, Bangkok, Songkhla, and Nakhon Pathom provinces, with several having attended school in more than one province. Each informant had completed the 12th grade, and seven had gone on to complete university. Six of the informants taught or had taught at the Anusarnsunthorn School for the Deaf in Chiang Mai.

More information on the language informants from all groups is available in Appendix C. Appendix B provides the original questionnaire used to gather information.

No data was found that indicated a dialectal difference between informants in relation to predicates’ situation types. Chapter 3 presents the results of testing the full set of 30 predicates with the Original Group. Chapter 4 covers the results from testing with Groups A and B and compares the results with those of the first group.

37

Chapter 3 Situation Types in Thai Sign Language

3.1 Overview The following results are grouped according to situation type. Section 3.2 presents the results for individual-level states. These are followed by stage-level states in Section 3.3, activities in 3.4, accomplishments in 3.5, achievements in 3.6, and semelfactives in 3.7. Section 3.8 summarizes the results.

3.2 Individual-level states States are non-dynamic, durative situations. These stative situations consist of transitory, stage-level states like be sick and enduring, individual-level states like be tall (Carlson, 1980, p. 106). In ThSL, the distinction between these two situation types can be seen by their behavior with three of the tests: pace adverbials (Test 3), temporal extent adverbials (Test 4), and repeatability (Test 6). Thus, although the feature of telicity does not apply to states (Smith, 1997, p. 20), the results from testing the ThSL stative predicates with MANY.TIMES are included here. The second test for telicity (STILL.THERE) is not used.

The predicates that represent individual-level states for this study are BEAUTIFUL, HONEST, SHORT, SMART, and TALL. In Section 3.2.1, these predicates are tested for dynamism, and in Section 3.2.2 they are tested for duration. Section 3.2.3 shows the results of testing them with MANY.TIMES to identify their status as individual-level states.

3.2.1 Testing for dynamism – individual-level states Since states are non-volitional, stative predicates tend to be incompatible with commands (Test 1) and interruption (Test 2). Clauses (19) through (23) show the predicates with the first test frame.

(19) *BEAUTIFUL GO ‘Go be beautiful!’

38

(20) *HONEST GO ‘Go be honest!’

(21) *SHORT GO ‘Go be short!’

(22) *SMART GO ‘Go be smart!’

(23) *TALL GO ‘Go be tall!’

In (19) through (23), the predicates BEAUTIFUL, HONEST, SHORT, SMART, and TALL are all incompatible with command forms, indicating they are [-dynamic]. The predicates are shown with the second test frame, INTERRUPT, in (24) through (28).

(24) *BEAUTIFUL INTERRUPT ‘(My) being beautiful was interrupted by …’

(25) *HONEST INTERRUPT ‘(My) being honest was interrupted by …’

(26) *SHORT INTERRUPT ‘(My) being short was interrupted by …’

(27) *SMART INTERRUPT ‘(My) being smart was interrupted by …’

(28) *DOCTOR TALL INTERRUPT ‘The doctor’s being tall was interrupted by …’

In (24) through (28), the predicates BEAUTIFUL, HONEST, SHORT, SMART, and TALL are incompatible with INTERRUPT. These results again indicate that the predicates are [-dynamic]. Section 3.2.2 shows the results of testing the durative value of individual-level states.

3.2.2 Testing for duration – individual-level states Although individual-level states are durative, they are incompatible with pace adverbials meaning ‘slow’. The incompatibility is a result of their stative nature, which lacks an underlying premise of motion. Because of this, a state cannot occur

39

slowly. This factor often caused the sign SLOW to be re-interpreted as ‘for a long time’. Examples (29) through (30) illustrate this phenomenon.

(29) BEAUTIFUL SLOW ‘She (kept herself beautiful) for a long time.’

(30) CHEAT ZERO HONEST SLOW ALREADY ‘(He) has not cheated at all – he has (acted) honestly for a long time already.’

In (29) and (30), the predicates BEAUTIFUL and HONEST are compatible with SLOW when it is interpreted to mean ‘for a long time’. This interpretation was evidenced by the paraphrases that informants offered (e.g., ‘beautiful for one month’ or ‘honest year after year’). The shift is unusual in light of Smith’s (1997) principle of external override, which states that the meaning of the predicate will shift, not the meaning of the adverbial. Potentially, then, the different interpretation reflects a case of polysemy in which one sign has two related meanings. The meaning is then determined based on the situation type of the clause.

The same interpretation of SLOW, however, was only marginally acceptable with SMART under a coerced interpretation of the situation. The two clauses in (31) and (32) demonstrate.

(31) ?INDEX-3 FRIEND SMART SLOW ‘There, (my) friend, has (acted) wisely for a long time.’

(32) INDEX-3 SMART SLOW ‘He became smart slowly.’

In (31), the meaning of SMART is interpreted as a series of dynamic events – ‘acted wisely’. This interpretation shifts the stative situation into a derived-level activity. Even with this shifted meaning the sign SLOW is still interpreted to mean ‘for a long time’, and the clause is only marginally acceptable. In (32), SMART is again coerced into an event, ‘become smart’. This dynamic interpretation allows the situation to be modified by ‘slowly’. The incompatibility of a stative interpretation of the predicate with a pace adverbial indicates it is [-dynamic].

The results of testing SHORT and TALL also reflect the tension between pace adverbials and stative predicates. The clauses in (33) and (34) show the predicates in this test frame.

40

(33) *SHORT SLOW ‘(I) was short slowly.’

(34) *MAN TALL SLOW ‘The man was tall slowly.’

Both SHORT in (33) and TALL in (34) were infelicitous with SLOW, indicating they are [-dynamic]. Their incompatibility, in this case, is due to their stative nature.

The second test for duration is compatibility with temporal extent adverbials (Test 4). These adverbials imply the termination of the situation, which conflicts with the enduring nature of individual-level states. The clauses in (35) through (39) reflect this conflict.

(35) *BEAUTIFUL ONE.MONTH ‘(I) was beautiful for a month.’

(36) *INDEX-1 FRIEND HONEST ONE.WEEK ‘My friend was honest for a week.’

(37) *SHORT ONE.WEEK ‘(I) was short for a week.’

(38) *SMART ONE.WEEK ‘(I) became smart in a week.’

(39) *DOCTOR TALL ALL.DAY DAY.ONE ‘The doctor became tall in one day.’

In (35) through (39), the stative predicates BEAUTIFUL, HONEST, SHORT, SMART, and TALL are incompatible with temporal extent adverbs, indicating they are individual- level (rather than stage-level) states.

Although all of the clauses in (35) through (39) are infelicitous, their specific results vary slightly. First, the presence of the temporal extent adverbials produces ill- formed clauses in (35) through (37). In (38) and (39), the clauses are ill-formed, and an inchoative reading is given to the predicates, changing their meanings to ‘become smart’ and ‘become tall’. This shift coerces the situations from states into accomplishments. However, the test used time frames (‘one week’ and ‘one day’) that were logically insufficient to produce the indicated changes. This made the constructions semantically ill-formed. The predicates SHORT and SMART were used

41

in extended testing with time frames of several years. The results for extended testing can be found in Section 4.2.1.‎

The results from the two tests for duration provide several insights into the featural values of the first predicate set. First, the predicates do not occur with SLOW when it has the meaning ‘slow’. The incompatibility indicates the predicates are [-dynamic]. The marginal acceptability of three of the predicates (BEAUTIFUL, HONEST, and SMART) with SLOW when it is interpreted to mean ‘for a long time’ indicates the predicates are [+durative]. The incompatibility of the predicates with adverbials of temporal extent (Test 4) indicates they are individual-level states. Section 3.2.3 presents the results from testing the predicates with MANY.TIMES.

3.2.3 Testing for state type with MANY.TIMES Individual-level states tend to continue over the lifetime of the referent. This tendency makes them incompatible with the test MANY.TIMES. The clauses in (40) through (43) show the predicates in this test frame.

(40) *FRIEND INDEX-3 BEAUTIFUL MANY.TIMES ‘(My) friend, she has been beautiful many times.’

(41) *INDEX-1 FRIEND SHORT MANY.TIMES ‘My friend has been short many times.’

(42) *INDEX-1 FRIEND SMART MANY.TIMES ‘My friend was smart many times.’

(43) *DOCTOR TALL MANY.TIMES ‘The doctor was tall many times.’

As the examples show, the predicates BEAUTIFUL in (40), SHORT in (41), SMART in (42), and TALL in (43) are all incompatible with MANY.TIMES. The results reflect their enduring nature as individual-level states.

Only one of the individual-level states, HONEST, occurred felicitously with MANY.TIMES. The clause in (44) shows the predicate in this test frame.

(44) HONEST MANY.TIMES CHEAT ZERO ‘(I) have acted honestly many times, (I’ve) not cheated at all.’

42

The presence of MANY.TIMES in (44) causes a dynamic interpretation of HONEST, ‘I have acted honestly’. The shift is in line with Croft’s (2012, p. 38) observation that dispositional predicates can be construed as activities (e.g., compare Sam is kind with Sam is being kind). The value of the adverbial shifts the situation into a derived level activity in keeping with the principle of external override. The shift contrasts with the basic-level value of the HONEST that was seen in (20) and (25) where its incompatibility with GO and INTERRUPT indicate it is [-dynamic].

The results of testing ThSL individual-level states are summarized in Table 4.

Table 4 Test results for individual-level states in ThSL

Predicate

TIMES

THERE

.

.

GO INTERRUPT ADVERBIALS PACE TEMPORALEXTENT STILL MANY

BEAUTIFUL X X * X n/a X HONEST X X * X n/a X* SHORT X X X X n/a X SMART X X X X n/a X TALL X X X X n/a X X – incompatible  - compatible * – further considerations

The predicates in the far left column of Table 4 are incompatible with the command GO and the sign INTERRUPT, indicating they are [-dynamic].

The results for the next test, SLOW, are mixed. While both BEAUTIFUL and HONEST occurred with SLOW, the adverbial was interpreted to mean ‘for a long time’ and the predicates were coerced into dynamic situation types (e.g., ‘acted honestly’). For SMART, the predicate was interpreted as a derived-level activity but was still only marginally acceptable with SLOW. The final two predicates were infelicitous with SLOW. Despite these differences in details, the results all point to a [-dynamic] value for the predicates.

All the predicates were incompatible with adverbials of temporal extent. The results indicate that the predicates are individual-level states rather than stage-level ones.

43

Finally, most of the predicates were incompatible with MANY.TIMES, indicating they are individual-level states. The sign HONEST was the only exception. When it occurred with MANY.TIMES, it was interpreted as a series of events (‘act honestly’). As Section 3.3 shows, this phenomenon is more common with stage-level states. However, if HONEST were indeed a stage-level state, it should have been compatible with adverbials of temporal extent. As seen in (36), this is not the case. The results may indicate that HONEST is a less central example of an individual-level state. The coerced interpretation aligns with the behavior of dispositional predicates, which “denote ability or preference” (Smith, 1997, p. 34). Croft (2012, p. 38) observes that such states are especially prone to coercion into activities. For example, friendly becomes an activity in the clause John is being friendly (Dowty, 1979, p. 114).

3.3 Stage-level states Stage-level states are transitory. This feature distinguishes them from individual- level states. The predicates chosen to represent stage-level states in ThSL are AGITATED, ANGRY, BORED, SAD, and SICK. They are tested for dynamism in Section 3.3.1 and duration in Section 3.3.2. They are then used with the test MANY.TIMES in Section 3.3.3.

3.3.1 Testing for dynamism – stage-level states The stative value of stage-level states makes them incompatible with commands (Test 1). The clauses in (45) through (49) show the predicates in this test frame.

(45) *AGITATED GO ‘Go be agitated!’

(46) *ANGRY GO ‘Go be angry!’

(47) *BORED GO ‘Go be bored!’

(48) *SAD GO ‘Go be sad!’

(49) *SICK GO ‘Go be sick!’

44

As (45) through (49) show, the predicates AGITATED, ANGRY, BORED, SAD, and SICK respectively, are all infelicitous with GO. The results indicate that the predicates are [-dynamic]. In (50) through (54) the predicates are shown with the second test INTERRUPT.

(50) *TEACHER ANGRY INTERRUPT ‘The teacher was angry when interrupted by …

(51) *GRANDMA SICK INTERRUPT ‘Grandma was sick when interrupted by … ’

(52) ?AGITATED INTERRUPT ‘(I) was interrupted from being agitated by …’

(53) ?BORED INTERRUPT ‘(I) was interrupted from being bored by …’

(54) ?SAD INTERRUPT ‘I was interrupted from being sad by …’

The predicates ANGRY in (50) and SICK in (51) are unacceptable with INTERRUPT. The remaining predicates – AGITATED in (52), BORED in (53), and SAD in (54) – were partially acceptable. Informants observed that a sudden event could interrupt the indicated emotion. For example, a kind word could calm agitation, or an unexpected visit could end boredom. However, these cases involved implications of feeling or thinking (e.g., ‘I was feeling agitated’, ‘I was thinking sad thoughts’) that coerced the situations into more dynamic events. Even then, informants indicated that other phrases expressed this kind of termination more naturally (e.g., ‘decided not to be bored’, ‘calmed down from being agitated’). These observations, along with the incompatibility of the predicates with commands, indicate they are [-dynamic].

3.3.2 Testing for duration – stage-level states Unlike individual-level states, stage-level states are compatible with pace adverbials. The clauses in (55) through (59) show the predicates with this test frame.

(55) INDEX FRIEND AGITATED SLOW ‘(My) friend there has been agitated for a long time.’

(56) TEACHER ANGRY SLOW ‘The teacher was angry for a long time.’

45

(57) BORED SLOW ENDURE NEG ‘(I) have been bored for so long, (I) can’t take it anymore.’

(58) SAD SLOW ALREADY ‘Has (she) already been sad for a long time?’

(59) GRANDMA SICK SLOW ‘Grandma has been sick for a long time.’

As (55) through (59) illustrate, the predicates occurred felicitously with SLOW. However, the adverbial was consistently interpreted to mean ‘for a long time’ instead of ‘slow’. The interpretation was again evidenced by paraphrases in which informants used adverbs such as ‘for several days’. Since the predicates did not occur with SLOW meaning ‘slow’, they pattern as [-dynamic] situations. Their occurrence with SLOW meaning ‘for a long time’ indicates they are [+durative].

As transitory situations, stage-level states are also compatible with other adverbials of temporal extent. The clauses in (60) through (64) show the predicates with this test.

(60) INDEX-1 FRIEND INDEX-3 AGITATED ONE.HOUR AGITATED WHAT ‘My friend, he has been agitated for an hour. What is he agitated about?’

(61) TEACHER ANGRY MINUTE-FIVE CALM.DOWN ‘The teacher was angry for five minutes and then calmed down.’

(62) BORED ALL.DAY ‘(I) was bored all day.’

(63) FRIEND INDEX-3 INDEX-1 CLOSE INDEX-3 SAD ALL.DAY ‘A friend, he and I are close, he was sad all day.’

(64) GRANDMA SICK DAY-TWO ‘Grandma has been sick for two days.’

In (60) through (64), the stative predicates AGITATED, ANGRY, BORED, SAD, and SICK are compatible with adverbials of temporal extent. The compatibility indicates they are [+durative].

46

3.3.3 Testing for state type with MANY.TIMES Unlike an individual-level state, a stage-level state can be experienced multiple times over a participant’s lifetime. Theoretically, then, stage-level states should be compatible with MANY.TIMES. The clauses in (65) through (67) show the predicates AGITATED, BORED, and SAD with this test.

(65) INDEX-1 FRIEND INDEX-3 AGITATED MANY.TIMES ‘My friend, he has been agitated many times.’

(66) FRIEND BORED (BORED) MANY.TIMES ALREADY ‘My friend has been bored many times.’

(67) FRIEND SAD (SAD) MANY.TIMES ‘My friend has been sad many times.’

In (65) through (67), the states AGITATED, BORED, and SAD combine with MANY.TIMES to create well-formed sentences. This compatibility is expected since transitory states can be repeated. For AGITATED, which already involves repeated movement (see Appendix A), only a single iteration of the predicate was necessary. For BORED, most informants preferred a single articulation. One informant preferred a double movement. For SAD, most informants preferred a single movement, stating that a double movement looked too similar to a sign meaning ‘repeatedly hungry’.

The remaining predicates, ANGRY and SICK are shown with MANY.TIMES in (68) through (71).

(68) *TEACHER ANGRY MANY.TIMES AFRAID ‘The teacher was angry many times, (so I am) afraid (of him).’

(69) *GRANDMA SICK MANY.TIMES ALREADY ‘Grandma was sick many times already.’

(70) TEACHER ANGRY ANGRY MANY.TIMES AFRAID ‘The teacher has been angry many times, (so I am) afraid (of him).’

(71) GRANDMA SICK SICK MANY.TIMES ALREADY ‘Grandma has been sick many times already.’

In (68) and (69), a single iteration of the predicates ANGRY and SICK are used with MANY.TIMES. Informants reported that these clauses were infelicitous. In contrast, the

47

predicates are each signed twice in (70) and (71). This reduplication made the predicates compatible with MANY.TIMES.

Several factors may lie behind the reduplicated forms. First, on a syntactic level, it may be that only certain signers require reduplication of stage-level states in multi- event clauses. Second, emotive states like ANGRY, BORED, and SAD may be more likely to undergo reduplication. A third possibility is that phonological restrictions govern reduplication. The distinctive behavior of AGITATED supports this analysis. It was the only predicate that was never reduplicated and was the only predicate whose basic form already involved complex movement. Further data may illuminate which factor or factors underlie the required reduplication of some stage-level states when they occur with MANY.TIMES.

Table 5 captures the results from testing AGITATED, ANGRY, BORED, SAD, and SICK.

Table 5 Test results for stage-level states in ThSL

Predicate

TIMES

THERE

.

.

GO INTERRUPT ADVERBIALS PACE TEMPORALEXTENT STILL MANY

AGITATED X ? *  X  ANGRY X X *  X * BORED X ? *  X  SAD X ? *  X  SICK X X *  X * X – incompatible  - compatible ? – questionable * – further considerations

As the second column of Table 5 shows, the predicates were incompatible with command forms, indicating they are [-dynamic]. The predicates were generally judged to be incompatible with the second test, INTERRUPT, as well. The results again indicate they are both [-dynamic]. Marginal cases of acceptability rested on coerced, dynamic interpretations of the predicates.

The predicates also occurred felicitously with SLOW when it was interpreted to mean ‘for a long time’. The incompatibility of the predicates with the meaning ‘slow’ again indicates they are [-dynamic]; their compatibility with the meaning ‘for a long time’ indicates they are [+durative]. Each of the predicates occurred felicitously with

48

adverbials of temporal extent, again indicating they are [+durative]. The results from the tests further identify the predicates as stage-level states. This conclusion is supported by the compatibility of the predicates with MANY.TIMES. Two of the predicates, ANGRY and SICK, had to be reduplicated to occur with the test. The reason for this requirement is unclear. It may be semantic, or phonological.

The remaining predicates tested in the study belong to dynamic situation types. Predicates from the first of these situation types, activities, are tested in Section 3.4.

3.4 Activities Activities are dynamic, durative, atelic situations. The feature of dynamism makes them compatible with commands and interruption (Tests 1 and 2), as long as the subject referent is volitional. The concept of motion associated with dynamic events, and the durative value associated with activities, makes them compatible with pace adverbials (Test 3). The durative value of activities also allows them to occur with adverbials of temporal extent (Test 4). Finally, since activities are atelic, they are compatible with expressions of continuation (Test 5) and can be repeated with the same participant (Test 6).

The predicates chosen to represent activities in ThSL are DRIVE.MOTORCYCLE, PAINT, RIDE.BICYCLE, RUN, and SWIM. In Section 3.4.1 these predicates are shown with tests for dynamism, in Section 3.4.2 with tests for duration, and in Section 3.4.3 with tests for telicity.

3.4.1 Testing for dynamism – activities The predicates are shown with the first test, commands, in (72) through (76).

(72) DRIVE.MOTORCYCLE GO ‘Go drive a motorcycle (to 7-11 for snacks).’

(73) PAINT GO ‘Go paint!’

(74) RIDE.BICYCLE GO ‘Go ride a bicycle!’

(75) RUN GO ‘Go run!’

49

(76) SWIM GO ‘Go swim!’

The compatibility of the predicates in (72) through (76) with GO indicates they are [+dynamic]. They are shown with the test INTERRUPT in (77) through (81).

(77) DRIVE.MOTORCYCLE INTERRUPT ‘(I) was driving a motorcycle when interrupted by …’

(78) GRANDMA PAINT INTERRUPT ‘Grandma was painting when she was interrupted by …’

(79) INDEX-3 FRIEND RIDE.BICYCLE INTERRUPT WHAT FLAT.TIRE ‘My friend was riding a bicycle when interrupted by the tires going flat.’

3 (80) RUN RUNDEPICTING.VERB INTERRUPT PERSON.FALL SKID.ON.FACE ‘(My friend) was running when interrupted by falling and skidding on her face.’

