UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Laryngeal phenomena in Tahltan Bob, Tanya Marie 1999

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1999-0458.pdf [ 6.76MB ]
Metadata
JSON: 831-1.0089100.json
JSON-LD: 831-1.0089100-ld.json
RDF/XML (Pretty): 831-1.0089100-rdf.xml
RDF/JSON: 831-1.0089100-rdf.json
Turtle: 831-1.0089100-turtle.txt
N-Triples: 831-1.0089100-rdf-ntriples.txt
Original Record: 831-1.0089100-source.json
Full Text
831-1.0089100-fulltext.txt
Citation
831-1.0089100.ris

Full Text

LARYNGEAL PHENOMENA IN TAHLTAN by TANYA MARIE BOB B. A., The University of British Columbia, 1997 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES Department of Linguistics We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October 1999 © Tanya Marie Bob, 1999 in presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of U n C ^ A ^ E S The University of British Columbia Vancouver, Canada Date OctoV-irrr \S ,. \0 i f t f t DE-6 (2788) Abstract This thesis investigates the phonetic and phonological properties of laryngeal distinctions in the consonant inventory of Tahltan, a northern Athapaskan language. This thesis does not examine the phonetic properties of all Tahltan consonants. Instead, this thesis focuses on the phonetic acoustic properties of plain stop consonants, which have been described inconsistently in Tahltan, to determine their laryngeal specification. This thesis also examines the observed patterns of behavior governing syllable structure to help determine the laryngeal specification of consonants in Tahltan. In addition, several morphophonemic processes are examined to determine the phonological laryngeal specification of consonants in Tahltan. Based on the phonetic findings, and observed patterns of behavior governing syllable structure, I will argue that stop consonants in Tahltan exhibit four laryngeal articulations: voiced, voiceless unaspirated voiceless aspirated and glottalized. Based on the morphophonemic evidence, I will argue that fricative consonants exhibit two laryngeal articulations: voiced and voiceless. Furthermore, I will argue that glottal stop is specified for the laryngeal specification [constricted glottis] (henceforth [CG]) and that [h] is specified for the laryngeal specification [spread glottis] (henceforth [SG]). Table of Contents Page Abstract i i Table o f Contents i i i List o f Figures v Acknowledgements v i Chapter One Introduction 1 1.1 Language Information 1 1.2 Tahltan Consonant Inventory 1 1.3 Dual Mechanism Hypothesis 2 1.4 Laryngeal Mechanism Hypothesis 3 1.5 Outline 4 Chapter Two Phonetic Analysis o f Stops 6 2.1 Objective 6 2.2 Background 6 2.2.1 Tahltan Stop Consonants 6 2.2.2 Voiced and Voiceless Unaspirated Stops 8 2.2.3 Acoustic Properties o f Voiced and Voiceless unaspirated stops 9 2.3 Methods 10 2.4 Results 13 2.4.1 Post-Sibilant Stop Consonants 13 2.4.2 Post-Nasal Stop Consonants 15 2.4.3 Intervocalic Stop Consonants 17 2.4.4 Summary 19 2.5 Discussion 19 2.6 Conclusion 24 Chapter Three Syllable Structure Constraint 25 3.1 Theoretical Assumptions 25 3.2 Background 26 3.2.1 Tahltan Consonant Inventory 26 3.2.2 Orthographic Conventions 26 3.2.3 Tahltan Syllable Structure 27 3.3 Data 30 3.3.1 Consonants in Syllable-Initial Position 30 3.3.2 Consonants in Syllable-Final Position 31 3.3.3 Summary 32 3.4 Syllable Structure Constraint 32 3.5 Conclusion 33 Chapter Four Classifiers 34 4.1 The Classifier Prefixes and Laryngeal Specification 34 4.2 Function 34 4.3 Tahltan Verb 35 4.4 Theoretical Assumptions 36 4.5 Hypothesis 36 iii 4.6 h-C lass i f i e r 37 4.6.1 M o r p h o p h o n e m i c E f fec ts 37 4.6.2 A n a l y s i s , 41 4.6.3 S u m m a r y 44 4.7 1-Classi f ier 44 4.7.1 M o r p h o p h o n e m i c E f fec ts 44 4.7.2 A n a l y s i s 47 4.7.3 S u m m a r y 48 4.8 d - C l a s s i f i e r 48 4.8.1 M o r p h o p h o n e m i c E f fec ts 48 4.8.2 A n a l y s i s 51 4.8.3 Summary 53 4.9 0 - C l a s s i f i e r 53 4.9.1 M o r p h o p h o n e m i c E f fec ts 55 4.9.2 A n a l y