(81) GRANDPA SWIM INTERRUPT ‘Grandpa was swimming when he was interrupted by … ’

In (77) through (81), the predicates are compatible with INTERRUPT, again indicating that they are [+dynamic]. Since the event itself is interrupted, the results also indicate that the predicates are [+durative]. Section 3.4.2 presents the results from testing the durative value of the predicates.

3.4.2 Testing for duration – activities Since activities are both dynamic and durative, they are compatible with pace adverbials. The clauses in (82) through (86) show the activity predicates with this test frame.

(82) DRIVE.MOTORCYCLE SLOW DRIVE.MOTORCYCLE TURTLE SLOW ‘(My friend) drives his motorcycle so slowly! He drives like a turtle – slow!’

(83) GRANDMA PAINT SLOW ‘Grandma painted slowly.’

3 As noted in Section 2.3, SWIM and RUN have depicting verb forms (i.e., iconic forms) and non- depicting forms. Often, both forms were signed in succession in a multi-verb construction. This seems to be a narrative device and was often hard to avoid when eliciting examples.

50

(84) INDEX-3 RIDE.BICYCLE SLOW FAT ‘He rides a bicycle slowly (because he’s) fat.’

(85) INDEX-3 UNCLE RUN VERY.SLOW OLD ALREADY ‘(My) uncle there runs very slowly (because) he is already old.’

(86) GRANDPA SWIM SLOW ‘Grandpa swims slowly.’

In (82) through (86), each of the activity predicates is compatible with SLOW. When signers paraphrased these clauses they tended to drop the adverbial and instead articulate the verb more slowly, indicating it was interpreted as ‘slow’ and not ‘for a long time’. The results indicate both a [+dynamic] and [+durative] value for the predicates.

The examples in (87) through (91) show the predicates with the second test for duration – expressions of temporal extent.

(87) DRIVE.MOTORCYCLE FAR ONE.HOUR ARRIVE SHOP GAS ‘(I) drove on and on for an hour and finally arrived at the gas station.’

(88) GRANDMA PAINT MINUTE-THIRTY ‘Grandma painted for thirty minutes.’

(89) RIDE.BICYCLE ABOUT ONE.HOUR ‘(I) rode a bicycle for about an hour.’

(90) UNCLE STRONG RUN RUNDEPICTING.VERB ONE.HOUR CAN ‘(My) uncle is strong – he can run for an hour!’

(91) SWIM MINUTE-THIRTY ‘(I) swam for thirty minutes.’

As seen in (87) through (91), the predicates DRIVE.MOTORCYCLE, PAINT, RIDE.BICYCLE, RUN, and SWIM, respectively, are compatible with adverbials of temporal extent. This felicitous co-occurrence again indicates they are [+durative]. Section 3.4.3 presents the results from testing the predicates for telicity.

51

3.4.3 Testing for telicity – activities An atelic event does not involve a result state. This factor makes activities felicitous with the fifth test, questions of continuation. The clauses in (92) through (96) show the predicates in this test frame.

(92) INDEX-3 FRIEND DRIVE.MOTORCYCLE INDEX-3 ‘(My) friend there is driving his motorcycle.’ DRIVE.MOTORCYCLE STILL.THERE ‘Is he still driving?’

(93) EASEL GRANDMA PAINT ‘Grandma is painting on an easel.’ PRACTICE STILL.THERE PAINTING ‘Is she still practicing painting?’

(94) INDEX-1 FRIEND RIDE.BICYCLE ‘My friend is riding his bicycle.’ RIDE.BICYCLE STILL.THERE ‘Is he still riding it?’

(95) MORNING FRIEND RUN INDEX ‘In the morning my friend went running over there.’ RUN STILL.THERE ‘Is he still running?’

(96) MORNING UNCLE GO SWIM ‘(This) morning, (my) uncle went swimming.’ PRACTICE4 STILL.THERE ‘Is he still practicing?’

The compatibility of the predicates in (92) through (96) with the question form of the test STILL.THERE indicates they do not include entry into a result state. If they did, the question of continuation would be pragmatically odd. Thus, the predicates are [-telic].

The clauses in (97) through (101) show the clauses with the next test, MANY.TIMES.

4 Although PRACTICE was used in the question instead of SWIM, the reference is to swimming, the same event. It was thus judged to be an acceptable example of the atelic value of SWIM.

52

(97) INDEX-1 DRIVE.MOTORCYCLE MANY.TIMES ALREADY ‘I have driven a motorcycle many times.’

(98) GRANDMA PAINT MANY.TIMES BORED ALREADY ‘Grandma has painted many times and is tired of it.’

(99) INDEX-1 FRIEND RIDE.BICYCLE MANY.TIMES ALREADY ‘My friend has ridden a bicycle many times.’

(100) INDEX-1 1-SEE-3 MORNING UNCLE RUNCONSTRUCTED.ACTION RUN MANY.TIMES ALREADY INDEX-3 ‘I’ve seen that in the morning (my) uncle often goes running.’

(101) SWIM MANY.TIMES ALREADY ‘(I) have swum many times.’

In (97) through (101), the predicates occur felicitously with MANY.TIMES. Further, each iteration of the event is understood to involve the same participant, indicating that no result state is involved and that the events are [-telic].

Table 6 summarizes the results of testing DRIVE.MOTORCYCLE, PAINT, RIDE.BICYCLE, RUN, and SWIM for their dynamic, durative, and telic values.

Table 6 Test results for activities in ThSL

EXTENT

Predicate

TIMES

THERE

.

.

GO INTERRUPT ADVERBIALS PACE TEMPORAL STILL MANY

DRIVE.MOTORCYCLE       PAINT       RIDE.BICYCLE       RUN       WALK        – compatible ? – questionable X – incompatible * – further considerations

As Table 6 shows, the predicates DRIVE.MOTORCYCLE, PAINT, RIDE.BICYCLE, RUN, and WALK are compatible with each of the tests. Their compatibility with command forms, interruption, and the pace adverbial SLOW indicates they are [+dynamic].

53

Their compatibility with SLOW and with temporal extent adverbials further shows they are [+durative]. Finally, the ability to question the continuation of the event using STILL.THERE and to indicate repetitions of the event with MANY.TIMES indicates the predicates are [-telic].

3.5 Accomplishments Accomplishments are dynamic, durative, telic situations. Like other events, the feature of dynamism makes activities compatible with commands and interruption (Tests 1 and 2). Their dynamism and duration further allow them to occur with pace adverbials (Test 3). Their durative value also accounts for their compatibility with adverbials of temporal extent (Test 4). Since accomplishments are telic, they are incompatible with assertions of continuation (Test 5) if the event is portrayed as completed (see Section 1.3.2). Likewise, a telic event cannot repeat (Test 6) unless the result state is first reversed, or a new participant is involved.

Telic predicates include three categories: property scales (e.g., widen, cool), path scales (e.g., cross the road, climb a hill), and incremental theme verbs (e.g., build a boat, eat an orange, destroy a city). However, Rappaport Hovav (2006) argues persuasively that the telicity of clauses with incremental theme verbs is a property of the object of the clause rather than the predicate. Thus, the current study does not include incremental theme verbs. Instead, it focuses on property scale predicates.

Property scale predicates express the degree of change that an object undergoes during an event (Christopher Kennedy & Levin, 2008, p. 156). Examples include widen, cool, grow up, expand, lighten, and dry. However, these predicates are not all telic. Their telicity hinges on the type scale to which they belong.

The scale types are as follows. First, totally closed scales have a minimum and maximum value. For example, fill the tank has a minimal value (an empty tank) and a maximal value (a full tank). Second, a bottom-closed scale only has a minimal value. The predicates tangle and get wet belong to this category. They have the minimal values untangled and dry, respectively, but no maximal values (an entity can get more and more tangled or more and more wet). Third, top-closed scales have maximal values but no minimal values. For example, straighten and dry have upper limits (completely straight, completely dry), but no lower limits. Finally, totally open scales lack both a maximal and minimal value. Predicates like widen and expand belong to this category.

54

Predicates that belong to closed scales may be telic or atelic (Kuhn, 2017, p. 16). Those that fall along completely open scales are atelic unless pragmatic implications create closed readings. Atelic property scale predicates share the same featural values as activities (i.e., they are dynamic, durative, and atelic). However, their internal contour is different since they involve directed movement along a scale. Because of this, Croft (2012, p. 44) calls atelic scalar predicates ‘directed activities’. This term is adopted in the following discussion.

Two of the predicates chosen to represent accomplishments belong to bottom-closed scales. These are BEND.BAR and GROW.BEARD. The events have clear minimal values – completely straight, completely beardless – but lack maximal ones. One predicate, STRAIGHTEN.BAR, is top-closed. It possesses a clear maximal value – perfectly straight. In contrast, the final two predicates, SKY.DARKEN and ICE.MELT, belong to totally- closed scales and have both minimal values (completely light, completely solid) and maximal values (completely dark, completely liquid). No predicates from totally- open scales were selected since these always pattern as atelic events.

Section 3.5.1 presents the results of testing these predicates for dynamism. Section 3.5.2 provides the results for the tests of duration and Section 3.5.3 the telicity test results.

3.5.1 Testing for dynamism – accomplishments Compatibility with GO depends on the semantic quality of volition as well as dynamism. The clauses in (102) through (105) show the results of testing four of the accomplishment predicates with GO.

(102) BEND.BAR GO ‘Go bend the bar!’

(103) GROW.BEARD GO ‘Grow a beard.’

(104) SKY GOD 1-SEE-3 1-COMMAND-3 SKY.DARKEN ‘The sky, God saw it and commanded, ‘Darken!’’

(105) STRAIGHTEN.BAR GO ‘Go straighten the bar!’

55

In (102) through (105), the predicates BEND.BAR, GROW.BEARD, SKY.DARKEN, and STRAIGHTEN.BAR occur felicitously with the command GO, indicating they are [+dynamic]. The remaining predicate, ICE.MELT, is shown with this command in (106).

(106) *ICE ICE.MELT GO ‘Ice, melt!’

The clause in (106) is infelicitous due to the non-volitional nature of ICE. This lack of volition prevents the test from identifying the dynamic value of ICE.MELT. However, the compatibility of the ICE.MELT with SLOW (see Section 3.5.2) demonstrates it is indeed dynamic.

The second test, INTERRUPT, can only occur with accomplishment predicates if the event is portrayed as uncompleted (see Section 1.3.2). When the full movement of the predicate is used, the clause is no longer acceptable. The clause in (107) illustrates this restriction.

(107) *SUPERMAN METAL STRAIGHTEN.BAR INTERRUPT ‘Superman straightened the metal bar when he was interrupted …’

In (107), the signer’s hands complete the movement of straightening a bar. The completed movement indicates the maximal degree of a closed scale has been reached (see Kuhn, 2017, p. 19). Thus, the result state has been reached, and the event can no longer be interrupted.

In contrast, uncompleted portrayals of the event do not include the endpoint. The clauses in (108) to (111) use a shortened form of the verbs, represented by

VERB UNCOMPLETED, that provide an uncompleted portrayal of the event.

(108) SUPERMAN METAL BAR BEND.BAR UNCOMPLETED INTERRUPT ‘Superman was bending the metal bar when he was interrupted by …’

(109) SKY.DARKEN UNCOMPLETED INTERRUPT SUN ‘The sky was darkening when interrupted by the sun shining …’

(110) SUPERMAN METAL BAR STRAIGHTEN.BARUNCOMPLETED INTERRUPT ‘Superman was straightening the metal bar when he was interrupted by …’

(111) ?ICE ICE.MELT UNCOMPLETED INTERRUPT ‘The ice was melting when interrupted by …’

56

In (108) to (111), the movement of the predicate signs BEND.BAR, SKY.DARKEN, STRAIGHTEN.BAR, and ICE.MELT are all cut short. The altered form presents the event as uncompleted, allowing the process leading up to the result state to be interrupted. The altered forms of the first three signs – BEND.BAR, SKY.DARKEN, STRAIGHTEN.BAR – are then compatible with INTERRUPT, indicating they are [+dynamic]. The predicate

ICE.MELTUNCOMPLETED in (111) is only marginally acceptable. The reason for this appears to be pragmatic since informants observed that once ice begins melting, it rarely stops (see Section 4.5.1).

The last predicate, GROW.BEARD, involves movement along a top-open scale and, hence, lacks a maximal value. The clause in (112) shows this final predicate with INTERRUPT.

(112) INDEX-1 GROW.BEARD INTERRUPT INDEX-3 3-MEAN-1 CUT.BEARD ‘I was growing a beard when (the process) was interrupted because someone was mean to me and cut it off.’

In (112), the movement for GROW.BEARD is unmodified. The predicate can occur with INTERRUPT since the scale along which the event progresses (i.e., ‘beard length’) lacks a maximal degree. Thus, the results provide two pieces of information regarding GROW.BEARD. First, its compatibility with INTERRUPT indicates it is indeed [+dynamic]. Second, the fact that it occurs with INTERRUPT in unmodified form indicates it is [-telic]. The results are discussed further in Section 3.5.3. Section 3.5.2 reviews the results of testing the durative value of the predicates.

3.5.2 Testing for duration – accomplishments Since accomplishments are durative, they can occur felicitously with pace adverbials. The clauses in (113) through (117) show the predicates in this test frame.

(113) SUPERMAN METAL BAR BEND.BAR SLOW ‘Superman bent the metal bar slowly.’

(114) INDEX-3 GROW.BEARD SLOW ‘His beard is growing slowly.’

(115) INDEX AMERICA AIR COLD TEMPERATURE.LOW ICE ICE.MELT VERY.SLOW ‘There in America it is really cold, so ice melts very slowly.’

57

(116) STRANGE SKY.DARKEN SLOW ‘That’s strange – night is falling slowly.’

(117) SUPERMAN METAL BAR STRAIGHTEN.BAR SLOW ‘Superman straightened the metal bar slowly.’

In (113) through (117), the predicates occur felicitously with SLOW and VERY.SLOW, indicating they are [+durative]. Since pace adverbials are associated with motion, the results in (113) through (117) also show that the predicates are [+dynamic].

The predicates are shown with the next test, temporal extent adverbials, in (118) through (122).

(118) SUPERMAN METAL BAR BEND.BAR MINUTE-FIVE ‘Superman bent the metal bar in five minutes.’

(119) INDEX-1 GROW.BEARD SLOW FIVE.YEARS ‘I grew my beard slowly over the course of five years.’

(120) SUN HOT ICE ICE.MELT MINUTE-TWO SHORT.TIME ‘When the sun is out and it’s hot, ice melts in two minutes – so quickly!’

(121) SUPERMAN METAL BAR STRAIGHTEN.BAR MINUTE-FIVE ‘Superman straightened the metal bar in five minutes.’

(122) SKY.DARKEN TIME MINUTE-THIRTY ENOUGH ‘The sky will become dark in thirty minutes – it’s enough (time to go for a run).’

The predicates in (118) through (122) also occur felicitously with adverbials of temporal extent, again indicating they are [+durative]. Section 3.5.3 shows the test results for the final featural value, telicity.

3.5.3 Testing for telicity – accomplishments When an accomplishment is presented as a completed event, it implies a result state has been reached. The change of state makes completed accomplishments incompatible with continuation – once the result state is reached, the event can no longer continue. The clauses in (123) through (127) show the predicates with Test 5, compatibility with continuation.

58

(123) *SUPERMAN METAL BAR BEND.BAR STILL.THERE ‘Superman bent the metal bar and is still bending it.’

(124) *ICE ICE.MELT STILL.THERE ‘The ice melted and is still there melting.’

(125) SKY.DARKEN ALREADY ‘The sky has darkened already.’ *SKY.DARKEN STILL.THERE ‘Is it still darkening?’

(126) *SUPERMAN METAL BAR STRAIGHTEN.BAR STILL.THERE ‘Superman straightened the metal bar and is still straightening it.’

(127) INDEX-3 INDEX-1 FRIEND GROW.BEARD ‘My friend grew a beard/has a long beard.’ INDEX-3 STILL.THERE ‘Is it still there (does he still have it)?’

In (123) through (126), the predicates BEND.BAR, ICE.MELT, SKY.DARKEN, and STRAIGHTEN.BAR are incompatible with STILL.THERE, indicating they are [+telic]. These results are only expected for the last three predicates – ICE.MELT, SKY.DARKEN, and STRAIGHTEN.BAR – since each one has a maximal value. With the fourth predicate, BEND.BAR, the maximal value appears pragmatically set at about 90 degrees. This maximum provides an endpoint to the situation. The last predicate GROW.BEARD, shown in (127), seems to lack this pragmatic endpoint, making it compatible with STILL.THERE. The compatibility is expected for scalar predicates that lack a maximal value. It is also possible that the sign GROW.BEARD in (127) is treated as a nominal meaning ‘long beard’. This interpretation would also be compatible with a question of continuation since it relates to the result state (having a beard) rather than the process (growing a beard).

The second test of telicity is the repetition of the event with the same participant(s). Examples (128) through (132) show the predicates with this test.

(128) *SUPERMAN METAL BAR BEND.BAR MANY.TIMES ‘Superman bent the bar many times.’

(129) *GROW.BEARD MANY.TIMES ‘(I) have grown a beard many times.’

59

(130) *ICE ICE.MELT MANY.TIMES ‘The ice melted many times.’

(131) *SKY.DARKEN MANY.TIMES ALREADY ‘The sky has darkened many times already.’

(132) *SUPERMAN STRAIGHTEN.BAR MANY.TIMES ALREADY ‘There, Superman, he’s really skilled. (He’s) straightened the bar many times already.’

As (128) through (132) show, all the predicates are infelicitous with MANY.TIMES. The results indicate a [+telic] value.

If the clause explicitly indicates that the event is reversed and then repeated, it can occur with MANY.TIMES. The clauses in (133) through (136) show examples that the participants provided.

(133) INDEX-3 EXPERIENCE ALWAYS USED.TO ALREADY BEND.BAR FAST BEND.BAR STRAIGHTEN.BAR BEND.BAR STRAIGHTEN.BAR MANY.TIMES ALREADY INDEX-3 ‘He (Superman) is already experienced and used to (bending bars). He bends them quickly. Bend, straighten, bend, straighten – he’s done it many times already.’

(134) MORNING ICE.MELT NIGHT ICE.FREEZE ICE.MELT ICE.FREEZE ICE.MELT STRANGE ‘In the morning, the ice melts. At night it freezes – melts, freezes, melts. It’s strange.’

(135) KNOW TIME SLEEP REST SKY.DARKEN SKY.DARKEN TOO.QUICKLY DAY.PASS SKY.DARKEN SKY.LIGHTEN SKY.DARKEN SKY.LIGHTEN SKY.DARKEN MANY.TIMES ALREADY ‘You know at night when it’s time to sleep? Night falls so quickly. The day goes by and then night falls. Day comes, night falls, day comes, night falls. It keeps happening like that.’

(136) INDEX SUPERMAN SKILLED STRAIGHTEN.BAR BEND.BAR STRAIGHTEN.BAR BEND.BAR MANY.TIMES ALREADY ‘There, Superman, he’s really skilled. (He’s) already straightened a bar many times/straightened many bars.’

60

In (133) through (136), the results of the events BEND.BAR, ICE.MELT, SKY.DARKEN, and STRAIGHTEN.BAR are explicitly reversed. This reversal allows them to occur felicitously with MANY.TIMES. In addition, as in (136), if the clause indicates multiple participants (e.g., multiple bars), the event can also occur with MANY.TIMES. Both results show that the predicates are [+telic].

An exception to this pattern is found with GROW.BEARD. The event does not have to be explicitly reversed to occur with MANY.TIMES, as (137) shows.

(137) GROW.BEARD GROW.BEARD MANY.TIMES ALREADY ‘(My friend) has grown a beard many times.’

In (137), GROW.BEARD is articulated twice, allowing it to occur felicitously with MANY.TIMES. Even though this behavior is distinct from that of the other accomplishments, it is also different than their atelic counterparts, activities. As Section 3.4.3 shows, activity predicates can occur felicitously with MANY.TIMES without reduplication. Thus, ThSL appears to distinguish directed activities like GROW.BEARD from both accomplishments and (undirected) activities in clauses indicating multi-events. Testing additional directed activity predicates in ThSL would affirm or refute this conclusion.

Table 7 shows the results of testing the featural values of BEND.BAR, GROW.BEARD, ICE.MELT, SKY.DARKEN, and STRAIGHTEN.BAR.

Table 7 Test results for accomplishments in ThSL

Predicate

TIMES

THERE

.

.

GO INTERRUPT ADVERBIALS PACE TEMPORALEXTENT STILL MANY

BEND.BAR  *   X X GROW.BEARD       ICE.MELT X* X*   X X SKY.DARKEN  *   X X STRAIGHTEN.BAR  *   X X  – compatible ? – questionable X – incompatible * – further considerations

61

Most of the predicates in Table 7 occur felicitously with command forms and with INTERRUPT, indicating they are [+dynamic]. The exception is ICE.MELT, which is incompatible with commands since it takes a non-volitional subject. It also fails the test INTERRUPT on pragmatic grounds since once ice begins melting, it rarely stops. The compatibility of the predicates with SLOW and VERY.SLOW and with temporal extent adverbials indicates they are [+durative].

In terms of telicity, four of the predicates (BEND.BAR, ICE.MELT, SKY.DARKEN, STRAIGHTEN.BAR) were incompatible with STILL.THERE. They were also incompatible with MANY.TIMES when the same participants were involved. Both sets of results indicate the predicates are [+telic]. The last sign, GROW.BEARD, was compatible with STILL.THERE and MANY.TIMES, reflecting its nature as an open-ended scalar predicate. It is, thus, not an accomplishment but instead falls under Croft’s (2012, p. 44) category of “directed activities.” However, GROW.BEARD did not behave the same way with MANY.TIMES as the activity predicates did (see Section 3.4.3), suggesting that ThSL may treat directed and undirected activities as two distinct, covert lexical categories.

Achievements, the other type of telic event, are instantaneous. Predicates from this category are tested in Section 3.6.

3.6 Achievements Achievements are dynamic, instantaneous, telic situations. Their dynamic value makes them compatible with commands (Test 1) and interruption (Test 2). Their instantaneous value, however, means that when they occur with INTERRUPT, it is the process leading up to the event that is interrupted. Achievements can undergo coerced interpretations when used with pace adverbials (Test 3) or with adverbials of temporal extent (Test 4). The coercion is again a product of their inherently instantaneous nature. Since achievements are telic, they are incompatible with expressions of continuation (Test 5). Their telic value also makes them incompatible with repetition (Test 6) unless the result state is reversed, or a new participant is selected.

However, not all result states are reversible. This fact divides achievements into two subtypes: reversible and irreversible. Reversible events such as turn on the light can be repeated if the result state is reversed (i.e., the light is turned off). Irreversible events like break a window can only be repeated with a new participant. The examples in (138) and (139) illustrate these different behaviors.

62

(138) Zane turned on the porch light four times last night.

(139) *Jackson popped the red balloon four times yesterday.

In (138), the event turned on the porch light is a reversible achievement, making it compatible with the repetitions indicated by four times. It is implied, however, that the light was turned off between each instantiation of the event. In contrast, popped the red balloon in (139) is an irreversible event. The implication of reversal is, thus, unavailable, and the event is incongruous with four times.

The distinctive behaviors of reversible and irreversible achievements are also seen in the testing of ThSL predicates. Two of these predicates – EXPLODE and OPEN.CAN – are irreversible. The remaining predicates - BLOCK (on social media), ERASE (Line), and TURN.OFF.LIGHT – are reversible. Section 3.6.1 describes the results of running these predicates through the tests for dynamism. Section 3.6.2 presents them alongside the tests for duration, and Section 3.6.3 provides the results of testing them for telicity.

3.6.1 Testing for dynamism – achievements Most of the achievement predicates tested involve volitional subjects, allowing them to occur felicitously with the first test of dynamism – compatibility with commands. Examples (140) through (143) show the results.

(140) INDEX-2 SWEETHEART ARGUE 1-REQUEST-2 GO 2-BLOCK-3 ‘Your sweetheart that you argue with, please go block her (on Facebook).’

(141) LINE ERASENOD ‘Erase Line!’

(142) PEPSI OPEN.CAN GO ‘Go open a can of Pepsi!’

(143) LIGHT TURN.OFF.LIGHT GO ‘Go turn off the light!’

As seen in (140) through (143), the predicates BLOCK, ERASE, OPEN.CAN, and TURN.OFF.LIGHT each occur felicitously with commands, indicating they are [+dynamic]. The remaining predicate, EXPLODE, cannot occur with a command if an instrument is not specified. Thus, the clause in (144) is unacceptable. The addition of an instrument in (145) makes the command felicitous.

63

(144) *CAR EXPLODE GO ‘Go explode the car!’

(145) CAR ATTACH.OBJECT.UNDER.CAR EXPLODE GO ‘Go place a bomb under the car and blow it up!’

In (144), EXPLODE is used with GO but no instrument. Informants indicated that this construction was unacceptable since the addressee would not know how to carry out the command (e.g., explosives, gas). In (145), a classifier construction meaning ATTACH.OBJECT.UNDER.CAR is added, supplying an instrument. This construction is then acceptable with GO.

The clauses in (146) through (150) show the predicates with the second test for dynamism, INTERRUPT.

(146) INDEX-1 1-BLOCK-3 INTERRUPT ‘I was interrupted from blocking (my sweetheart) on Facebook by …’

(147) INDEX-1 FRIEND LINE ERASE INTERRUPT ‘My friend was interrupted from erasing Line by …’

(148) EXPLODE INTERRUPT BAD COME ALREADY ‘(I) was interrupted from blowing up (the car) – an enemy was already coming.’

(149) OPEN.CAN INTERRUPT ‘(I) was interrupted from opening the can by …’

(150) MOTHER LIGHT TURN.OFF.LIGHT INTERRUPT ‘Mother was interrupted from turning off the light by …’

The compatibility of the predicates in (146) through (150) with interruption indicates they are [+dynamic]. However, since instantaneous events lack duration, there is no internal process that can be interrupted. Instead, it is the period leading up to the event that is interrupted. The interruption prevents the event itself from occurring (at least temporarily).

The presence of INTERRUPT often caused the articulation of the predicate to be cut short. The modified form is reminiscent of the ‘unrealized inceptive’ identified in ASL by Liddell (1984, p. 262). An unrealized inceptive conveys the meaning ‘was going to, but …’ The uncompleted form of ThSL predicates expresses the same

64

meaning. When this shortened form is used, the sign INTERRUPT can be dropped, and the meaning carried by the modified predicate alone.

Section 3.6.2 presents the results for testing the durative value of the predicates.

3.6.2 Testing for duration – achievements The instantaneous nature of achievements is incompatible with the durative value of pace adverbials like SLOW. Examples (151) through (156) show the ThSL achievement predicates with this test.

(151) INDEX-2 FRIEND INDEX-3 FACEBOOK BLOCK-3 SLOW INDEX-3 DO WHAT ‘Your friend, he’s been slow to block (someone) on Facebook. What’s he doing?’

(152) FRIEND LINE ERASE SLOW ‘(My) friend erased Line slowly (because he couldn’t remember how).’ ‘(My) friend was slow to erase Line (because he was undecided).’

(153) OPEN.CAN SLOW ‘(She) opened the can slowly (so as to not damage her manicured nails).’ ‘(She) was slow to open the can (because she stopped to chat with someone).’

(154) WALK POKE TURN.OFF.LIGHT VERY.SLOW ‘(Grandma) went, poked (the switch) and turned off the lights. She did it very slowly.’

(155) SLOW TURN.OFF.LIGHT TIME ALREADY ‘Why are you so slow? It’s already time to turn off the light!’

(156) *CAR EXPLODE SLOW ‘The car was slow to explode.’

For the four predicates in (151) through (155), SLOW causes a delayed inceptive interpretation of the situation. The shift changes the events into derived-level accomplishments. The delay may be associated with a pre-process (e.g., walking slowly up to the light switch) or to an intentional delay in the execution of the event (i.e., putting it off). The context determines which interpretation is selected. The fact that SLOW cannot modify the events themselves indicates the predicates are [-durative].

65

The last predicate EXPLODE in (156) is unacceptable with SLOW. The difference may again be due to the absence of an instrument in the clause. Alternatively, it may be semantic: the pre-process for EXPLODE is conceptually more distinct from the event than the pre-process of other predicates since it involves two agents – the placer of the bomb and the bomb itself.

The instantaneous nature of achievements also conflicts with Test 4, temporal extent adverbials. The clauses in (157) through (160) show the predicates in this test frame.

(157) *FRIEND LINE ERASE MINUTE-FIVE ‘My friend erased Line for five minutes.’

(158) *CAR EXPLODE MINUTE-FIVE ‘The car exploded for five minutes.’

(159) *OPEN.CAN MINUTE-FIVE ‘(I) opened the can for five minutes.’

(160) *MOTHER LIGHT TURN.OFF.LIGHT MINUTE-FIVE ‘Mother turned off the light for five minutes.’

The examples in (157) through (160) show the incompatibility of ERASE, EXPLODE, OPEN.CAN, and TURN.OFF.LIGHT with temporal extent adverbials. The results indicate the predicates are [-durative].

In some instances, however, temporal extent adverbials can coerce the predicates into accomplishments by associating them with preliminary stages. The clauses in (161) and (162) illustrate this phenomenon.

(161) INDEX-3 OPEN.CAN MINUTE-FIVE PROBLEM WHAT

INDEX-3 LONG.NAILS OPEN.CANCAREFULLY ‘She took five minutes to open the can! What was the problem? Long nails – she opened the can carefully (in order not to break them).’

(162) 1-BLOCK-3 TIME MINUTE-FIVE WHY VERY.SLOW ‘He’s been (trying to) block (someone) on (Facebook) for five minutes. Why (so long)? He’s so slow!’

The presence of MINUTE-FIVE in (161) with OPEN.CAN and in (162) with BLOCK creates delayed inceptive interpretations of the events. The interpretation aligns with the

66

principle of external override since the meaning of the adverbial overrides the instantaneous value of the predicate. The change creates a derived-level accomplishment.

3.6.3 Testing for telicity – achievements Achievements end in a result state. This makes them incompatible with Test 5, STILL.THERE. The clauses in (163) through (167) illustrate this clash.

(163) FRIEND 1-BLOCK-3 ALREADY ‘(I) have blocked my friend on Facebook.’ *1-BLOCK-3 STILL.THERE ‘Are you still blocking them?’

(164) FRIEND LINE ERASE ‘(My) friend erased Line.’ *ERASE STILL.THERE ‘Is he still erasing it?’

(165) CAR EXPLODE ‘The car exploded.’ *EXPLODE STILL.THERE ‘Is it still exploding?’

(166) MOTHER OPEN.CAN ‘Mom opened the can.’ *OPEN.CAN STILL.THERE ‘Is she still opening it?’

(167) MOTHER TURN.OFF.LIGHT ‘Mom turned off the lights.’ *TURN.OFF.LIGHT STILL.THERE ‘It she still turning it off?’

In (163) through (167), the questions of continuation are semantically odd. The incompatibility of the question with the statements indicates that the predicates are [+telic].

67

The distinction between reversible and irreversible achievements is seen in their behavior with the final test, MANY.TIMES. The irreversible achievements EXPLODE and OPEN.CAN are shown with this test in (168) and (169).

(168) *CAR CL.CAR EXPLODE MANY.TIMES ‘The car exploded many times.’

(169) *INDEX-1 OPEN.CANX OPEN.CANX MANY.TIMES ALREADY ‘I have opened the can many times.’

In (167) the addition of the classifier CL.CAR indicates a specific referent, ‘the car’. The presence of a specific referent makes EXPLODE incongruous with MANY.TIMES, indicating that EXPLODE is [+telic]. In (169), the sign OPEN.CAN is signed twice in the same location in signing space, which also indicates a specific referent. The specification of the referent makes the predicate incompatible with MANY.TIMES and reveals that it is also [+telic].

The reversible achievements BLOCK and ERASE are shown with MANY.TIMES in (170) through (173).

(170) *FACEBOOK 1-BLOCK-3 MANY.TIMES ALREADY ‘I have blocked him on Facebook many times already

(171) *INDEX-1 FRIEND LINE ERASE MANY.TIMES ‘My friend erased Line many times …’

(172) HEY INDEX-2 1-BLOCK-3 ALREADY ‘Hey, did you block him on Facebook already?” INDEX-1 YES NOW FACEBOOK 1-BLOCK-3 1-BLOCK-3 1-BLOCK-3 MANY.TIMES ALREADY ‘Yes, now I have already blocked him on Facebook many times.’

(173) INDEX-1 FRIEND LINE ERASE ERASE MANY.TIMES ‘My friend erased Line many times.’

The clauses in the first two examples, (170) and (171), are ill-formed due to the single iteration of the predicates BLOCK and ERASE. The results indicate a [-telic] value. However, when the predicates are reduplicated in (172) and (173), they then occur felicitously with MANY.TIMES. In contrast, the irreversible event OPEN.CAN in (169) was incompatible with MANY.TIMES even when it was reduplicated. These

68

contrastive results indicate that ThSL distinguishes reversible versus irreversible achievements.

In (174) the final predicate, TURN.OFF.LIGHT, is tested with MANY.TIMES.

(174) *SOHM LIGHT TURN.OFF.LIGHT MANY.TIMES ‘Sohm turned off the light many times …’

The single iteration of TURN.OFF.LIGHT in (174) is incompatible with the multi-event implication of MANY.TIMES, which makes the clause unacceptable. However, the predicate cannot merely be reduplicated to resolve the incongruity. This limitation is due to the phonological form of the sign. A single iteration is signed by holding the dominant hand at about shoulder height, with the fingers outspread and the palm facing down. The fingers are then closed so that the tips come together (see still shots in Appendix A). If the sign is repeated, the fingers must first re-open. However, this opening movement is the articulation for the sign TURN.ON.LIGHT. By repeating the sign then, the reversal of the state is explicitly reversed (i.e., ‘turn off light, turn on light …’). This makes the predicate compatible with MANY.TIMES, as (175) shows.

(175) TURN.OFF.LIGHT TURN.ON.LIGHT TURN.OFF.LIGHT TURN.ON.LIGHT MANY.TIMES ‘The lights flashed many times.’

The results in (168) through (176) identify several factors that affect the compatibility of achievements with MANY.TIMES. First, if the achievement is irreversible, it is incompatible with MANY.TIMES. The incompatibility identifies these predicates as [+telic]. Reversible achievements, on the other hand, can occur with MANY.TIMES if the predicate is reduplicated. Pragmatically, however, it is implied in these examples that the result state is reversed in between each instantiation of the event, again indicating the event is [+telic]. Finally, for some predicates, the reversal of the state is explicitly expressed due to phonetic restrictions, as seen with the sign TURN.OFF.LIGHT.

Table 8 summarizes the results of testing the predicates BLOCK, ERASE, EXPLODE, OPEN.CAN, and TURN.OFF.LIGHT.

69

Table 8 Test results for achievements in ThSL

TIMES THERE

Predicate .

.

GO INTERRUPT PACE ADVERBIALS TEMPORAL EXTENT STILL MANY

BLOCK (on FB)  * * * X * ERASE (Line)  * * X X * EXPLODE X * X X X X OPEN.CAN  * * * X X TURN.OFF.LIGHT  * * X X *  – compatible ? – questionable X – incompatible * – further considerations

In Table 8, the compatibility of most of the predicates with GO and INTERRUPT indicates they are [+dynamic]. The exception is EXPLODE which, since the subject semantically lacks volition, cannot occur directly with GO. The compatibility of EXPLODE and the other predicates with INTERRUPT, however, confirms they are all [+dynamic]. When used with INTERRUPT, however, the clause indicates that the process leading up to the event is interrupted, not the event itself. For duration, the predicates generally occurred felicitously with pace adverbials. However, their presence created a delayed inceptive interpretation of the situation. The only predicate that did not allow this interpretation was EXPLODE. It was unacceptable with SLOW. The results indicate that the events are [-durative]. All the predicates were incompatible with adverbials of temporal extent, further indicating they are [-durative]. Instances where the predicates were acceptable with these adverbials revolved around coerced interpretations of the event involving preliminary stages.

None of the events were compatible with STILL.THERE, indicating they are [+telic]. The irreversible achievements EXPLODE and OPEN.CAN also did not occur with the final test, MANY.TIMES, when a single participant was involved. These results also indicates a [+telic] value. When reversible achievements were reduplicated, they were compatible with MANY.TIMES. Their compatibility was due to the assumption that the result state was reversed between each occurrence of the event. Finally, for phonetic reasons, reduplication of TURN.OFF.LIGHT formed the sign LIGHT.FLASH. This led to an explicit reversal of the state if the sign was repeated, making it compatible with MANY.TIMES.

Similar patterns of reduplication were also found with semelfactive predicates. These are discussed in Section 3.7.

70

3.7 Semelfactives Semelfactives are dynamic, instantaneous, atelic situations. Like other events, their dynamic value makes them compatible with commands (Test 1) and interruption (Test 2). However, since semelfactives are instantaneous, INTERRUPT affects the preliminary stages of the event, not the event itself. The instantaneous value of semelfactives also makes them incongruous with pace adverbials (Test 3) and adverbials of temporal extent (Test 4). Semelfactive events do not have a result state. Thus, they are compatible with expressions of continuation (Test 5) and repetition (Test 6). However, as instantaneous events, the situations can only continue by repetition. Unlike telic instantaneous events, each repetition can involve the same participant without any implication of reversal.

The ThSL predicates chosen for the study are CLAP, KNOCK.ON.DOOR, PERSON.DUNK (under water), PUNCH.PERSON, and SHOUT. Section 3.7.1 demonstrates their dynamic value, Section 3.7.2 their durative value, and Section 3.7.3 their telic value.

3.7.1 Testing for dynamism – semelfactives As dynamic situations, semelfactives are compatible with Test 1, commands. The clauses in (176) through (180) show the predicates in this test frame.

(176) SOHM MODEL CLAPNOD ‘Sohm is modeling, clap (for her)!’

(177) TIME ALREADY KNOCK.ON.DOOR GO ‘It’s time already, go knock (on the patient’s door)!’

(178) MOMENT PERSON.DUNK CAN MOMENT ‘It’s only for a moment – you can dunk! It’s just a moment!’

(179) INDEX-2 GO PUNCH.PERSON GO ‘You go punch that person!’

(180) SHOUT GO ‘Go on, shout!’

71

The predicates in (176) through (180) are each compatible with commands, indicating they are [+dynamic].5

In (181) through (185) the predicates are shown with the second test for dynamism, INTERRUPT.

(181) CLAP INTERRUPT ‘(I) was interrupted from clapping by …’

(182) NURSE HOSPITAL KNOCK.ON.DOOR INTERRUPT ‘The hospital nurse was interrupted from knocking on the door by …’

(183) PERSON.DUNKUNREALIZEDINCEPTIVE INTERRUPT WHAT CRAB ‘(My friend) was interrupted from dunking by a crab (pinching his toe).’

(184) PUNCH.PERSON INTERRUPT ‘I was interrupted from punching him by …’

(185) INDEX-1 PERSEVERE SHOUTUNREALIZEDINCEPTIVE INTERRUPT INDEX-2 EAT ‘I was persevering and about to shout when interrupted (by someone asking), ‘Would you like to eat?’’

As (181) through (185) show, each of the predicates occurs felicitously with INTERRUPT, again indicating they are [+dynamic]. Most of the predicates in the examples occurred in their basic form or with an unrealized inceptive form, as in (183) with PERSON.DUNK. Since the events are instantaneous, INTERRUPT refers to the stages leading up to the event, not the event itself. This interpretation indicates the events are [-durative] as well as [+dynamic]. The tests in Section 3.7.2 further demonstrate their durative value.

3.7.2 Testing for duration – semelfactives The instantaneous nature of semelfactives and the durative implications of pace adverbials (Test 3) are incompatible. The clauses in (186) through (190) show the predicates with this test frame.

5 In (178) PERSON.DUNK occurs with CAN rather than a command. Since this sign carries the same implication of volition and, hence, dynamism, it is accepted here as a demonstration of the dynamism of PERSON.DUNK.

72

(186) INDEX-1 FRIEND PERSON.DUNK SLOW ‘My friend dunked slowly.’ ‘My friend was slow to dunk.’

(187) INDEX-3 PUNCH.PERSON SLOW ‘He punched (someone) slowly.’ ‘He was slow to punch (someone).’

(188) SHOUT SLOW ‘(He) has shouted for a long time (it is part of his nature).’ ‘(He) was slow to shout (because he is old).’

(189) CLAP SLOW ‘(I) clapped slowly.’ ‘(I) was slow to clap.’

(190) NURSE HOSPITAL KNOCK.ON.DOOR SLOW ‘The hospital nurse knocked slowly on the door.’ ‘The hospital nurse was slow to knock on the door.’

In (186) through (190) the use of SLOW with the selected predicates causes a delayed inceptive interpretation. The assumed delay may be attributed to delayed execution or slow execution of the event. In one instance, (188), the meaning of the adverbial shifted to ‘for a long time’ and the event was interpreted as a dispositional predicate (i.e., ‘it is his nature to shout’). The results reflect a clash between the meaning of SLOW and the [-durative] nature of the predicates. The examples present another instance where compatibility is a product of the featural value of the adverbial overriding the value of the predicate (see Smith, 1997, pp. 45, 53).

The durative value of adverbials of temporal extent (Test 4) can create iterative interpretations of semelfactive events. To block this interpretation, the sign ONE.TIME was added to this test frame when it was used with semelfactives. The examples in (191) through (195) illustrate the resulting clauses.

(191) *FRIEND CLAP ONE.TIME MINUTE-FIVE ‘(My) friend clapped one time for five minutes.’