s i s 55 4.9.3 Summary 56 4.10 C o n c l u s i o n 56 Chapter F i v e N o u n Stem-Init ial and Stem-Fina l V o i c i n g A l te rna t ions 57 5.1 Theoret i ca l A s s u m p t i o n s 57 5.2 Cont inuant V o i c i n g 57 5.2.1 Da ta 58 5.2.2 A n a l y s i s 60 5.2.3 Summary 63 5.3 Word-F ina l D e v o i c i n g 63 5.3.1 Da ta 64 5.3.2 A n a l y s i s 65 5.3.3 Stem-Fina l S top Consonants 66 5.3.4 S u m m a r y 67 5.4 C o n c l u s i o n 67 Chapter S i x M o r p h o p h o n e m i c E f fects in Op t ima l i t y Theo r y 68 6.1 Theore t i ca l A s s u m p t i o n s 68 6.2 Summary o f M o r p h o p h o n e m i c Processes 69 6.3 Constra ints and Const ra in t R a n k i n g 70 6.4 M o r p h o p h o n e m i c Processes in Op t ima l i t y Theo ry 75 6.4.1 1-classifier 75 6.4.2 d- c lass i f i e r 76 6.4.3 h- c lass i f i e r 77 6.4.4 Cont inuant V o i c i n g 80 6.4.5 Word-F ina l D e v o i c i n g 84 , 6.5 C o n c l u s i o n 85 Chapte r Seven C o n c l u s i o n 86 B i b l i o g r a p h y 87 A P P E N D I X A : E l i c i t a t i on 90 A P P E N D I X B: W a v e f o r m s and Spectrograms 92 iv List of Figures Page Figure (1) Post-sibilant [b]; [esba:ke] 'my moccasins' 92 Figure (2) Post-sibilant [d]; [esdihe] 'my grouse' 92 Figure (3) Post-sibilant [g]; [esgendame] 'my horse' 93 Figure (4) Post-nasal [b]; [inba:tke] 'your moccasins' 93 Figure (5) Post-nasal [g]; [ingawe] 'your drum' 94 Figure (6) Intervocalic [b]; [mebede] 'his food' 94 Figure (7) Intervocalic [d]; [medegane] 'his sockeye' 95 Figure (8) Intervocalic [g]; [megahe] 'his rabbit' 95 v Acknowledgements I wish to thank all the members of my thesis committee, Dr. Patricia Shaw, Dr. John Alderete and Dr. Doug Pulleyblank for all their support, encouragement and invaluable advice. I would like to extend many thanks to my advisor Patricia Shaw for graciously sharing her Tahltan fieldnotes and for her guidance through out my thesis. Again, I would like to extend my thanks to John Alderete for all the encouragement over the past year, for being such a good listener and for organizing our trip to Telegraph Creek. I would also like to extend my thanks to Doug Pulleyblank for being such an excellent professor over the years and for his support and encouragement. Furthermore, I would like to thank Dr. Guy Carden and Dr. Bryan Gick for their help with the phonetic analysis study. I would like to express my sincere appreciation to my friends and family in Telegraph Creek and Iskut, British Columbia, who graciously shared their knowledge of the Tahltan language and culture with me. I would also like to thank the Tahltan band council, who funded me while working toward my masters degree. I would also like to thank my friends and classmates for all their help and encouragement over the past two years. I especially would like to thank my friend Christine, who after linguistics 100 convinced me to take just one more year of linguistics! Finally, this thesis would not of been possible without the love and support of my family. Thank you, mom, dad and Yat. I've appreciated everything you've done for me. vi C h a p t e r O n e Introduction Th i s thesis investigates the phonet ic and phono log i ca l propert ies o f laryngeal d is t inc t ions in the consonant inventory o f Tah l tan , a northern A thapaskan language. T h i s thesis does not examine the phonet ic propert ies o f a l l Tah l tan consonants. Instead, this thesis examines the phonet ic acoust ic propert ies o f p l a in stop consonants , w h i c h have been descr ibed incons is tent ly in Tah l t an , to determine their la ryngeal spec i f i ca t ion . T h i s thesis a lso examines the observed patterns o f behav ior gove rn ing sy l l ab le structure to he lp determine the laryngeal spec i f i ca t ion o f consonants in Tah l tan . F i na l l y , several m o r p h o p h o n e m i c processes are examined to determine the phono log i ca l la ryngeal spec i f i ca t ion o f consonants in Tah l tan . 