(192) *HOSPITAL NURSE KNOCK.ON.DOOR ONE.TIME MINUTE-FIVE ‘The nurse knocked on the door one time for five minutes.’

73

(193) *UNCLE PERSON.DUNK ONE.TIME MINUTE-FIVE ‘(My) uncle dunked one time for five minutes.’

(194) *BULLY PUNCH.PERSON ONE.TIME MINUTE-FIVE ‘The bully punched (someone) one time for five minutes.’

(195) *WOMAN SHOUT ONE.TIME MINUTE-FIVE ‘The woman shouted one time for five minutes.’

The clauses in (191) through (195) show the incompatibility of the predicates with expressions of temporal extent. The results reflect the [-durative] nature of the predicates.

3.7.3 Testing for telicity – semelfactives Semelfactive events terminate immediately after they begin. In basic form, then, semelfactives are incompatible with the implication of continuation. However, if an iterative interpretation of the event is allowed, the event can be questioned with is x still there doing y? The clauses in (196) through (200) show the predicates with this test frame.

(196) FRIEND WHERE ‘Where is your friend?’ CLAP (CLAP) STILL.THERE ‘He is still over their clapping.’

(197) INDEX-3 NURSE KNOCK.ON.DOOR ‘There the nurse knocked on the door.’ KNOCK.ON.DOOR STILL.THERE ‘Is she still knocking?’

(198) HEY CHILD BOY WHERE ‘Hey, where is the little boy?’ INDEX PLAY PERSON.DUNK (PERSON.DUNK) STILL.THERE PLAY STILL.THERE ‘(He’s) playing over there, dunking (in the water). He’s still playing.’

74

(199) INDEX-3 FRIEND PUNCH.PERSON ‘There (my) friend punched someone.’ PUNCH.PERSON STILL.THERE ‘Is he still punching them?’

(200) HEY INDEX SHOUT ‘Hey, over there (someone) shouted.’ HEY SHOUT STILL.THERE ‘Oh! Are they still shouting?’

In (196) through (200), each of the predicates are compatible with questions of continuation, indicating they are [-telic].

The atelic value of semelfactives also makes them repeatable. However, since repetition involves a series of events, a single iteration of the predicate is unacceptable, as (201) through (205) show.

(201) *CLAP MANY.TIMES ALREADY ‘(I) have clapped many times already.’

(202) *KNOCK.ON.DOOR MANY.TIMES ALREADY ‘(I) have knocked on the door many times already.’

(203) *INDEX-1 FRIEND PERSON.DUNK MANY.TIMES ‘My friend dunked many times.’

(204) *INDEX-1 PUNCH.PERSON MANY.TIMES ‘I punched (someone) many times.’

(205) *INDEX-3 FRIEND SHOUT MANY.TIMES ‘(My) friend there shouted many times.’

The single iteration of the predicates in (201) through (205) is incompatible with the multi-event implication of MANY.TIMES. Reduplicating the predicates, as done in (206) through (210), resolves the ill-formedness.

(206) CLAP CLAP MANY.TIMES ALREADY ‘(I) have clapped many times.’

(207) KNOCK.ON.DOOR KNOCK.ON.DOOR MANY.TIMES ALREADY ‘(I) have knocked on the door many times.’

75

(208) INDEX-1 FRIEND PERSON.DUNK PERSON.DUNK MANY.TIMES ‘My friend has dunked many times.’

(209) PUNCH.PERSON PUNCH.PERSON MANY.TIMES ALREADY AGITATED ENDURE NEG ‘I punched him many times already. I was agitated and couldn’t take it anymore.’

(210) INDEX-3 FRIEND SHOUT SHOUT MANY.TIMES ‘(My) friend there shouted many times.’

The predicates in (206) through (210) are each reduplicated, allowing a multi-event reading with MANY.TIMES. The participants in the clauses remain the same, indicating that no result state is involved. The results show the [-telic] value of the predicates.

Table 9 displays the results for testing CLAP, KNOCK.ON.DOOR, PERSON.DUNK, PUNCH.PERSON, and SHOUT.

Table 9 Test results for semelfactives in ThSL

Predicate

TIMES

THERE

.

.

GO INTERRUPT ADVERBIALS PACE TEMPORALEXTENT STILL MANY

CLAP  * * X  * KNOCK.ON.DOOR  * * X  * PERSON.DUNK  * * X  * PUNCH.PERSON  * * X  * SHOUT  * * X  *  – compatible ? – questionable X – incompatible * – further considerations

As Table 9 shows, the predicates CLAP, KNOCK.ON.DOOR, PERSON.DUNK, PUNCH.PERSON, and SHOUT are compatible with GO and INTERRUPT, showing they are [+dynamic]. For INTERRUPT, the preliminary stages of the event are interrupted, reflecting the [-durative] nature of the event. Likewise, even though the predicates occurred with SLOW, the adverbial caused delayed inceptive interpretations meaning ‘slow to do x’ or ‘did x slowly’. None of the predicates could occur with temporal extent

76

adverbials, showing they are [-durative]. The sign ONE.TIME was added to this test frame to block an iterative interpretation of the event. The predicates occurred felicitously with questions of continuation, indicating they are [-telic]. This conclusion was further affirmed by their compatibility with MANY.TIMES. However, predicates had to be reduplicated to occur felicitously with this test since it indicates a multi-event.

3.8 Summary of results This section provides a summary of the behavior of ThSL predicates with the selected test frames. Instead of presenting the results according to situation type, they are presented by test. This arrangement provides a different view of the data – one that captures the general behavior of predicates from each situation type with each test. The results highlight the fact that the temporal features of ThSL predicates are covert lexical categories.

The first two tests identified the dynamic value of predicates. Table 10 summarizes the behavior of the predicates with Test 1, compatibility with commands.

Table 10 Compatibility of ThSL predicates with commands (Test 1)

Situation type Compatibility with commands Feature value Individual-level states Incompatible [-dynamic] Stage-level states Incompatible [-dynamic] Activities Compatible [+dynamic] Accomplishments Compatible [+dynamic] Achievements Compatible [+dynamic] Semelfactives Compatible [+dynamic]

As Table 10 shows, neither stage-level nor individual-level states are compatible with commands, indicating they are [-dynamic]. In contrast, the activity, accomplishment, achievement, and semelfactive predicates were compatible with this test, indicating they are [+dynamic]. Exceptions to this pattern were the result of semantically non-volitional arguments rather than a stative value of the predicate.

The second test, compatibility with INTERRUPT, reflects the dynamic value of the predicate. Table 11 shows the results of testing the predicates with INTERRUPT.

77

Table 11 Compatibility of ThSL predicates with INTERRUPT (Test 2)

Situation type Compatibility with INTERRUPT Feature value Individual-level states Incompatible [-dynamic] Stage-level states Incompatible [-dynamic] Activities Compatible [+dynamic] The event itself is interrupted [+durative] Accomplishments Compatible [+dynamic] The event itself is interrupted [+durative] Achievements Compatible [+dynamic] Preliminary stages are interrupted [-durative] Semelfactives Compatible [+dynamic] Preliminary stages are interrupted [-durative]

As Table 11 shows, stative predicates were incompatible INTERRUPT, indicating they are [-dynamic]. Exceptions involved coerced interpretations of the states as dynamic events, such as ‘thinking sad thoughts’ instead of ‘being sad’. However, even these coerced examples were only marginally acceptable.

Activities and accomplishments were compatible with the test, as long as the accomplishments were portrayed as uncompleted. The compatibility of these predicates with interrupt indicates they are [+dynamic]. The interruption of the event itself also indicates that they are [+durative].

In contrast, when achievement and semelfactive predicates were tested with INTERRUPT, the resulting clauses indicated that the preliminary stages of the event was interrupted. The predicates felicitous cooccurrence with the test indicates they are [+dynamic]. The fact that the preliminary stages are interrupted, not the event itself, indicates they are also [-durative].

The results from the third test, pace adverbials, are listed in Table 12.

Table 12 Compatibility of ThSL predicates with pace adverbials (Test 3)

Situation type Compatibility with pace adverbials Feature value Individual-level states Incompatible with ‘slow’ [-dynamic] Incompatible with ‘for a long time’ Individual-level state Stage-level states Incompatible with ‘slow’ [-dynamic] Compatible with ‘for a long time’ Stage-level state

78

Situation type Compatibility with pace adverbials Feature value Activities Compatible [+durative] [+dynamic] Accomplishments Compatible [+durative] [+dynamic] Achievements Causes a delayed inceptive interpretation [-durative] of the event Semelfactives Causes a delayed inceptive interpretation [-durative] of the event

As seen in Table 12, when pace adverbials were used with states, they were interpreted to mean ‘for a long time’. The results reflect the [-dynamic] nature of stative predicates, which is incompatible with pace modifiers. Both types of stative predicates caused the adverbials to be interpreted as ‘for a long time’. However, only stage-level states, which are transitory, were compatible with this new adverbial. Thus, the test served to distinguish stage-level from individual-level states.

Durative, dynamic predicates (activities and accomplishments) were compatible with the test since they involve motion and have multiple stages. In contrast, while instantaneous events (achievements and semelfactives) also involve motion, they consist of a single stage. Thus, the predicates were interpreted as delayed inceptives when placed in this test frame.

The results from Test 4, compatibility with temporal extent adverbials, are shown in Table 13.

Table 13 Compatibility of ThSL predicates with temporal extent (Test 4)

Situation type Compatibility with temporal extent Feature value Individual-level states Incompatible Individual-level state Stage-level states Compatible Stage-level state Activities Compatible [+durative] Accomplishments Compatible [+durative] Achievements Incompatible [-durative] Semelfactives Incompatible [-durative]

As Table 13 shows, the fourth test produced relatively straight-forward results. First, the test was incompatible with individual-level states, since these indicate a long- term property of a participant. In contrast, the temporary nature of stage-level states made them felicitous with the test. Activity and accomplishment predicates were

79

also compatible, indicating they are [+durative]. In contrast, the instantaneous nature of achievements and semelfactives made them incompatible with the test. Cases in which achievement predicates did occur felicitously involved coerced interpretations of the events as delayed inceptives.

Tests 5, STILL.THERE, was designed to identify the telic value of dynamic events. The results from this test are summarized in Table 14.

Table 14 Compatibility of ThSL predicates with STILL.THERE (Test 5)

Situation type Compatibility with STILL.THERE Feature value Activities Compatible [-telic] Accomplishments Incompatible [+telic] Achievements Incompatible [+telic] Semelfactives Compatible [-telic]

As Table 14 shows, the activity predicates are compatible with questions of continuation, indicating they are [-telic]. In contrast, neither accomplishment nor achievement predicates occurred felicitously with STILL.THERE, indicating they are [+telic]. Semelfactives were also felicitous with STILL.THERE if the predicate was repeated to allow an iterative interpretation. The repetition requirement reflects the [-durative] nature of the predicates. However, the same participant was involved in each instantiation of the event, indicating the semelfactives predicates are indeed [-telic].

The final test was compatibility with MANY.TIMES. Although the test does not apply to states as a test for telicity, it did serve to distinguish stage-level from individual- level states. Thus, it was also applied to stative predicates. The results are summarized in Table 15.

Table 15 Compatibility of ThSL predicates with MANY.TIMES (Test 6)

Situation type Compatibility with MANY.TIMES Feature value Individual-level states Incompatible Individual-level state Stage-level states Compatible Stage-level state Activities Compatible [-telic] Accomplishments Incompatible [+telic] Achievements Incompatible [+telic] Semelfactives Compatible [-telic]

80

As Table 15 shows, the enduring nature of individual-level states made them incompatible with MANY.TIMES. Stage-level states, which express transitory properties, were compatible, although some had to be reduplicated due to the multi- event implication of the test.

For events, both activities and semelfactives were felicitous with MANY.TIMES, indicating they are [-telic]. The semelfactive predicates had to be repeated to allow iterative interpretations. Accomplishment predicates were incompatible with MANY.TIMES, indicating they are [+telic]. Irreversible achievements were also incompatible with the test, showing that they are [+telic]. Reversible achievements occurred with MANY.TIMES under the implication that the result state was reversed between each instantiation of the event.

81

Chapter 4 Extended Testing

4.1 Overview Twelve of the 30 predicates tested with the Original Group (see Chapter 3) were tested with two new groups, Group A and Group B. Each group had four Deaf participants. In order to easily indicate which data came from which group, the examples each have two data points. For example, (211a) has the data from Group A and (211b) has the data from Group B.

The results are again presented in order of situation type, with individual-level states covered in Section 4.2, stage-level states in Section 4.3, activities in Section 4.4, accomplishments in Section 4.5, achievements in Section 4.6, and semelfactives in Section 4.7. Each section has two parts. In the first part, the data from Groups A and B is presented. In the second part, it is compared with the results of the Original Group. A general summary of findings is provided in Section 4.8.

The predicates selected for extended testing were ones that 1) behaved as central- level exemplars of situation types in the first round of testing, and 2) were used by the Deaf in Chiang Mai. Table 16 shows the selected predicates.

Table 16 Predicates used in extended testing of situation types in ThSL

Situation type represented ThSL Predicates used for extended testing

States – individual SHORT, SMART States – stage ANGRY, SICK Activities DRAW/PAINT, RUN Accomplishments BEND.BAR, ICE.MELT Achievements ERASE, OPEN.CAN Semelfactives PUNCH.PERSON, SHOUT

Among the 12 central-level predicates shown in Table 16, five differed to some degree between the signers in Bangkok (the Original Group) and those in Chiang Mai (Groups A and B). First, the sign DRAW was used for both ‘paint’ and ‘draw’ by Group A. Since the form and concept of the two signs were judged to be close enough for comparison, DRAW was used in place of PAINT for this group. Second,

82

Group A used the sign LOCK instead of BLOCK to express the meaning ‘block on Facebook’. Since the forms are quite different (e.g., BLOCK is an agreeing verb, LOCK is not), the predicate was replaced with OPEN.CAN for extended testing. Third, among Groups A and B, the sign ERASE did not mean ‘uninstall’ when used with LINE but instead ‘erase message’. Since the form of the sign was the same, it was still used. Fourth, for some informants, the pinky of the non-dominant hand in the sign PUNCH.PERSON was not extended. Since this is an allophonic variation of the person classifier in ThSL, the sign was still used for extended testing. Finally, although Groups A and B used the sign ICE.MELT, they often replaced it in discussions with a predicate that shows an ice cube shrinking rather than turning to water. For this chapter, data using the original variation was preferred.

4.2 Individual-level states An individual-level state expresses a long-term property of a participant. Two predicates from this situation type – SHORT and SMART – were used in extended testing. The results are presented in Section 4.2.1, then compared to the results of the Original Group in Section 4.2.2.

4.2.1 Testing featural values for individual-level states The predicates for individual-level states were first tested with commands to determine their dynamism. The clauses in (211) and (212) show the judgments of Groups A and B.

(211) a. *SHORT GO ‘Go be short!’

b. *SHORT GO ‘Go be short!’

(212) a. SMART GO ‘Go become smart!’ ‘You smart ones, go (to the competition)!’

b. INDEX-2 SMART GO ‘Go become smart!’ ‘You are smart, go do it!’

83

As (211a) and (211b) show, SHORT could not be used with a command, indicating it is [-dynamic]. Likewise, SMART was only marginally acceptable with commands, as (212a) and (212b) show. For Group A, the command led to an inchoative interpretation ‘become smart’ or nominal interpretation ‘smart ones’. For Group B, the command was interpreted either as an inchoative ‘go be smart’ or as two separate clauses, ‘you are smart’ and ‘go do it’. Neither group, however, allowed a static interpretation of SMART with GO, indicating it is [-dynamic].

The clauses in (213) and (214) show the predicates with the second test, INTERRUPT.

(213) a. *FRIEND SHORT INTERRUPT ‘(My) friend’s being short was interrupted by …’

b. *FRIEND SHORT INTERRUPT ‘(My) friend’s being short was interrupted by …’

(214) a. *FRIEND SMART INTERRUPT ‘(My) friend’s being smart was interrupted by …’

b. *SMART INTERRUPT ‘(His) being smart was interrupted by …’

In both (213) and (214) the predicates are incompatible with interruption. The results again indicate that the predicates are [-dynamic].

The predicates were next tested with pace adverbials. The test primarily determines the durative value of events. However, it also identifies the dynamic value of predicates since only events are compatible with the implication of motion. The predicates are shown with this test frame in (215) and (216).

(215) a. ?SHORT VERY.SLOW ‘(He) has been short for a long time.’

b. ?SHORT VERY.SLOW ‘The short one is very slow.’

(216) a. *SMART SLOW ‘(He) is smart slowly.’

b. *SMART VERY.SLOW ‘(He) is smart slowly.’

84

As (215a) shows, when SHORT was tested with VERY.SLOW, Group A interpreted the adverbial as ‘for a long time’. When tested with Group B, as seen in (215b), the predicate was understood as a nominal, ‘the short one’. Both results reflect the incompatibility of stative predicates with pace adverbials and indicate SHORT is [-dynamic].

In contrast, and as seen in (216a) and (216b), neither the meaning of the adverbial nor the predicate shifted when SMART was used with this test. The clauses were simply deemed unacceptable. The judgments reflect the [-dynamic] nature of SMART.

The predicates are shown with Test 4, adverbials of temporal extent, in (217) and (218).

(217) a. SHORT TWO.YEAR ‘(He) has been that short for 2 years.’

b. *INDEX-3 FRIEND SHORT TWO.YEAR ‘That friend was short for two years.’

(218) a. SMART LONG.TIME ALREADY FIVE.YEARS ‘(He) has been smart for a long time already – five years!’

b. *INDEX-3 FRIEND SMART TWO.YEARS 'That friend was smart for two year.’

For Group A, both SHORT in (217a) and SMART in (218a) were acceptable with temporal extent adverbials, indicating they are [+durative]. The results are unexpected since individual-level states are generally incompatible with delimiting time frames. However, Group A created contexts for the clauses that gave them the meaning ‘has been and still is x’. These contexts removed the delimiting implication that is usually present with temporal extent adverbials.

With Group B, this context was not created. Thus, the clauses in (217b) and (218b) were deemed unacceptable. However, informants pointed out that the clause would be acceptable if an object was added. This is shown in (219).

(219) SIGN.LANGUAGE SMART THREE.YEARS ‘She learned sign language in three years.’

The addition of the object ‘sign language’ in (219) causes an inchoative interpretation of the event (e.g., ‘she learned to sign in three years’).

85

The results from this test point to an underlying conflict between the natures of SMART and SHORT and the meaning of temporal extent adverbials. Such a conflict is expected for individual-level states.

Both SHORT and SMART were also tested with MANY.TIMES. The clauses in (220) and (221) show the results.

(220) a. *INDEX-1 FRIEND SHORT MANY.TIMES ‘(My) friend has been short many times.’

b. *INDEX-1 FRIEND SHORT MANY.TIMES ‘(My) friend has been short many times.’

(221) a. *INDEX-1 FRIEND SMART MANY.TIMES ‘My friend was smart many times.’

b. *INDEX-1 FRIEND SMART MANY.TIMES ‘My friend was smart many times.’

As seen in (220) and (221), both groups indicated that SHORT and SMART were unacceptable with MANY.TIMES. The results reflect the long-term nature of individual- level states, which makes them incompatible with the implication of repetition.

4.2.2 Comparison of results for individual-level states The results from testing SHORT and SMART across groups are summarized in Table 17.

Table 17 Comparison of results for individual-level states in ThSL

Test Predicate Original Group Group A Group B Commands SHORT X X X SMART X X X INTERRUPT SHORT X X X

SMART X X X Pace Adverbials SHORT X X X SMART X X X Temporal Extent SHORT X  X SMART X  X STILL.THERE SHORT n/a n/a n/a

SMART n/a n/a n/a

86

Test Predicate Original Group Group A Group B MANY.TIMES SHORT X X X SMART X X X  – compatible ? – questionable X – incompatible * – further considerations

As Table 17 indicates, the results of testing SHORT and SMART were similar across test groups. Each group indicated the predicates were incompatible with commands, interruption, and pace adverbials, revealing that they are [-dynamic].

The groups differed in their judgments with Test 4 – adverbials of temporal extent. With the Original Group, a relatively short time frame (‘one week’) was used. This time frame was incongruous with the persistent nature of SHORT and SMART and created an inchoative – but still infelicitous – interpretation. With Groups A and B, a time frame of several years was used instead. Group A created contexts for the clauses that prevented a temporally delimiting interpretation (i.e., the contexts indicated that the state still held). Thus, the constructions were acceptable. In Group B, this kind of context was not created, and the adverbial was interpreted as a temporal boundary (i.e., the state no longer held). This interpretation made the clauses infelicitous.

The results, although different in their details across groups, all reflect the long-term nature of the predicates as individual-level states. Their incompatibility with MANY.TIMES points to the same conclusion.

4.3 Stage-level states Unlike individual-level states, stage-level states express temporary properties. The results of testing two such states – ANGRY and SICK – with Groups A and B are presented in Section 4.3.1. The results are compared with those of the Original Group in Section 4.3.2.

4.3.1 Testing featural values for stage-level states The predicates are shown with the first test of dynamism, compatibility with commands, in (222) and (223).

(222) a. *ANGRY GO ‘Go be angry!’

87

b. *ANGRY GO ‘Go be angry!’

(223) a. *SICK GO ‘Go be sick!’

b. *SICK GO ‘Go be sick!’

As expected, neither group accepted ANGRY in (222) or SICK in (223) with commands. The results indicate the predicates are [-dynamic].

In (224) and (225) the predicates are shown with the second test for dynamism, INTERRUPT.

(224) a. ?TEACHER ANGRY INTERRUPT ‘The teacher was getting angry when interrupted by…’

b. *FATHER ANGRY INTERRUPT ‘Father was interrupted from being angry by …’

(225) a. *GRANDMA SICK INTERRUPT ‘Grandma was interrupted from being sick by … ’

b. *GRANDMA SICK INTERRUPT ‘Grandma was interrupted from being sick by … ’

The first predicate, ANGRY in (224a), was interpreted by Group A as a coerced situation, ‘get angry’. Although this coerces the state into an event, the construction was still questionable. Alternative ways of expressing the inchoative meaning were preferred. For Group B, the same construction was considered infelicitous. Both results indicate ANGRY is [-dynamic]. The use of SICK with INTERRUPT in (225) was unacceptable to both groups, also indicating it is [-dynamic].

The predicates differed in their compatibility with Test 3, pace adverbials. The results are seen in (226) and (227).

(226) a. ?TEACHER ANGRY VERY.SLOW ‘The teacher was angry for a long time.’

b. *FATHER ANGRY VERY.SLOW ‘Father slowly became angry.’

88

(227) a. GRANDMA SICK VERY.SLOW ‘Grandma has been sick for such a long time.’

b. LAY.DOWN SICK VERY.SLOW DISAPPEAR NOT.YET ‘(She) has been sick in bed for a long time – she’s still not better.’

As seen in (226a) and (226b), Groups A and B had different interpretations of VERY.SLOW when it was used with ANGRY. For Group A, VERY.SLOW became a temporal extent indicator, ‘for a long time’. However, this construction was still considered pragmatically odd since anger is a fleeting emotion. With Group B, VERY.SLOW was understood as a pace adverbial, and the predicate was interpreted as an inchoative, ‘became angry’. This construction was also unacceptable pragmatically since anger comes on in a moment. The judgments indicate that the predicate is [-dynamic].

In contrast, and as (227) shows, both groups indicated that VERY.SLOW was felicitous with SICK. However, the compatibility was contingent on the adverbial meaning ‘for a long time’, not ‘slow’. Overall, the results indicate that both predicates are [-dynamic].

In (228) and (229) the predicates are shown with Test 4, temporal extend adverbials.

(228) a. TEACHER ANGRY MINUTE-TWO-THREE ‘The teacher was angry for a few minutes.’

b. INDEX-3 ANGRY FIVE MINUTE GO TRAVEL ‘After being angry for five minutes, I went out traveling.’

(229) a. GRANDMA SICK DAY-TWO ALREADY ‘Grandma was sick for two days.’

b. WATCH.PERSON SICK INDEX-3 DAY-TWO INDEX-1 TAKE.PERSON INJECTION ‘I saw they had been sick for two days, so I took them to get an injection.’

As (228) and (229) show, both groups accepted ANGRY and SICK with adverbials of temporal extent, indicating they are [+durative]. Their compatibility also indicates their status as stage-level states.

89

The predicates were also tested with MANY.TIMES, as shown in (230) and (231).

(230) a. TEACHER ANGRY (ANGRY) MANY.TIMES AFRAID ‘The teacher has been angry many times, (so I am) afraid (of him).’

b. FATHER ANGRY MANY.TIMES ‘Father has been angry many times.’

(231) a. GRANDMA SICK MANY.TIMES ALREADY ‘Grandma has been sick many times already.’

b. SICK SICK MANY.TIMES ‘(Grandma) has been sick many times.’

As (230) and (231) show, both ANGRY and SICK occur felicitously with MANY.TIMES. The results reflect the stage-level nature of the predicates since the state can repeatedly occur over the participant’s lifetime.

4.3.2 Comparison of results for stage-level states The acceptability judgments of Groups A and B for stage-level states closely matched those of the Original Group. The results are compared in Table 18.

Table 18 Comparison of results for stage-level states in ThSL

Test Predicate Original Group Group A Group B

Commands ANGRY X X X SICK X X X

INTERRUPT ANGRY X X X SICK X X X

Pace Adverbials ANGRY * X* X* SICK * * *

Temporal Extent ANGRY    SICK   

STILL.THERE ANGRY n/a n/a n/a SICK n/a n/a n/a

MANY.TIMES ANGRY * * * SICK * * *  – compatible ? – questionable X – incompatible * – further considerations

90

As with the Original Group, informants in Groups A and B judged ANGRY and SICK to be incompatible with commands and interruption. For Test 3, the Original Group accepted the use of a pace adverbial with ANGRY when the adverbial was interpreted as ‘for a long time’. Group A also accepted this reading but noted it was pragmatically odd since anger is ordinarily short-lived. With Group B, the adverbial was interpreted as ‘slowly’ and the meaning of the predicate was shifted to an inchoative (‘became angry’). The results all indicate that ANGRY is [-dynamic], making it incompatible with pace adverbials. Likewise, SICK was only acceptable with the test if the adverbials was interpreted as ‘for a long time’. These results indicate that SICK is also [-dynamic].

For Test 4, all groups accepted clauses in which the states occurred with temporal extent adverbials. The compatibility of the predicates with this test first shows they are [+durative] since (in general) only durative predicates are compatible with adverbials of temporal extent. It also indicates that they are stage-level states since individual-level states are not compatible with temporal extent adverbials.

The compatibility of the predicates with the final test, MANY.TIMES, confirms this analysis. However, the groups often indicated that one or both predicates needed to be reduplicated in this context. Table 19 summarizes the preferences among the groups for reduplication.

Table 19 Reduplication of stage-level states in multi-event clauses

Original Group Group A Group B

ANGRY … MANY.TIMES Double or triple Single or double Single articulation articulation articulation

SICK … MANY.TIMES Double or triple Single articulation Double or triple articulation articulation

As captured in Table 19, none of the groups’ judgments for reduplication matched precisely. In general, however, signers from Bangkok preferred reduplicated forms, while signers from Chiang Mai were more likely to accept single iterations. Whether the results arise from regional variation or not requires further research.

4.4 Activities The predicates PAINT and RUN were initially selected to test with Groups A and B. However, Group A used the sign DRAW for both ‘paint’ and ‘draw’. Since the form of

91

DRAW is similar to PAINT, and the meaning is the same, it was used for testing with Group A. With Group B, PAINT was used. The results of testing these predicates are given in Section 4.4.1, then compared with the judgments of the Original Group in Section 4.4.2.

4.4.1 Testing featural values for activities Unlike states, activities are [+dynamic], a value that makes them compatible with Test 1, commands. The examples in (232) and (233) show the selected predicates with this test.

(232) a. DRAW GO ‘Go draw!’

b. STRESSED NO PAINT GO ‘Don’t be stressed, go paint!’

(233) a. RUN GO ‘Go run!’

b. RUN GO ‘Go run!’

The compatibility of DRAW in (232a) and PAINT in (232b) with GO indicates they are both [+dynamic]. Likewise, as the clauses in (233) show, both groups indicated that RUN was felicitous with commands, indicating it is also [+dynamic].

The predicates are shown with the second test for dynamism, INTERRUPT, in (234) and (235).

(234) a. GRANDMA DRAW INTERRUPT ‘Grandma was drawing when interrupted by …’

b. PAINT INTERRUPT EXAMPLE 3-CALL-1 ‘Painting could be interrupted, for example, by someone calling you.’

(235) a. RUN INTERRUPT ‘I was running when interrupted by ...’

b. RUN POKE.FOOT INTERRUPT ‘I was running when interrupted by stepping on something pokey.’

92

The clauses in (234) and (235) were all judged to be felicitous. The compatibility of the predicates with INTERRUPT in these examples again shows that they are [+dynamic].

The dynamism of the predicates and their durative value is also seen in their compatibility with pace adverbials. The clauses in (236) and (237) show the predicates with this test frame.

(236) a. GRANDMA DRAW SLOW ‘Grandma paints/draws slowly.’

b. PAINT VERY.SLOW ‘(Grandma) paints very slowly.’

(237) a. UNCLE RUN VERY.SLOW ‘(My) uncle runs very slowly.’

b. UNCLE RUN VERY.SLOW ‘(My) uncle runs very slowly.’

Both Group A and B accepted the predicates with pace adverbials, as (236) and (237) show. The judgments indicate the predicates are both [+dynamic] and [+durative].

The predicates are shown with the second test for duration, temporal extent adverbials, in (238) and (239).

(238) a. GRANDMA DRAW MINUTE-THIRTY ‘Grandma drew (a picture) in thirty minutes.’

b. INDEX-3 GRANDMA PAINT SELF PAINT FINISH.UP INDEX-1 SEE MINUTE .THIRTY ‘I saw that grandma finished up (the painting) on her own in thirty minutes.’

(239) a. FRIEND RUN THREE.HOUR ‘(My) friend ran (20 kilometers) in/for three hours.’

b. RUN MINUTE-THIRTY 4 POINT 5 KILOMETERS ‘(He) ran 4.5 kilometers in thirty minutes.’

93

The felicitous occurrence of DRAW in (238a), PAINT in (238b), and RUN in (239) with temporal extent adverbials indicates the predicates are [+durative].

Interestingly, the informants in Group A added telic interpretations in the presence of temporal extent adverbials. In (238a), they said the implication was that a specific picture was painted in the time specified (i.e., thirty minutes). In (239a), they interpreted the clause as either an atelic event (ran for three hours) or implied a certain distance was run in that time, creating a telic interpretation. For Group B, constituents were added that made the clauses overtly telic. For example, in (238b) FINISH.UP was added, while in (239b), an explicit distance (4.5 kilometers) was supplied. Based on the informants’ discussions, the tendency to imply or add telicity to the clauses seemed to be a product of the precise nature of the adverbials used, such as ‘thirty minutes’ or ‘three hours’. Less specific time frames (e.g., ‘for a long time’ or ‘for hours and hours’) may not produce the same tendency. This possibility points to the need for further study on the interaction of temporal extent adverbials and telicity (see Section 5.5.2).

The predicates are shown with the first telicity test, STILL.THERE, in (240) and (241).

(240) a. GRANDMA DRAW STILL.THERE ‘Grandma is still there painting/drawing.’

b. GRANDMA PAINT STILL.THERE ‘Grandma is still there painting.’

(241) a. UNCLE RUN STILL.THERE ‘(My) uncle is still out running.’

b. RUN STILL.THERE ‘(He) is still out running.’

As (240) and (241) show, both groups accepted the use of STILL.THERE with DRAW, PAINT, and RUN. The results indicate the predicates are [-telic].

The predicates are shown with the second test for telicity, MANY.TIMES, in (242) and (243).

(242) a. GRANDMA DRAW MANY.TIMES ALREADY CHANGE PLANT ARRANGE ‘Grandma has already drawn many times. (She) wants to arrange flowers instead.’

94

b. INDEX-3 PAINT MANY.TIMES CORRECT ‘He paints all the time, that’s true.’

(243) a. RUN MANY.TIMES CHANGE RIDE.BICYCLE ‘(I) have run many times, I want to bicycle instead.’

b. RUN MANY.TIMES ‘(He) has run many times.’

As the clauses in (242) and (243) show, DRAW, PAINT, and RUN were all felicitous with MANY.TIMES. The results again indicate that the predicates are [-telic].

4.4.2 Comparison of results for activities The featural values of DRAW/PAINT and RUN indicated by Group A and B’s acceptability judgments matched those of the Original Group. Table 20 shows the results from each group.

Table 20 Comparison of results for activities in ThSL

Test Predicate Original Group Group A Group B

Commands PAINT/DRAW   

RUN    INTERRUPT PAINT/DRAW    RUN    Pace Adverbials PAINT/DRAW    RUN    Temporal Extent PAINT/DRAW   

RUN    STILL.THERE PAINT/DRAW    RUN    MANY.TIMES PAINT/DRAW    RUN     – compatible ? – questionable X – incompatible * – further considerations

As Table 20 illustrates, activity predicates were judged to be compatible with all tests by all groups. The compatibility of the predicates with commands and interruption indicates they are [+dynamic], and their compatibility with pace

95

adverbials and temporal extent adverbials indicates they are [+durative]. Lastly, their felicitous occurrences with STILL.THERE and MANY.TIMES indicate they are [-telic].

4.5 Accomplishments The accomplishment predicates BEND.BAR and ICE.MELT were chosen for extended testing. The results from testing these predicates with Groups A and B are presented in Section 4.5.1. They are then compared with the responses of the Original Group in Section 4.5.2.

4.5.1 Testing featural values for accomplishments As dynamic events, accomplishments are generally compatible with Test 1, commands. The results from testing the two accomplishment predicates with this test are shown in (244) and (245).

(244) a. BEND.BAR GO ‘Go bend a bar!’

b. BEND.BAR GO ‘Go bend a bar!’

(245) a. *ICE ICE.MELT GO ‘Ice, melt!’

b. *ICE ICE.MELT GO ‘Ice, melt!’

As (244) shows, both Group A and B accepted BEND.BAR with GO, indicating it is [+dynamic]. As (245) shows, however, the inanimate nature of ICE made the second predicate incompatible with GO.

As telic events, neither predicate can occur with INTERRUPT when the full form of the sign is used. The clauses in (246) and (247) reflect this constraint.

(246) a. *SUPERMAN METAL BAR BEND.BAR INTERRUPT ‘Superman bent the metal bar when he was interrupted …’

b. *BEND.BAR INTERRUPT ‘(Someone) bent a metal bar when interrupted by …’

96

(247)6 a. *ICE.MELT INTERRUPT ‘The ice melted when interrupted by …’

In (246) and (247), the full forms of BEND.BAR and ICE.MELT are used with INTERRUPT. As expected, this form of the predicates is infelicitous with the test since it indicates that the event has reached its final endpoint.

If the form of the predicate is modified to indicate the event has not yet reached its endpoint, it then occurs felicitously with INTERRUPT. The clauses in (248) and (249) demonstrate.

(248) a. SUPERMAN METAL BAR BEND.BARUNCOMPLETED INTERRUPT ‘Superman was bending the metal bar when he was interrupted by …’

b. BEND. BARUNCOMPLETED INTERRUPT ‘(Someone) was bending a metal bar when interrupted by …’

(249) a. ICE.MELTUNCOMPLETED INTERRUPT PUT.IN.FREEZER ‘The ice was melting when interrupted by placing it in the freezer…’

b. ICE.MELTUNCOMPLETED INTERRUPT PUT.IN.FREEZER ‘The ice was melting when interrupted by placing it in the freezer…’

The clauses in (248) and (249) show that both predicates occur felicitously with INTERRUPT when the uncompleted forms of the signs are used. The results indicate that the predicates are [+dynamic].

The predicates are shown with Test 3, pace adverbials, in (250) and (251).

(250) a. SUPERMAN BEND.BAR SLOW ‘Superman bent the metal bar slowly.’

b. SUPERMAN BEND.BAR SLOW ‘Superman bent the metal bar slowly.’

(251) a. ICE ICE.MELT VERY.SLOW ‘The ice melted very slowly.’

b. ICE ICE.MELT VERY.SLOW WHY WEATHER COLD ‘The ice melted very slowly because of the cold weather.’

6 Data point unavailable for Group B. Only the uncompleted form of the predicate was tested.

97

As seen in (250) and (251), both predicates were felicitous with pace adverbials. Often, however, the predicates were articulated more slowly in this context and their uncompleted forms were used. The compatibility of the predicates with pace adverbials indicates they are [+durative].

The same durative value is reflected in the co-occurrence of the predicate with temporal extent adverbials. The clauses in (252) and (253) show them with this test frame.

(252) a. HUMAN PEOPLE BEND.BAR TIME MINUTE-TWO MINUTE-THREE ‘Ordinary people can bend a bar in two or three minutes.’

b. INDEX-1 INDEX-3 BEND.BAR TWO.MINUTES ENOUGH.TIME ‘I could bend that bar in two minutes.’

(253) a. ICE ICE.MELT MINUTE-ONE ‘The ice melted in a minute.’

b. CYLINDER INDEX.LEFT WATER HOT INDEX.RIGHT WATER COLD ICE DROP.LEFT-DROP.RIGHT

MELTLEFT MINUTE ONE MELTRIGHT MINUTE TWO ‘If one cylinder had hot water and another had cold water and you dropped an ice cube in each, the one in the hot water would melt in a minute and the one in cold water would melt in two minutes.’

The compatibility of BEND.BAR in (252) and ICE.MELT in (253) with temporal extent adverbials indicates they are [+durative] events.7

In (254) and (255), the predicates are shown with the first test for telicity, STILL.THERE.

(254) a. *SUPERMAN METAL BAR BEND.BAR STILL.THERE ‘Superman bent the metal bar and is still bending it.’

b. *BEND.BAR STILL.THERE ‘(Someone) bent a bar and is still bending it.’

7 The sign used for MELT in (253b) shows ice melting in a cylindrical object (e.g., a cup or test tube). No examples were available from Group B that used the sign ICE.MELT for ice melting on a flat surface.

98

(255) a. *ICE ICE.MELT STILL.THERE ‘The ice melted and is still there melting.’

b. *ICE ICE.MELT STILL.THERE ‘The ice melted and is still there melting.’

As (256) and (257) show, both predicates were incompatible with STILL.THERE. The results indicate a [+telic] value.

The predicates are shown with the final test, MANY.TIMES, in (256) and (257).

(256) a. *SUPERMAN BEND.BAR MANY.TIMES ‘Superman bent the bar many times.’

b. *BEND.BAR MANY.TIMES ‘(Someone) bent the bar many times.’

(257) a. *ICE ICE.MELT MANY.TIMES ‘The ice melted many times.’

b. *ICE ICE.FALL ICE.MELT MANY.TIMES ‘The ice fell and melted many times.’

As the clauses in (256) and (257) demonstrate, both BEND.BAR and ICE.MELT are incompatible with MANY.TIMES. Their incompatibility indicates that both predicates are [+telic].

4.5.2 Comparison of results for accomplishments The results from testing BEND.BAR and ICE.MELT with Groups A and B matched the results of the Original Group. Table 21 provides a summary of the judgments of each group.

Table 21 Comparison of results for accomplishments in ThSL

Test Predicate Original Group Group A Group B

Commands BEND.BAR    ICE.MELT X* X* X*

INTERRUPT BEND.BAR * * * ICE.MELT ?* * *

99

Test Predicate Original Group Group A Group B

Pace Adverbials BEND.BAR    ICE.MELT   

Temporal Extent BEND.BAR    ICE.MELT   

STILL.THERE BEND.BAR X X X ICE.MELT X X X

MANY.TIMES BEND.BAR X X X ICE.MELT X X X  – compatible ? – questionable X – incompatible * – further considerations

As Table 21 shows, all groups accepted BEND.BAR with commands, indicating it is [+dynamic]. For ICE.MELT, however, the lack of semantic agency in ICE makes the predicate incompatible with the test. In order to be compatible with the second test, INTERRUPT, the predicates had to occur in an uncompleted form. This form presents the event as completed and allows the process leading up to the result state to be interrupted. The ability to interrupt the event itself indicates it is both [+dynamic] and [+durative].

All three groups accepted the predicates with pace adverbials and temporal extent adverbials. The judgments indicate a [+durative] value. Finally, none of the groups accepted constructions involving STILL.THERE or MANY.TIMES with the predicates in their original (i.e., completed) forms. These results show the predicates are [-telic].

4.6 Achievements Achievements are dynamic, instantaneous, telic events. The predicates ERASE and OPEN.CAN were used to represent this situation type in extended testing. The results are discussed in Section 4.6.1, then compared with the results from the Original Group in Section 4.6.2.

4.6.1 Testing featural values for achievements The predicates are shown with the first test, commands, in (258) and (259).

(258) a. GO OPEN.CAN PEPSI INDEX-1 EAT NOT.YET ‘Go open a Pepsi for me – I haven’t eaten yet.’

100

b. OPEN.CAN GO ‘Open the can!’

(259) a. INDEX-3 WRONG INDEX-2 ERASE ‘That’s wrong, erase it.’

b. GO ERASE QUICKLY ‘Quick! Erase that!’

As (258) and (259) show, both OPEN.CAN and ERASE are acceptable with commands. Their compatibility with the test indicates that they are [+dynamic].

When the predicates were used with Test 2, INTERRUPT, the informants used the unrealized inceptive form of the predicates. The shortened form is seen in (260) and (261).