1.1 Language In format ion The A thapaskan language f a m i l y consists o f three subgroups, w h i c h are spread out across N o r t h A m e r i c a : No r the rn , Pac i f i c Coast and A p a c h e a n . The A p a c h e a n subgroup o f languages are spoken in the southwest o f the U n i t e d States, wh i l e the Pac i f i c Coas t languages are spoken a long the pac i f i c coast o f the Un i t ed States. The Nor the rn A thapaskan languages are spoken in the northwest o f the Un i t ed States and Canada . M o r e spec i f i ca l l y , the Nor the rn A thapaskan languages are spoken i n : inter ior A l a s k a , No r thwes t Terr i tor ies , Y u k o n , inter ior B r i t i sh C o l u m b i a , northern A lbe r t a , northern Saskatchewan and northern M a n i t o b a . T h i s thesis focuses on the northern A thapaskan language Tah l tan . Tah l tan is spoken by the F i rs t Na t i ons people w h o occupy the S t ik ine P lateau area o f northwestern B r i t i sh C o l u m b i a . Tah l tan is spoken in the present day commun i t i e s o f Te legraph Creek , Iskut and Dease L a k e . T h i s thesis presents data f r om Tah l tan speaker residents o f Te legraph C reek and Iskut, B r i t i sh C o l u m b i a . 1.2 Tah l tan Consonant Inventory The Tah l tan consonant inventory is presented in (1). S im i l a r to other A thapaskan languages, in Tah l tan , stop consonants exh ib i t a three-way laryngeal contrast, f r icat ives exh ib i t a two-way laryngeal contrast and there is a s ingle set o f sonorant consonants. 1 (1) Tah l tan Consonant Inventory labial coronal dorsal laryngeal labial alveolar inter-dental alveolar lateral palatal velar labio-velar uvular glottal a. P t t8 ts ti dz k kw q ? b. th teh tsh t i h ts h k h kwh q h c. t' t9' ts' t i ' ts' k' k w ' q' d. 9 s i s 5 X x w X h e. 6 z 1 z y Y Y w f. m n y w a. plain stop d. voiceless fricative b. voiceless aspirated stop e. voiced fricative c. glottalized stop f. sonorant In the A thapaskan l iterature, the three series o f stop consonants are t yp i ca l l y phonet i ca l l y descr ibed as vo ice less unaspirated (p la in stops), vo ice less aspirated and g lot ta l ized . The two series o f f r icat ive consonants are descr ibed as vo i c ed and vo ice less . In the Tah l tan l iterature, the series o f p l a in stop consonants have been descr ibed inconsistent ly . A l t h o u g h never the focus o f an invest igat ion, Tah l tan p la in stop consonants have been descr ibed as vo ice less unaspirated ( H a r d w i c k 1984) and as vo i c ed ( Shaw 1981 ; Shaw 1982; 1983; Na te r 1989) . 1 In Tah l t an , the r ema in ing series o f stop consonants are descr ibed as vo ice less aspirated and g lo t ta l ized , wh i l e the f r icat ives are descr ibed as vo ice less and vo i ced . These descr ipt ions are consistent w i t h the descr ipt ions in the A thapaskan l iterature. 1.3 D u a l M e c h a n i s m Hypothes i s The D u a l M e c h a n i s m Hypothes i s proposed by R i c e (1994) is a hypothes is w h i c h concerns the phono log i ca l laryngeal treatment o f the plain/aspirated stop contrast and the vo i ced/ vo ice less f r icat ive contrast in A thapaskan languages. R i c e (1994) argues fo r the D u a l M e c h a n i s m Hypothes i s based on var ied phono log i ca l ev idence f r o m S lave , N a v a j o and K o y u k o n . 2 U n d e r the D u a l M e c h a n i s m Hypothes i s , vo ice less unaspirated stops are spec i f i ed fo r the feature [stop] and unmarked fo r laryngeal features. T h e aspirated stops are spec i f i ed for the feature [stop] and the laryngeal feature [ SG ] . W i t h regards to the f r icat ives , vo ice less f r icat ives are unspec i f ied for 1 The phonetic symbol for voiceless unaspirated stops wi l l be used to represent this series o f stop consonants. 2 More specifically, Rice (1994) argues that there are lexical and postlexical processes in Athapaskan languages and that lexical processes support the Dual Mechanism Hypothesis, but postlexical do not. See Rice (1994) for details. 