(260) a. OPEN.CANUNREALIZED INTERRUPT ‘I was about to open a can when interrupted by …’

b. OPEN.CANUNREALIZED WAIT TALK INTERRUPTED ‘I was about to open a can when someone came to talk to me, so I was interrupted.’

(261) a. LANGUAGE WANT ERASEUNREALIZED NO INTERRUPT ‘I wanted to erase messages online, ‘No!’ (someone said), so I was interrupted.’

b. TEXT ERASEUNREALIZED 3-CALL-1 VIDEO.CHAT TALK ERASE NOT.YET ‘I was texting and about to erase (a message) when someone called me to video chat, so I didn’t get to erase it.’

In (260) and (261), using the unrealized inceptive form of the predicates made them felicitous with INTERRUPT. The resulting clauses indicate that the events themselves did not occur. This shows that the predicates are [-durative] since the event itself cannot be interrupted.

The predicates are shown with Test 3, pace adverbials, in (262) and (263).

(262) a. ?ERASE SLOW STUCK ‘I was too slow in erasing (the messages), so I was caught.’

101

b. *FRIEND LINE ERASE VERY.SLOW ‘My friend erased something on Line very slowly.’

(263) a. ?OPEN.CAN VERY.SLOW ‘(She) opened the can very slowly.’

b. *OPEN.CAN VERY.SLOW ‘(She) opened the can very slowly.’

As (262a) and (262b) show, ERASE was marginally acceptable or unacceptable with SLOW. In both cases SLOW was interpreted as modifying the period leading up to the event. The results show that the event proper is [-durative]. Likewise, Group A indicated that (263a) was marginally acceptable but that the construction indicated the event itself (‘open can’) had not yet occurred. Group B said the construction was incompatible, as (263b) shows. Again, the responses of both groups indicate that OPEN.CAN is [-durative].

The predicates are shown with the second test for duration, temporal extent adverbials, in (264) and (265). The sign ONE.TIME was added to most of the test frames to prevent an iterative interpretation of the events.

(264) a. *FRIEND LINE ERASE ONE.TIME MINUTE-FIVE ‘(My) friend erased a Line (message) one time for five minutes.’

b. *LINE ERASE ONE.TIME MINUTE-TWO ‘(She) erased a Line (message) one time for two minutes.’

(265) a. ?OPEN.CAN MINUTE-ONE ‘(He) opened the can in a minute.’

b. *OPEN.CAN ONE.TIME MINUTE-TWO ‘(He) opened the can once for two minutes.’

As the clauses in (264) show, the use of ERASE with temporal extent adverbials was unacceptable to both groups. Both clauses utilized ONE.TIME to block an iterative interpretation. Group B also reported that the clause in (265b) was infelicitous with the temporal extent adverbial when ONE.TIME was present. In (265a), ONE.TIME was not used. This made the clause marginally acceptable under a delayed inceptive interpretation. However, Group A indicated that the time frame should be no more than a minute and that, typically, the event only took a moment. All of the results reflect the [-durative] nature of the predicates.

102

In (266) and (267) the predicates are shown with the first test for telicity, STILL.THERE.

(266) a. *ERASE STILL.THERE ‘(She) is still there erasing (a message).’

b. *ERASE STILL.THERE ‘(He) is still erasing the message.’

(267) a. FRIEND OPEN.CAN ‘(My) friend opened a can.’ *OPEN.CAN STILL.THERE ‘Is he still opening it?’

b. OPEN.CAN STILL.THERE ‘He is off opening a can.’ ‘He is still over there opening cans.’

As seen in (266), both groups indicated that STILL.THERE was incompatible with ERASE. For OPEN.CAN, Group A judged the predicate to be infelicitous with STILL.THERE, as (267a) shows. As (267b) shows, Group B allowed the predicate to occur with STILL.THERE under one of two interpretations: if it referred to a broader context (e.g., ‘off opening a can’) or involved plural patients (i.e., ‘opening cans’). Although varied in their details, the results all indicate that the events are [+telic].

The predicates are shown with the second test for telicity, MANY.TIMES, in (268) and (269).

(268) a. *LINE ERASE MANY.TIMES ‘I erased the Line (message) many times.’

b. *LINE ERASE MANY.TIMES ‘I erased the Line (message) many times.’

(269) a. *INDEX-1 FRIEND OPEN.CAN MANY.TIMES ‘My friend opened the can many times …’

b. *PEPSI OPEN.CAN MANY.TIMES ‘(She) opened the can many times …’

As (268) and (269) show, the predicates were incompatible with MANY.TIMES. This again reflects a [+telic] value.

103

4.6.2 Comparison of results for achievements Table 22 summarizes the results of testing ERASE and OPEN.CAN. Although the results differed across the three groups in some of their details, the featural values indicated are consistent.

Table 22 Comparison of results for achievements in ThSL

Test Predicate Original Group Group A Group B

Commands ERASE    OPEN.CAN   

INTERRUPT ERASE * * * OPEN.CAN * * *

Pace Adverbials ERASE * ?* X* OPEN.CAN * ?* X*

Temporal Extent ERASE X X X OPEN.CAN * X X

STILL.THERE ERASE X X X OPEN.CAN X X X*

MANY.TIMES ERASE * X X OPEN.CAN X X X  – compatible ? – questionable X – incompatible * – further considerations

Across all groups, OPEN.CAN and ERASE occurred felicitously with commands, indicating they are [+dynamic]. When used with INTERRUPT, the predicate took an unrealized inceptive form that indicated the event itself had not yet occurred. The fact that the events themselves could not be interrupted shows the predicates are [-durative].

For Test 3, occurrence with pace adverbials, the groups varied in their acceptability judgments. The Original Group accepted SLOW with the predicates under a coerced reading involving preliminary stages. It was these stages that unfolded slowly, not the event itself. Group A gave similar responses but rated the constructions as only marginally acceptable. Group B did not accept them. The differing judgments reflect a clash between the featural values of the adverbial and those of the predicate and point to a [-durative] value.

104

For the second test of duration, temporal extent adverbials, the groups also indicated that the clauses were either infelicitous or had coerced interpretations. The results again indicate that the predicates are [-durative].

Finally, the groups typically indicated that the predicates were infelicitous with STILL.THERE. Group B, however, accepted OPEN.CAN with STILL.THERE if the clause referred to an episode (i.e., ‘off opening a can’) or involved multiple objects (i.e., ‘opening cans’). These results point to the [+telic] nature of the events proper. The predicates were also unacceptable with MANY.TIMES, again indicating a [-telic] value.

4.7 Semelfactives Semelfactives are dynamic, instantaneous, atelic events. The predicates PUNCH.PERSON and SHOUT were selected for extended testing of semelfactives. The results of testing these predicates with Groups A and B are discussed in Section 4.7.1. In Section 4.7.2 the results are compared with those of the Original Group.

4.7.1 Testing featural values for semelfactives The predicates are shown with the first test of dynamism, compatibility with commands, in (270) and (271).

(270) a. PUNCH.PERSON GO ‘Go punch that person!’

b. GO PUNCH.PERSON ‘Go punch that person!’

(271) a. SHOUT GO ‘Go shout!’

b. 1-GIVE-2 SHOUT GO TEST ‘Go ahead and shout, try it!’

As (270) and (271) show, both PUNCH.PERSON and SHOUT were compatible with command forms. The results indicate that the predicates are [+dynamic].

The predicates are shown with the second test, INTERRUPT, in (272) and (273).

105

(272) a. PUNCH.PERSONUNREALIZED FRIEND WAIT STOP

PUNCH.PERSONUNREALIZED INTERRUPT ‘I was about to punch someone when a friend said, ‘Wait, stop!’ So, I was interrupted from punching them.’

b. WANT PUNCH.PERSONUNREALIZED [NO PUNCH NO] INTERRUPT NO ‘I wanted to punch someone but (someone said), ‘Don’t punch them!’ So, I was interrupted and didn’t do it.’

(273) a. ?SHOUT INTERRUPT ‘(I) won’t shout.’

b. *SHOUT INTERRUPT ‘(I) was interrupted from shouting by …’

As (272a) shows, when PUNCH.PERSON was used with INTERRUPT, informants automatically used the unrealized inceptive form of the predicate. This form was compatible with the command. The inability to interrupt the event itself shows it is [-durative]. It’s compatibility with INTERRUPT indicates it is [+dynamic].

The second predicate, SHOUT, did not take an unrealized inceptive form, as seen in (273). Informants gave the construction either a questionable (273a) or unacceptable (273b) rating when the sign was used with INTERRUPT. Both groups noted that it was more natural to express the unrealized event by signing ‘no’ (i.e., ‘no, I won’t shout’) or using only the unrealized inceptive form of the predicate without adding INTERRUPT. The existence of an unrealized inceptive form indicates that SHOUT is indeed [+dynamic] since states are generally incompatible with this form. The inability to interrupt the event itself indicates it is [-durative].

The predicates are shown with the first test for duration, pace adverbials, in (274) and (275).

(274) a. *PUNCH.PERSON SLOW ‘He punched (someone) slowly.’

b. *PUNCH.PERSON VERY.SLOW ‘He was slow to punch someone.’

(275) a. ?SHOUT GO SHOUT VERY.SLOW ‘Come on and shout! You are so slow to shout!’

106

b. *FRIEND SHOUT VERY.SLOW ‘(My) friend shouted slowly.’

As (274) shows, both groups indicated that PUNCH.PERSON was infelicitous with pace adverbials, thus indicating it is [-durative]. When SHOUT was placed in the same test frame, some informants from Group A accepted the construction while others did not, making the clause in (275a) questionable. Those who accepted the clause noted that the adverbial modified the period leading up to the event (i.e., ‘slow to x’) rather than the event itself. For Group B, the pace adverbial was understood to modify the event itself, and the clause, as (275b) shows, was unacceptable. The results from both groups indicate that SHOUT is [-durative].

The predicates are shown with the second test for duration, temporal extent adverbials, in (276) and (277). The sign ONE.TIME is used in the examples to prevent a coerced, iterative interpretation of the events.

(276) a. *PUNCH.PERSON ONE.TIME MINUTE-FIVE ‘(He) punched (someone) one time for five minutes.’

b. *INDEX-3 FRIEND PUNCH.PERSON ONE.TIME MINUTE-TWO ‘My friend punched someone one time for two minutes.’

(277) a. *WOMAN SHOUT ONE.TIME MINUTE-FIVE ‘The woman shouted one time for five minutes.’

b. *WOMAN SHOUT ONE.TIME MINUTE-TWO ‘The woman shouted one time for two minutes.’

As (276) and (277) show, both predicates were unacceptable with temporal extent adverbials. The results indicate that they are [-durative].

The examples in (278) and (279) show the predicates with the first test for telicity, STILL.THERE.

(278) a. INDEX FRIEND SWEETHEART PUNCH.PERSON ‘(My) friend’s sweetheart punched her.’ PUNCH.PERSON PUNCH.PERSON STILL.THERE ‘Is he still punching her?’

107

b. SOHM WHERE ‘Where is Sohm?’ SOHM SWEETHEART PUNCH.PERSON PUNCH.PERSON STILL.THERE PUNCH.PERSON PUNCH.PERSON GO 2-HELP-3 ‘Sohm is still beating up his sweetheart – go help!’

(279) a. SHOUT SHOUT LOUD STILL.THERE ‘(Someone) is still there shouting loudly.’

b. HEY INDEX-3 DOG SHOUT STILL.THERE ‘Hey, the dog is still there barking.’

In (278), PUNCH.PERSON is felicitous with questions of continuation using STILL.THERE, indicating that it is [-telic]. Both groups required reduplication of the predicate in the question portion. This requirement is expected since instantaneous events must be repeated to continue.

In (279), SHOUT was also felicitous with STILL.THERE. As (279a) shows, Group A also required this predicate to be repeated to allow an iterative interpretation. Group B allowed an iterative interpretation of the event even with a single repetition of the predicate, as (279b) shows. The ability to repeat the event with the same participants indicates that SHOUT is also a [-telic] situation.

In (280) and (281), the predicates are shown with the final test, MANY.TIMES.

(280) a. *PUNCH.PERSON MANY.TIMES ALREADY ‘I punched (that person) many times already.’

b. *PUNCH.PERSON MANY.TIMES ‘(He) punched someone many times.’

(281) a. *SHOUT MANY.TIMES ‘(She) shouted many times.’

b. *SHOUT MANY.TIMES ‘(She) shouted many times.’

In (280) and (281), The single iterations of the predicates is infelicitous with MANY.TIMES. This is expected since MANY.TIMES creates a multi-event interpretation of the clauses. If the predicates are repeated, they are then compatible with this construction, as (282) and (283) show.

108

(282) a. PUNCH.PERSON PUNCH.PERSON MANY.TIMES ALREADY ‘I punched (that person) many times already.’

b. PUNCH.PERSON PUNCH.PERSON MANY.TIMES BE.CAREFUL ‘(He) has punched (someone/people) many times, be careful.’

(283) a. SHOUT SHOUT SHOUT ENOUGH MANY.TIMES ‘(You’ve) shouted many times – enough!’

b. SHOUT SHOUT MANY.TIMES PERSONALITY INDEX-3 PERSON.CL ‘She shouts a lot, it’s her personality.’

Repetition of the predicates in (282) and (283) makes them felicitous with MANY.TIMES. The fact that the events are repeated with the same participants indicates they are [-telic].

4.7.2 Comparison of results for semelfactives The results for testing PUNCH.PERSON and SHOUT across groups are presented inTable 23. Although the groups again differed in their acceptability judgments of coerced situations, the featural values indicated by their judgments were consistent.

Table 23 Comparison of results for semelfactives in ThSL

Test Predicate Original Group Group A Group B

Commands PUNCH.PERSON   

SHOUT   

INTERRUPT PUNCH.PERSON * * * SHOUT * ?* X

Pace Adverbials PUNCH.PERSON * X ?

SHOUT * X X

Temporal PUNCH.PERSON X X X Extent SHOUT X X X STILL.THERE PUNCH.PERSON  * * SHOUT  * * MANY.TIMES PUNCH.PERSON * * * SHOUT * * *  – compatible ? – questionable X – incompatible * – further considerations

109

Across groups, the semelfactive predicates PUNCH.PERSON and SHOUT were acceptable with commands, indicating they are both [+dynamic]. In general, the predicates were also compatible with INTERRUPT when an uncompleted form of the sign was used. The uncompleted form indicates an unrealized event. The shifted meaning indicates that the predicates are [-durative] since the events themselves cannot be interrupted.

When used with pace adverbials, the predicates received a delayed inceptive interpretation. Although the Original Group accepted these constructions, Groups A and B found them unacceptable. However, both results (coerced interpretations and judgments of unacceptability) indicate the predicates themselves are [-durative]. The same results were indicated when testing the predicates with temporal extent adverbials. The sign ONE.TIME was added to the test frame to prevent iterative interpretations of the events. All groups indicated that the predicates were unacceptable in this frame, showing that they are [-durative].

Finally, both PUNCH.PERSON and SHOUT were compatible with STILL.THERE and MANY.TIMES. The repeatability of the events with the same participants indicates they are [-telic]. To occur felicitously with either test, however, most participants required a double articulation of the predicates.

4.8 Summary of results The indicated featural values of the 12 predicates used in extended testing were consistent across all groups. The details of their behavior within test frames varied slightly. First, the Original Group was more willing to accept clauses with coerced situation types that arose from a clash between the nature of the adverbial and that of the predicate. Group A was less accepting of these coerced clauses and often gave them only marginally acceptable ratings. Group B tended to indicate they were simply unacceptable.

The groups – and individual members within the groups – also varied in their judgments of when a predicate needed to be reduplicated with STILL.THERE and MANY.TIMES. The reason behind these variations requires further investigation.

110

Chapter 5 Conclusion

5.1 Overview Several findings emerged during this study on situation types in Thai Sign Language. First among these was the identification of situation types for ThSL predicates and their behavior in different test frames. Section 5.2 provides a summary of these findings. Some of the results from these tests have interesting repercussions for the claims of the Event Visibility Hypothesis (EVH). Section 5.3 discusses these. The study also brought up several considerations for testing situation types in sign languages. These are treated in Section 5.4. Section 5.5 then presents areas for further research.

5.2 Summary of findings on ThSL situation types This study has explored situation types in ThSL through a series of tests. The first test – compatibility with command forms – was used to distinguish states from events. The test is based on the tendency of events to be volitional, making them compatible with commands. ThSL event predicates occurred felicitously with this test, revealing that they are [+dynamic]. Exceptions to this involved non-volitional subjects. Stative predicates were incompatible with commands, showing that they are [-dynamic].

The second test was the sign INTERRUPT. The test identified the dynamic value of predicates as well as the durative value of events. Predicates expressing stative situations were incompatible with INTERRUPT, highlighting the fact that they are [-dynamic]. The remaining predicates occurred felicitously with INTERRUPT, indicating they are [+dynamic]. The durative value of the predicates predictably affected the interpretation of the clause. For durative predicates (activities and accomplishments), the event itself was interrupted, revealing that the predicates are [+durative]. For instantaneous events, INTERRUPT applied to preliminary stages of the event and indicated the event itself did not occur. Often, signers used unrealized inceptive forms of the predicate in these clauses. The results show that the predicates are [-durative].

111

The third test, compatibility with pace adverbials, also identified the dynamic and durative values of predicates. Specifically, since states lack motion, they are incompatible with pace adverbials. The discordancy caused the signs SLOW and VERY.SLOW to be interpreted as ‘for a long time’, as evidenced by paraphrases of the clause. Stage-level states were compatible with the adverbials under this interpretation, reflecting their status as transitory states. In contrast, individual-level states were incompatible with the adverbials under either a pace-based or temporal extent-based interpretation. The former indicates that they are [-dynamic]; the latter that they are individual-level states. Thus, the test distinguished 1) states from events and 2) individual-level from stage-level states.

The compatibility of the test with most of the remaining predicates indicates that they are [+dynamic]. However, the meaning of the clause varied depending on the durative value of the predicate. With durative events, the adverbials indicated the event itself unfolded slowly (i.e., ‘painted slowly’), affirming that they are [+durative]. With instantaneous events, the test tended to invoke preliminary stages. It was these preliminary stages that unfolded slowly, not the event itself. Smith’s (1997) principle of external override predicts this type of coercion.

The fourth test used adverbs of temporal extent to determine the durative value of predicates. The test also distinguished state types. Since stage-level states are transitory, they were compatible with specifications of temporal extent. Individual- level states, on the other hand, tend to endure throughout the life of the participant, making them incompatible with the test. For dynamic situation types, both activity and accomplishment predicates occurred felicitously with temporal extent adverbials, showing they are [+durative]. Achievement predicates were either incompatible or underwent coerced interpretations involving preliminary stages. Both results reflect a [-durative] value. Some of the semelfactives were acceptable with the test under an iterative interpretation. However, when the addition of ONE.TIME blocked this interpretation, the clauses became unacceptable, showing that the predicates are [-durative].

The fifth test, STILL.THERE, identified the telic values of the predicates. The test is incompatible with events that have reached a result state; once the event reaches the result state, it cannot continue. Since activities do not involve a result state, the activity predicates were compatible with the test, showing they are [-telic]. The other atelic event type, semelfactives, only occurred with the test when the predicate was reduplicated, due to their instantaneous nature. Finally, telic

112

predicates (accomplishments and achievements) did not occur with STILL.THERE when the events were: 1) portrayed as completed, and 2) a single participant was involved. The results demonstrate that they are [+telic].

The sixth and final test was MANY.TIMES. The test identified the telic value of dynamic situation types. However, it also served to distinguish stage-level from individual-level states. Namely, stage-level states – which may occur multiple times over the life of a participant – were compatible with the test. In general, however, the predicate had to be repeated to occur felicitously in the test frame. Individual- level states, in contrast, were incompatible with the test. For dynamic situations, the compatibility of activity and semelfactive predicates shows that they are [-telic]. Semelfactives, however, also had to be repeated to be used in the test frame. Reversible achievements occurred with MANY.TIMES if the signers repeated the predicate signs. For these predicates, however, it was implied that the result state was reversed between each occurrence of the event. Irreversible achievements and accomplishments were not incompatible with the test, indicating that they are [+telic].

The tests objectively identified the featural values of ThSL predicates. The identification of these values, in turn, allows the testing of the proposed morphological marking of telicity and duration (i.e., the Event Visibility Hypothesis) with ThSL predicates.

5.3 ThSL and the Event Visibility Hypothesis (EVH) The Event Visibility Hypothesis (EVH) proposes that temporal features are morphologically marked in sign languages, giving rise to a cross-linguistic family resemblance (Wilbur, 2008, p. 217). Several components of the hypothesis are not new, as Wilbur (2008, pp. 246–247) herself notes. However, telicity has been found cross-linguistically to be an inherent property of predicates. Thus, the EVH’s proposal that it is morphologically marked in sign languages is distinctive. Wilbur calls this marking the EndState morpheme and describes it as a rapid deceleration to a stop (see Figure 8). In addition to the EndState morpheme, she also posits the Extent morpheme as a marker of duration in non-spatial predicates (Wilbur, 2008, p. 220). The morpheme manifests as movement along a line, although the line need not be straight to any degree (see Figure 9). The definition is quite broad, allowing the Extent morpheme to manifest in several ways.

113

The test results in Chapter 3 provide a basis for exploring the validity of both claims in relation to ThSL. Let us begin with the EndState morpheme. Several factors suggest that the EVH analysis of this movement is incorrect. Comparing telic and atelic instantaneous events, semelfactives, and achievements, provides the most compelling evidence against the EndState morpheme marking telicity.

Concerning semelfactives, Wilbur (2008, p. 220) dismisses them as a distinct type. She bases this on the claim that semelfactives have an internally heterogeneous composition. However, this analysis is not well-founded. Specifically, the term ‘heterogenous’ refers to an event with distinct subintervals. All telic events are heterogenous since the endpoint of the event involves a new state. For instance, ‘die’ includes two distinct stages: alive  dead. Semelfactive events, however, are not heterogenous in this sense. Instead, they involve a single stage. On either side of the stage is a rest state. For example, one can schematize the event ‘blink’ as rest state  blink  rest state. Rest states, however, are not included in the schema of an event. For example, an activity does not have the schema ‘rest state  walk  walk  walk  rest state’ but rather ‘walk  walk  walk.’ Thus, semelfactives are homogenous events involving a single stage. If semelfactives in sign languages pattern like telic events (presumably in possessing the EndState morpheme), then this is evidence against the movement marking telicity. Two of the ThSL predicates – SHOUT and PUNCH.PERSON support this conclusion. Although both are articulated with the EndState morpheme, neither is telic.

First, the sign SHOUT has the same movement that Wilbur (2008, p. 232) identifies as a manifestation of EndState in the ASL sign POSTPONE. Figure 18 displays the two signs side by side. The illustration of POSTPONE comes from Wilbur (2008, p. 232).

Figure 18 ASL sign POSTPONE (left) and ThSL sign SHOUT (right)

The ASL sign POSTPONE is on the left in Figure 18. The signer touches his thumbs and fingers together, forming small circles. The remaining fingers are extended. The hands are held in front of the body with the ulnar (pinkie) side down, then one or

114

both hands are moved forward (Vicar, 2001). Wilbur (2008, p. 232) states that this movement morphologically marks the [+telic] value of POSTPONE. However, SHOUT is articulated with the same movement, as the last two frames in Figure 18 show. The signer places the back of a claw-shaped hand near his mouth, then moving it rapidly up and out. However, although the movement contour is the same as POSTPONE, the results in Section 3.7.3 show that SHOUT is atelic. Thus, the movement does not mark telicity in ThSL.

The ThSL sign PUNCH.PERSON provides further evidence for this analysis. The sign parallels the ASL sign HIT, which Wilbur (2008, p. 232) presents as another sign with morphologically marked telicity. Figure 19 shows the two signs side by side.

Figure 19 ASL sign HIT (left) and ThSL sign PUNCH.PERSON (right)

In Figure 19, the first image shows the ASL sign HIT. In the sign, the non-dominant hand forms an ASL classifier for a person: a closed hand with the index finger extended. The signer moves his other fist forcefully against the side of the extended index finger. From the drawing in Figure 19, it appears that HIT does not include a bounce-back motion after the hands come into contact. The absence contrasts with the ThSL sign PUNCH.PERSON, which the three still shots on the right in Figure 19 show. In this sign, the signer extends the thumb and pinkie of his non-dominant hand, forming a ThSL classifier for a person. The fist of the dominant hand is brought forcefully against the non-dominant hand and then bounces back. Despite the difference regarding the bounce-back motion, the overall movement is still a rapid deceleration to a stop – presumably, the EndState morpheme. However, although Wilbur (2008, p. 232) claims that HIT is telic in ASL, PUNCH.PERSON is atelic in ThSL (see Section 3.7.3). Again, then, rapid deceleration to a stop does not appear to mark telicity in ThSL predicates.

The evidence from the remaining semelfactives, CLAP, KNOCK.ON.DOOR, and PERSON.DUNK is less straightforward. First, CLAP and KNOCK.ON.DOOR have repeated movements. For instance, in CLAP, the hands are brought together, then apart, then

115

together and apart again. The sign PERSON.DUNK also involves multiple sub- movements: one to encode submerging and one to encode emerging. However, a closer look at the internal movements of these semelfactives suggests that deceleration to a stop may mark the boundaries between finer sub-events that compose the overall event.

This analysis matches the proposal of Klima and Bellugi’s (1979) that such movement (deceleration to a stop) marks the end of an event. They observe that modifications of the movement for the sign SICK have semantic effects. Notably, if the movement comes to a sudden stop, the predicate becomes inchoative, ‘get sick’. Since an inchoative is both instantaneous and telic, the movement could mark either telicity or a final boundary. The data from ThSL supports the boundary analysis.

Such a boundary marker is expected with instantaneous situations since their momentary nature means the end of the event occurs immediately after the beginning. In instantaneous events that often involve repetition (e.g., each clap in ‘clapping’) or roundtrip movement (e.g., submerging and emerging in DUNK), the movement may occur several times. Analyzing phonological deceleration to a stop as semantically indicating a final boundary is also in line with Supalla and Newport’s (1978, p. 103) observation that a “single movement in the sign corresponds to single, punctual, or perfective action.”

If deceleration to a stop marks a final boundary, it should occur with completed, telic events. It should not occur in telic events that are portrayed as ongoing since these do not include a final boundary. These predictions align with the completed and uncompleted forms of the accomplishment predicates tested in ThSL. The two forms of BEND.BAR in Figure 20 and Figure 21 illustrate this feature. Phonologically, the forms differ in movement length and whether the movement ends in a sudden stop. Semantically, the full movement indicates a completed event. The ongoing movement shows an uncompleted event. Figure 20 shows the first form.

Figure 20 Completed form of ThSL sign BEND.BAR

116

Figure 20 shows the ThSL sign BEND.BAR. The signer’s hands are first placed at shoulder height and held in fists as if grasping a bar. The right hand then moves sharply down to a point near the signer’s waist. In this form, the predicate is incompatible with STILL.THERE, indicating it is both telic and completed. When the movement drops off (instead of coming to a stop), it can then occur felicitously with STILL.THERE. Figure 21 shows this form.

Figure 21 Uncompleted form of ThSL sign BEND.BAR

For the predicate sign in Figure 21, the signer again holds both hands at shoulder height and closes his fists. However, rather than moving the dominant hand down sharply, the signer moves it down gradually and with effort. Before reaching the signer’s waist, the sign either stops (with no notable deceleration) or transitions into the next sign. Based on Kuhn’s (2017, p. 10) intuitive analysis, this modification portrays the event as uncompleted (e.g., a result state has not been reached).

The question is, does the absence of a result state in an uncompleted event change the telicity of the event or does it only remove (or alter) the final boundary? Both analyses explain compatibility with STILL.THERE. A telic event can continue as long as the endpoint has not been reached, while an atelic event can continue indefinitely. However, based on the evidence from semelfactives that the deceleration to a stop marks a final boundary, it is more consistent to adopt the final boundary analysis. Further, the analysis is more probable conceptually. That is, although incremental theme verbs lend themselves to telic and atelic clauses (e.g., he mowed the lawn, he mowed for hours), it is difficult to imagine the same pattern for closed scalar predicates (e.g., BEND.BAR and ICE.MELT) since these have minimal and maximal values. Based on the implications from the form and behavior of semelfactive, achievement, and accomplishment predicates, deceleration to a stop reflects a final boundary in ThSL predicates, not telicity.

117

Wilbur’s (2008) analysis of the Extent morpheme also has several issues. These are addressed first, followed by evidence from ThSL that path marking does not indicate temporal extent.

First, Wilbur (2008, pp. 242–243) provides the examples SICK-RESULTATIVE (i.e., ‘get sick’) and LOOK.AT as predicates with the Extent morpheme. Figure 22 shows the sign for ‘get sick’ alongside the stative form, SICK. Both images come from Klima and Bellugi (1979, p. 265).

Figure 22 ASL signs SICK (left) and BECOME.SICK (right)

The first frame of Figure 22 shows the unmodified form of SICK. The signer places the middle finger of one hand against her forehead and the middle finger of the other hand on her sternum. The only movement involved in the sign is transitional (i.e., placing the hands in position) and not considered a phonological part of the sign itself. In contrast, the sign BECOME.SICK shown on the right of Figure 22 has a path movement. The hand is moved swiftly along this path until it makes contact with the forehead. Klima and Bellugi (1979, p. 261) state that this form “focuses on the completion of a change of state.” That is, the changed movement coerces the state into an inchoative event meaning ‘get sick’. The form is used in instantaneous clauses such as ‘yesterday at noon John took sick’ (Klima & Bellugi, 1979, p. 267). Note that the punctual time frame ‘at noon’ indicates an instantaneous event, ‘get sick’. Despite this explanation from the original research (Klima & Bellugi, 1979), Wilbur (2008, p. 243) states that the added path movement indicates “the passage of significant time” and makes the event durative (e.g., ‘became sick over an extended period’). However, the claim is unsupported (apart from EVH internal postulations), conflicts with the description and analysis of the original work (Klima & Bellugi, 1979), and is incongruent with the fact that the form co-occurs within an instantaneous time frame. Her conclusion that the movement marks temporal extent is thus questionable.

118

Another sign used by Wilbur (2008, p. 244) to illustrate both the EndState and Extent morphemes is the sign LOOK.AT. Figure 23 shows the sign.

Figure 23 ASL sign LOOK.AT (Wilbur, 2008, p. 244)

Figure 23 illustrates the ASL sign LOOK.AT. In this sign, the index and middle fingers are extended and bent at the first knuckle so that the tips of the extended fingers point forward. The signer holds her hand at about chin-height and moves it forward before ending in a sudden stop. Wilbur (2008, p. 244) claims the sign is both telic and durative – an accomplishment. However, this is unusual. Smith (1997, p. 56) observes that verbs of perception can take on both stative and dynamic qualities. As dynamic events, they can be interpreted as activities (Smith, 1997, p. 57) such as he looked at the moon (for hours). This event would be durative, but atelic. Alternatively, such verbs can also be inchoative (Smith, 1997, p. 56) as in suddenly, he looked at her. If deceleration to a stop marks a final boundary, then the inchoative interpretation of LOOK.AT is more likely (neither states nor activities have inherent boundaries). Since inchoatives are instantaneous events (achievements), then it would follow that the path movement in LOOK.AT is not marking duration. However, based on the internal assumptions of the EVH, Wilbur (2008, p. 244) states that the sign is durative. She then uses the durative analysis of the movement to support the EVH. The result is a process of circular reasoning.

The external tests used for ThSL predicates avoid this snare and allow an objective analysis of form-meaning relationships. Concerning the proposed Extent morpheme, the test results for ERASE are particularly salient since the sign has the same movement used in GET.SICK and LOOK.AT. Figure 24 shows the sign ERASE.

119

Figure 24 ThSL sign ERASE

The sign ERASE in Figure 24 is formed by holding the non-dominant hand open with the fingers together and thumb extended. The signer also extends the thumb of his dominant hand and closes his remaining fingers into a fist. The pad of the thumb is placed on the palm of the non-dominant hand and moved swiftly along it. The movement continues well past the fingertips, utilizing the same straight path movement seen in GET.SICK and LOOK.AT. Conceptually, the sign is non-spatial. Thus, according to Wilbur (2008), the movement must convey temporal duration. However, as Section 3.6.2 shows, ERASE is [-durative]. Thus, path movement (as defined by Wilbur) in non-spatial predicates is not an indicator of temporal extent in ThSL. Instead, Supalla and Newport’s (1978, p. 103) analysis that single movements indicate a single, punctual, or perfective action again provides a better account of the data.

Further investigation in additional sign languages would add valuable insight into the relationship between semantic features and the phonological forms of sign language predicates. For the investigation to be objective, however, the featural values of predicates must be determined by external tests. Once these values are independently determined, the form of the signs can be explored to determine potential form-meaning relationships.

5.4 Testing for situation types in other sign languages This section presents guidelines for future studies of situation types in sign languages. These guidelines should facilitate test selection and data elicitation, making the process easier for researchers and informants alike. Thoughts on identifying tests are presented first, followed by a description of the potential forms a test may take (i.e., adverbial constituents versus verbal modification). Afterward, two approaches for combining the tests are presented.

120

First, identifying practical tests is a process of trial and error. In order to ease this process, a list of potential tests should be compiled. Then, a small set of predicates can be run through these to determine which tests work well. Although a single predicate from each situation type would potentially suffice for this step, it is recommended that two or three be prepared. Then, if a predicate does not belong to the anticipated situation type, the researcher will have ‘back up’ predicates readily available.

Several factors may make a test ineffective for the target languages. For example, the test may 1) resist adaptation (e.g., ‘what happened’ in ThSL), 2) have too many collocational restrictions (e.g., STRONG and NO.FEELING), or 3) not have the same co-occurrence restrictions as other languages (e.g., predicates from all situation types embedded under verbs of perception in ThSL). Ineffective tests can be omitted from future sessions.

When selecting predicates for both initial and expanded testing, it is prudent to consider several factors. First, the predicates should be central examples of the situation type (e.g., ‘walk’ or ‘run’ for activities) rather than potentially borderline cases (e.g., verbs of position like ‘sit’ or ‘stand’ or verbs of perception like ‘see’ and ‘smell’). Predicates that take volitional participants are also preferable since they work with a greater number of tests. Regarding specific situation types, include both stage-level and individual-level stative predicates in the initial testing to identify which tests distinguish the two types. For example, Test 3, 4, and 6 in ThSL served this purpose. For activities, intransitive verbs are more straightforward to work with than transitive ones since the specificity of the object is not a factor. Accomplishments that involve closed scales (e.g., dry, melt) provide less ambiguous results than accomplishments involving open scales (e.g., widen, expand). Specifically, predicates from the latter category may be either telic or atelic, depending on pragmatic factors. This flexibility leads to mixed results with tests for telicity. Thus, closed scale predicates are preferable.

Irreversible achievements are also likely to produce more contrastive results than reversible ones. For instance, ThSL reversible achievements patterned like semelfactives by occurring felicitously with MANY.TIMES (see Section 3.6.3). Irreversible achievements were infelicitous, making the [+telic] value of such predicates more evident. Next, semelfactive signs may involve a single or double movement. Those signs with a double movement in ThSL (CLAP and KNOCK.ON.DOOR) were coerced more easily into activities. If this is a cross-linguistic pattern for sign

121

languages, then semelfactives involving a single movement will be simpler to test. Finally, signs like JUMP can express achievements since signing space represents real space. For example, in ThSL, the movement for the sign ‘jump’ can be altered to mean ‘jump from here to there’, ‘jump up and come down’, ‘jump onto something’, and so forth. Thus, consider the location and movement of potential semelfactives in advance to make sure they are not indicating locations and, hence, encoding telic events.

When choosing predicates, consider their behavior in reduplicated form. Certain predicates are nominalized when repeated (e.g., SIT becomes CHAIR in ASL). Nominalization occurred with the ThSL sign OPEN.CAN since both the multi-event form ‘open cans’ and the nominalized form ‘tin can’ use a repeated movement. The meaning of the sign is clear from the context and non-manual features. However, informants often switched between the forms without realizing it, making elicitation more complicated. Predicates that did not undergo nominalization were less challenging to test. Selecting predicates that fall within these guidelines will make it simpler to identify useful tests.

The tests themselves may involve adverbials, verbal modifications, or both. The current study focused on tests from the first category. The reason for this was to keep the verb form consistent throughout testing since alteration of the form can change the situation type of the predicate (see Klima & Bellugi, 1979, Chapter 11). However, adverbials often caused the verb form to shift (e.g., INTERRUPT often caused the uncompleted form of the verb to be used while SLOW caused it to be articulated more slowly). This effect defeated the purpose of using adverbial tests. Starting with tests that use verbal modification may be more straightforward.

Potential tests involving verbal modification include manner morphemes, unrealized inceptives, and commands. First, manner morphemes are used to identify both dynamism and duration. They may be manifested as mouth morphemes, such as the ASL morphemes meaning ‘carelessly’ or ‘in a regular manner’, or modifications of the manual component, such as slower articulation to express ‘slowly’. Tests using these morphemes should be incompatible with states and instantaneous events. Second, if the sign language has unrealized inceptive forms of predicates, recruit these forms to test for dynamism. States should be incompatible with the form, and events should be compatible. Third, if commands are marked non-manually in the language (e.g., by a firm, clause-final head nod as in ThSL), the non-manual

122

indicator can be used to identify dynamism. Again, states should be incompatible with the command form, and dynamic events should be compatible.

Once practical tests and predicates have been identified, a final consideration is how the tests are applied. The researcher can take two potential approaches: hierarchical or binary.8 Although the approaches differ subtly, the one chosen determines which tests are selected, how they are applied, and how the results are presented. Further, depending on which tests work in the language, one approach may be more feasible than the other.

First, a hierarchical approach uses a flow-chart process for testing predicates. That is, the behavior of the predicate with the previous test determines which test to apply next. Figure 25 illustrates this method. Note that italicized words in grey font indicate when to apply a test.

Figure 25 Hierarchical approach to identifying situation types

Figure 25 shows the test-flow for a hierarchical approach. The first test identifies the dynamic value of the predicate. If the predicate is stative, then test again to determine if it is a stage-level state (transitory) or individual-level state (non- transitory). On the other hand, if it is dynamic, test to determine if it is durative or instantaneous. If it is durative, its telic value is then tested. If it is instantaneous, either use the same test or a different one to determine its telicity.

8 This is based on my own observations. Although other researchers do not identify their approach explicitly, their selection and application of tests tend to align with one of the approaches presented here.

123

A benefit of the hierarchical approach is that it avoids putting each predicate through each test. For example, if a predicate tests out as a state, it is then tested to determine if it is a stage-level or individual-level state. It is not tested for duration or telicity. A potential disadvantage to the approach is that the behaviors of predicates cannot be compared in the same test environment across situation types.

In contrast, a binary approach puts each predicate through each test, combining the results to identify the corresponding situation type of the predicate. Table 24 demonstrates.

Table 24 Binary approach to identifying situation types

Test for dynamism Test for duration Test for telicity Situation type

Individual-level [-dynamic] [+durative] [-telic]  state

[-dynamic] [+durative] [-telic]  Stage-level state

[+dynamic] [+durative] [-telic]  Activity

[+dynamic] [+durative] [+telic]  Accomplishment

[+dynamic] [-durative] [+telic]  Achievement

[+dynamic] [-durative] [-telic]  Semelfactive

In the binary approach that Table 24 illustrates, each predicate used with each test. The results (e.g., [+dynamic], [+durative], [-telic]) are then combined to identify the situation type of the predicate. Since each predicate is used with each test, the tests can be conducted in any order. The pros and cons of the approach are the opposite of the hierarchical one. Namely, the testing is less streamlined since each predicate is placed in each test frame. However, the comprehensive application compares the behavior of predicates across test environments.

The two approaches differ primarily in how they distinguish 1) stage-level from individual-level states and 2) distinguish telic events from atelic ones. For a hierarchical approach, a single test identifies types of states. In a binary approach, the results for tests of duration and telicity may serve this purpose. Second, regarding telicity, the hierarchical approach may use two separate tests to identify telicity in durative and instantaneous events. A binary approach uses a single test to identify the telicity of both event types. Because the tests available in the language may be limited, the researcher may find it necessary to choose one approach over

124

the other. If enough tests are available for both approaches, the researcher can choose the one that better suits their research interests. Specifically, those who wish to investigate a large number of predicates may find the hierarchical approach more efficient. Researchers interested in how predicates from different situation types behave in the same environment may find the binary approach more suitable.

5.5 Directions for further research in ThSL The current study has only touched the surface of situation types in Thai Sign Language. It has focused on which tests identify featural values in ThSL and applied these to a set of predicates. However, the process of testing these predicates has led to several interesting observations concerning the phonological movement of signs and the sub-event structure of situation types. Section 5.5.1 discusses these observations and directions for further research. Section 5.5.2 presents additional topics for further study.

5.5.1 Phonological movement and sub-event structure Although Supalla and Newport’s (1978, p. 103) observation that a single movement indicates a single, punctual, or perfective action aligns with the available ThSL data, their corresponding observation that repeated motion marks durative or iterative events was less applicable. Instead, the phonological movement of predicates was sensitive to the finer details of the sub-event structure of events. Due to the iconic origins of many predicate signs, this is not surprising.

These sub-event structures were most relevant for durative events (activities and accomplishments) since they involve multiple stages. In ThSL, the phonological form of these signs reflected the internal stages of the events they expressed. The reflections fell along a continuum ranging from distinct sub-events to indistinct, gradient change. First, the ThSL activity predicates – DRIVE.MOTORCYCLE, PAINT, RIDE.BICYCLE, RUN, and SWIM – all involved repeated phonological movements. Each movement reflected an internal stage of the event (e.g., the brush strokes in painting or the pedal strokes in riding a bike). The movements fell along a continuum from clear, individual cycles to finely grained repetitions that were almost gradient in nature. For example, RIDE.BICYCLE has a phonologically clear cyclic movement, as Figure 26 shows.

125

Figure 26 ThSL sign RIDE.BICYCLE

In Figure 26, the sign RIDE.BICYCLE is formed by holding both hands in loose fists, one slightly higher than the other, with palms facing down. The upper hand then moves forward and down while the lower hand moves backward and up as if pedaling a bicycle. The clear cycles involved phonologically and conceptually in the event contrast with the more finely grained movement involved in DRIVE.MOTORCYCLE. Figure 27 shows this sign.

Figure 27 ThSL sign DRIVE.MOTORCYCLE

For DRIVE.MOTORCYCLE in Figure 27, the signer holds his fists at chest-height as if grasping the handles of a motorcycle. He then moves his fists slightly back and forth, reflecting the vibrating movement experienced while driving a motorcycle. Although individual cycles can still be identified, the movement is slighter than that of RIDE.BICYCLE. Since movement in the real world falls along a continuum (e.g., compare each stroke in swimming to each vibration in driving a motorcycle), signs with iconic origins are expected to reflect this continuum in their phonological form.

These two signs present another interesting aspect of sign language predicates. Specifically, riding a bike and driving a motorcycle both involve the movement of a two-wheeled vehicle. However, the signs utilize different elements of movement inherent to the event to express meaning. In DRIVE.MOTORCYCLE, it is the movement experienced by the driver’s hands; in RIDE.BICYCLE, it is the movement

126

experienced by the rider’s feet; in classifier predicates portraying these events (where the shape of the hand represents the shape of the vehicle), it is the movement experienced by the vehicle as a whole. These variations create an interesting parallel between viewpoint aspect and the lexical expression of predicates in sign languages. Namely, just as viewpoint aspects allow different portions of the event to come into focus, lexical expression allows different movements involved in the event to come into focus.

Even more finely grained than the movements of these predicates (conceptually and phonologically) is the movement of durative, scalar predicates. These predicates did not align with Supalla and Newport’s (1978, p. 103) observation that duration is marked by repeated movement. Instead, most of the ThSL accomplishment predicates (BEND.BAR, STRAIGHTEN.BAR, ICE.MELT, and SKY.DARKEN) had only a single, smooth motion (see Appendix A), even though they encoded durative events (see Section 3.5.2‎ ).9 This pattern suggests a correspondence between the real-world internal structure of an event and the phonological movement of the sign. Cyclic movements express events with cyclic movements; gradient movement expresses events with gradient movement.

The connection between the movement of a predicate sign and the internal structure of events lends support to Kuhn’s (2017) iconic analysis of phonological movement. Further study involving a more extensive selection of predicates – each analyzed according to their inherent featural values and internal structure – would provide further evidence for or against this conclusion.

The link between the phonological movement of a dynamic predicate and its conceptual structure also raises an interesting question. If phonological movement in event predicates reflects real-world movement, shouldn’t stative predicates lack phonological movement altogether? In the ThSL data, this is not the case. Instead, stative predicates were expressed using a wide range of movement patterns (see Appendix A). This directly contrasts with Wilbur’s (2008, p. 231) proposal that the handshape, location, and orientation of stative predicates remains the same throughout the articulation of the sign. If these phonological movements do not reflect the internal stages of the situation, then what, if anything, do they express?

9 The exception is GROW.BEARD (see Appendix A for picture). The signer places his hand near his chin. His fingers are curled in slightly as if holding a small ball. He then bends and straightens his fingers slightly as he moves his hand downward. Thus, the articulation involves both repeated “scrunching” movement and path movement. However, I would argue that the repeated movement reflects the shape of the object (a wavy or curly beard) while the smooth, downward path reflects the motion of the event.

127

Can the movements be used to group stative predicates into different sub-types based on their internal contour? If so, what conceptual distinctions lie behind these groupings? Alternatively, do different movement patterns represent different viewpoint aspects for stative predicates, with some including entry into the state and others viewing the state only from within? Or are the phonological movement patterns of stative predicates entirely arbitrary? These questions remain to be explored.

5.5.2 Additional topics for further research A few additional topics present themselves for further research. First, when temporal adverbials (e.g., five minutes) were used with activity predicates (specifically RUN and PAINT), they either produced telic implications or the informants added constituents to make the clause overtly telic (see Section 4.4). This preference for a telic reading seems to be a result of using precise time frames, such as ‘thirty minutes’ or ‘three hours’. Less specific time frames (e.g., ‘for a long time’ or ‘for hours and hours’) may not lead to telic interpretations. Further study using less specific time frames would shed light on whether or not this is the case.

Second, informants varied in their judgments of when a predicate needed to be reduplicated. This was seen especially in testing with STILL.THERE and MANY.TIMES. The reason behind these differing judgments requires further investigation.

Finally, the current study has been limited to a single dimension of temporality – features inherent to the predicate. A broader study would investigate the interaction of these features with viewpoint aspects. Such an investigation would identify any morphological marking of viewpoint aspects. It would also look at the default temporal location of an event (if any) based on the aspectual viewpoint. Such work would provide a holistic treatment of how situations are encoded, portrayed, and temporally located in Thai Sign Language and facilitate future cross-linguistic comparison.

128

BIBLIOGRAPHY

Carlson, G. N. (1980). References to kinds in English. In Jorge Hankamer (Ed.) Outstanding dissertations in linguistic. New York: Garland Publishing.

Collins-Ahlgren, M. (1990). Spatial-locative predicates in Thai Sign Language. In C. Lucas (Ed.) Sign language research: Theoretical issues (pp. 103–117). Washington, D.C.: Gallaudet University Press.

Comrie, B. (1976). Aspect: An introduction to the study of verbal aspect and related problems. Cambridge, United Kingdom: Cambridge University Press.

Croft, W. (2012). Verbs: Aspect and causal structure. New York: Oxford University Press.

Danthanavanich, S. (2008). A grammar of Thai Sign Language. (Unpublished doctoral dissertation). Mahidol University, Bangkok.

Davidson, K., Kocab, A., Sims, A. D., & Wagner, L. (2018). The relationship between verbal form and event structure in sign languages. Unpublished manuscript.

Dowty, D. (1979). Word meaning and Montague grammar: The semantics of verbs and times in generative semantics and in Montague’s PTQ. Dordrecht, the Netherlands: Kluwer Academic Publishers.

Eberle, S. E. (2013). Locative expressions in signed languages: A cross-linguistic comparison (Master's thesis, University of North Dakota). Retrieved from https://arts-sciences.und.edu/academics/summer-institute-of- linguistics/theses/_files/docs/2013-eberle-sarah.pdf

ELAN (Version 5.5). (2019). Retrieved from https://tla.mpi.nl/tools/tla-tools/elan/

Fennig, C. D., & Simons, G. F. (Eds.). (2018). Ethnologue: Languages of the World (21st ed.). Dallas, TX: SIL International. Retrieved from https://www.ethnologue.com.

Frawley, W. (1992). Linguistic semantics. Hillsdale, New Jersey: Lawrence Erlbaum Associates Inc.

129

HandBrake (Version 1.2.1). (2019). Retrieved from https://handbrake.fr

Hay, J., Kennedy, C., & Levin, B. (1999). Scalar structure underlies telicity in “degree achievements”. In T. Matthews & D. Strolovitch (Eds.), Proceedings of SALT 9 (pp. 127–144). Ithaca, New York: Cornell University, CLC Publications.

Kennedy, C., & McNally, L. (2005). Scale structure and the semantic typology of gradable predicates. Language, 81(2), 345–381.

Kennedy, Christopher, & Levin, B. (2008). Measure of change: The adjectival core of degree achievements. In L. McNally & C. Kennedy (Eds.), Adjectives and Adverbs: Syntax, semantics and discourse (pp. 156-182). Oxford and New York: Oxford University Press.

Klima, E. S., & Bellugi, U. (1979). The signs of language. Cambridge, MA: Harvard University Press.

Krifka, M. (2000). Alternatives for aspectual particles: Semantics of still and already. In L. Conathan, J. Good, D. Kavitskaya, A. Wulf, & A. Lu (Eds.) 26th Annual Meeting of the Berkeley Linguistics Society: General Session and Parasession on Aspect (pp. 401–412). University of California at Berkeley: Berkeley Linguistic Society.

Kuhn, J. (2017). Telicity and iconic scales in ASL. Retrieved from https://www.semanticsarchive.net/Archive/Tc2ZmRiO/kuhn-ASL-telicity.pdf

Kullavanijaya, P., & Thepkanjana, K. (2001). Devices for forming entity-denoting signs in Thai Sign Language. In K. Tingsabadh & A. S. Abramson (Eds.), Essays in Tai linguistics (pp. 91–116). Bangkok, Thailand: Chulalongkorn University Press.

Lakoff, G. (1965). On the nature of syntactic irregularity. Cambridge, MA: Harvard University.

Liddell, S. (1984). Unrealized-inceptive aspect in American Sign Language: Feature insertion in syllabic frames. In D. Testen, V. Mishra, & J. Drogo (Eds.), Papers from the 20th regional meeting of the Chicago Linguistic Society (pp. 257–270). Chicago, IL: University of Chicago Press.

130

Loebner, S. (1989). German schon -erst- noch: An integrated analysis. Linguistics and Philosophy, 12(2), 167–212.

Lumtien, N. (1997). Thai Sign Language (TSL) structure of the Deaf students in the schools for the Deaf (Unpublished master’s thesis). Srinakharinwirot University, Bangkok, Thailand.

Meir, I., Padden, C., Aronoff, M., & Sandler, W. (2007). Re-thinking sign language verb classes: The body as subject. Journal of Linguistics, 43(3), 531–563.

Ministry of Education. Promulgation for recognition of Thai Sign Language as a national language for Deaf people. (1999).

Niwatapant, C. (2006). A contrastive analysis of verbs in Thai Sign Language and standard Thai (Unpublished master’s thesis). Kasetsart University, Bangkok, Thailand.

Niwatapant, C., & Tumtavitikul, A. (2005). The verb “open” in Thai Sign Language. Journal of Ratchasuda College for Research and Development of Persons with Disabilities, 1(2), 12–27.

Padden, C. (1988). Interaction of morphology and syntax in American Sign Language. New York, NY: Garland Press.

Parsons, T. (1990). Events in the semantics of English. Cambridge, MA: MIT Press.

Pelkey, J. (2004). Phowa verbal semantics (Master's thesis, Payap University). Retrieved from http://newtdc.thailis.or.th/docview.aspx?tdcid=344198

Phillips, A., & Thiengburanathum, P. (2007). Verb classes in Thai. Language and Linguistics, 8(1), 167–191.

Pradapwattanangune, S. (1998). การปฏิเสธความในภาษามือไทยของโรงเรียนเศรษฐเสถียร Denial in the Thai Sign Language of Sethsatian school (Master’s thesis, Chulalongkorn University). Retrieved from http://cuir.car.chula.ac.th/handle/123456789/11477

Rappaport Hovav, M. (2006). Lexicalized meaning and the internal temporal structure of events. Unpublished manuscript. Jerusalem: the Hebrew University of Jerusalem.

131

Rathmann, C. G. (2005). Event structure in American Sign Language (Doctoral dissertation, University of Texas, Austin). Retrieved from https://pdfs.semanticscholar.org/0598/29c3602e9ffb9f05ed58332b9e351b1 8a656.pdf

Smith, C. S. (1997). The parameter of aspect (2nd ed.). Dordrecht, the Netherlands: Kluwer Academic Publishers.

Supalla, T., & Newport, E. (1978). How many seats in a chair? The derivation of nouns and verbs in American Sign Language. In P. Siple (Ed.), Understanding language through sign language research (pp. 91–132). New York, NY: Academic Press.

Thailand Hearing Disability Statistics. (2017). Retrieved from http://nadt.or.th/en/statistics

Tumtavitikul, A., & Niwatapant, C. (2008). Syllables and words in Thai Sign Language. Journal of Ratchasuda College for Research and Development of Persons with Disabilities, 4(1), 95–113.

Tumtavitikul, A., Niwatapant, C., & Dill, P. (2009). Classifiers in Thai Sign Language. SKASE Journal of Theoretical Linguistics, 6(1), 27–44.

Van Valin, R. D. (2005). Exploring the syntax-semantics interface. Cambridge, United Kingdom: Cambridge University Press.

Vicar, W. (2001). American Sign Language: “postpone.” Retrieved from https://www.lifeprint.com/asl101/pages-signs/p/postpone.htm

Wallace, C. (2019). Spatial relations along the in-on continuum in Thai Sign Language. Journal of the Southeast Asian Linguistics Society, 12(1), 163–178.

Wallace, C. (2020). Testing for situation types in Thai Sign Language. Journal of the Southeast Asian Linguistics Society, 13(1), 108-128.

Wilbur, R. B. (2008). Complex predicates involving events, time and aspect: Is this why sign languages look so similar? In J. Quer (Ed.), Signs of the time: Selected papers from TISLR 2004 (pp. 217–250). Hamburg, Germany: Signum.

132

Woodward, J. (1996). Modern Standard Thai Sign Language, influence from ASL, and its relationship to original Thai Sign varieties. Sign Language Studies, 92(Fall), 227–252.

Woodward, J. (2003). Sign languages and Deaf identities in Thailand and Vietnam. In L. Monaghan, C. Schmaling, K. Nakamura, & G. H. Turner (Eds.), Many ways to be deaf: International variation in deaf communities (pp. 283–301). Washington, D.C.: Gallaudet University Press.

Wudthayagorn, J. (1998). Yes-no questions in the Thai Sign Language of Sethsatian school (Unpublished master’s thesis). Chulalongkorn University, Bangkok, Thailand.

Xiao, R., & McEnery, T. (2004). Aspect in : A corpus-based study. Amsterdam, the Netherlands: John Benjamin’s Publishing Company.

133

APPENDIX A PREDICATES IN THAI SIGN LANGUAGE

The following still shots illustrate the ThSL predicates used in the study. They are organized by situation type. Within the situation type, the predicates are arranged in alphabetical order by their English representation (gloss).

1. Individual-level State predicates

BEAUTIFUL

HONEST – Variant 110

10 Variant 1 is the form of HONEST that was used for testing with the Original Group. Other Deaf (personal friends) observed that this variation is more common among Christian Deaf. Variant 2 is more common outside the Christian community.

134

HONEST – Variant 2

SHORT

SMART

TALL

135

2. Stage-level State predicates

AGITATED

ANGRY

BORED

SAD

136

SICK

3. Activity predicates

DRAW/PAINT11

DRIVE.MOTORBIKE

PAINT

11 Informants in Group A used the sign DRAW for both ‘draw’ and ‘paint’.

137

RIDE.BICYCLE

RUN

SWIM

4. Accomplishment predicates

BEND.BAR

138

GROW.BEARD

ICE.MELT

SKY.DARKEN

STRAIGHTEN.BAR

139

5. Achievement predicates

BLOCK (on Facebook)

DUNK (under water)

ERASE (Line)

EXPLODE

140

OPEN.CAN

6. Semelfactive predicates

CLAP

KNOCK.ON.DOOR

PUNCH.PERSON

141

SHOUT

TURN.OFF.LIGHT

142

APPENDIX B INFORMANT BACKGROUND QUESTIONNAIRE

The following questionnaire was used to gather background information from the study’s language informants over video chat. Elicitation was done in ThSL.

Biographical Information Questionnaire

Informant Number: Date:

1. First name/Last Name: 2. Age: 3. Gender: 4. Occupation: 5. Which school(s) did you attend? For which grades? 6. Did you attend university? 7. If yes, what was your major? 8. Where were you born? Where have you lived since then? From what ages? 9. Which one do you identify as: deaf or hard of hearing? 10. Do you have Deaf family members? 11. If yes, who? 12. Does anyone in your family know Thai Sign Language? 13. If yes, where did they learn Thai Sign Language? 14. How old were you when you learned Thai Sign Language? 15. What other sign languages do you know?

143

APPENDIX C INFORMANT BACKGROUND INFORMATION Table C1 Informants’ general background # Status Sex Age Occupation ThSL age Other Sign Languages DFM SFM Original Group 01 Deaf M 30 Office staff 9 some ASL Y Y 02 Deaf M 31 Office staff 8/9 some ASL, some root N N 03 Deaf M 27 Translator 9 none N N 04 Deaf F 25 Admin. Officer 8 some ASL N N 05 Deaf M 30 Courier 9 some ASL Y Y 06 HoH F 27 Housekeeper 8/9 none N N Group A 07 Deaf F 33 Accountant 6/7 some ASL N N 08 Deaf M 39 Office Staff 6 none N N 09 Deaf F 39 Teacher 6/7 some ASL N N 10 Deaf M 33 Teacher <5 none Y Y Group B 11 HoH M 28 Teacher 12 none Y N 12 Deaf M 39 Teacher 10/11 none Y Y 13 Deaf F 36 Teacher 7 some ASL, some root N N 14 Deaf M 36 Job Coach 6 some ASL, JSL, root Y N Table Guide: # - Informant number Status – Deaf or Hard of Hearing (HoH) Sex – Male (M) or female (F) Age – Age of the informant when background information was collected Occupation – Informant’s primary occupation when background information was collected ThSL Age – Age at which the signer began acquiring ThSL Other Sign Languages – Other sign languages that the informant knows DFM – Deaf/deaf family member or relative SFM – Signing family member or relative (either deaf or hearing) Root – A type of highly mimetic/iconic signing used to communicate across language barriers JSL – Japanese Sign Language

Other Notes: 1. Informants 01 and 05 are twins. 2. Informant 10 learned ThSL from an older Deaf sister before attending a Deaf school. Informant 12 has a Deaf brother seven years younger who also knows ThSL. 3. Informants 11 and 14 have deaf relatives who do not sign

144

Table C1 – Informants’ educational background # K P P 1-6 Location M M 1-6 Location Post secondary PS location Original Group 01 MS 1-6 Prachin Buri SD 1-6 Prachin Buri SD Ratchasuda College Bangkok 1-2 Mainstreamed Vocational training Prachin Buri 02 MS 1-3 Prachin Buri SD 3-6 Prachin Buri SD Rajabhat University Nakhon Ratchasima 03 - 1-6 Nakhorn Pathom SD 1-3 Nakhorn Pathom SD Vocational training Suphanburi 1-3 Mainstreamed Nonthaburi 04 MS 1-6 Nonthaburi SD Nonthaburi 1-6 Nonthaburi SD University 1-2 Mainstreamed 05 MS 1-6 Prachin Buri SD Ratchasuda College Bangkok 2-6 Prachin Buri SD 06 SD 1-6 Chonburi SD N N/A N/A N/A

Group A 145

07 SD 1-6 Anusarnsunthorn SD, Chiang Mai 1-6 Anusarnsunthorn SD, Chiang Mai Rajabhat University Chiang Mai

La-Orutis Demonstration School Anusarnsunthorn SD, Chiang Mai 1-3 1-3 08 - Bangkok Rajabhat University Chiang Mai 4-6 Anusarnsunthorn SD, Chiang Mai 4-6 La-Orutis Demonstration School, Bangkok 09 SD 1-6 Anusarnsunthorn SD, Chiang Mai 1-6 Anusarnsunthorn SD, Chiang Mai Ratchasuda College Bangkok 1-3 Songkhla SD 10 SD 1-6 Songkhla SD Ratchasuda College Bangkok 4-6 Thungmahamek SD, Bangkok Group B 11 - 1-6 Mainstreamed 1-6 Anusarnsunthorn SD, Chiang Mai Rajabhat University Chiang Mai 12 SD 1-6 Anusarnsunthorn SD, Chiang Mai 1-6 Anusarnsunthorn SD, Chiang Mai Rajabhat University Lampang 1-3 Anusarnsunthorn SD, Chiang Mai 13 - 1-6 Anusarnsunthorn SD, Chiang Mai Non-formal educational school, Nakhorn Ratchasuda College Nakhorn Pathom 4-6 Pathom 14 SD 1-6 Anusarnsunthorn SD, Chiang Mai 1-6 Anusarnsunthorn SD, Chiang Mai N/A N/A

147

Table C1 Guide: # – Informant number K – Kindergarden MS – Mainstreamed with hearing students in hearing school SD – School for the Deaf. If the location is not included in the name of the school, then the province is listed after the name (E.g., Anusarnsunthorn SD, Chiang Mai). P – Grades 1 – 6 M – Grades 7 – 12 Post secondary – Name of post secondary school or vocational training PS location – Province where post secondary school or vocational training was located.

Other Notes: Informants 04 and 05 repeated grades.

146

RESUME

Name: Cassie Wallace

Date of Birth: 29 December 1988

Place of Birth: United States of America

Institutions Attended: 2010, BA Christian Ministry, Boise Bible College (Boise, Idaho, USA).

2010, 2011, 2016 – University of North Dakota (Grand Forks, North Dakota, USA).

2017-2020, Master of Arts in Linguistics, Payap University (Chiang Mai, Thailand)

147