2 laryngeal features, wh i l e the vo i c ed fr icat ives are spec i f i ed for the la ryngea l feature [voice] . The D u a l M e c h a n i s m Hypothes i s is l a id out in (2). In (2), x is used to indicate w h i c h feature(s) each group o f consonants is spec i f i ed for. (2) D u a l M e c h a n i s m Hypothes i s (R i ce 1994) 3 voiceless voiceless voiceless voiced unaspirated aspirated fricative fricative Spread glottis X Voice X Stop x X A s prev ious l y ment ioned , in the A thapaskan literature stop consonants are descr ibed phonet i ca l l y as vo ice less unaspirated, vo ice less aspirated and g lo t ta l i zed , w h i l e f r icat ives are descr ibed as vo i ced and vo ice less . The D u a l M e c h a n i s m Hypothes i s , w h i c h concerns the phono log i ca l laryngeal spec i f i ca t ion o f stops and f r icat ives , is consistent w i th the phonet ic descr ipt ion o f A thapaskan consonants. 1.4 La r yngea l M e c h a n i s m Hypothes i s In this thesis, based on the phonet ic findings, I w i l l argue fo r rev i s ions to the D u a l M e c h a n i s m Hypothes i s (R i ce 1994). Reca l l that p l a in stop consonants have been descr ibed inconsistent ly in the Tah l tan l iterature. T h e y have been descr ibed as vo ice less unaspirated ( H a r d w i c k 1984) and as vo i c ed ( Shaw 1981 ; Shaw 1982; 1983; Na te r 1989). T o determine the phonet ic laryngeal spec i f i ca t ion o f Tah l tan p la in stop consonants, a phonet ic study, w h i c h examined the acoust ic propert ies o f p l a in stops, was conducted . The results f r om the phonet ic study indicate that there are vo i c ed and vo ice less unaspirated stops in the Tah l t an ; thus based on phonet ic ev idence, I w i l l argue for the add i t ion o f the category v o i c e d stop to the D u a l M e c h a n i s m Hypothes i s . The rev ised D u a l M e c h a n i s m Hypothes i s (henceforth L a r yngea l M e c h a n i s m Hypothes i s ) is l a id out in (3). O n c e aga in , x is used to indicate w h i c h feature(s) each group o f consonants is spec i f ied for. 3 Rice (1994) does not identify the laryngeal specifications o f glottalized stops, glottal stop and [h]; thus the specifications o f these consonants are not present in figure (2). 3 (3) La r yngea l M e c h a n i s m Hypothes i s ( rev ised f r o m R i c e 1994) voiced stop voiceless aspirated stops voiceless unaspirated stops voiceless fricatives voiced fricatives stop X X X voice X X SG X In the D u a l M e c h a n i s m Hypothes i s R i c e (1994) , R i c e does not ident i fy the laryngeal spec i f i ca t ions o f g lot ta l ized stops, glottal stop and [h] (see (x)). T o prov ide a more comple te hypothes is , I hypothes ize that g lot ta l ized stops are spec i f i ed for the feature [stop] and the laryngeal feature [ C G ] ; the laryngeal consonant glotta l stop is spec i f i ed fo r the feature [ C G ] ; and the consonant [h] is spec i f i ed for the feature [ SG ] . T h i s is exemp l i f i ed in (4). (4) L a r yngea l M e c h a n i s m Hypothes i s ( rev ised f r o m R i c e 1994) voiced voiceless voiceless glottalized voiceless voiced glottal h stop aspirated stops unaspirated stops stops fricatives fricatives stop stop X X X X voice X X SG X X CG X X In the remainder o f this thesis, the c l a ims o f the La r yngea l M e c h a n i s m Hypo thes i s are tested against observed patterns o f behav ior gove rn ing sy l l ab le structure and m o r p h o p h o n e m i c ev idence. The morphophonemic ev idence cons idered inc ludes the morphophonemic process o f cont inuant v o i c i n g and word-f ina l d e v o i c i n g , and the m o r p h o p h o n e m i c effects o f the c lass i f i e r pref ixes . It w i l l be shown that observed patterns o f behav ior gove rn ing sy l lab le structure support the laryngeal spec i f i cat ions o f the stop consonants in the La ryngea l M e c h a n i s m Hypothes i s . In add i t ion , it w i l l be shown that morphophonemi c processes o f cont inuant v o i c i n g and word-f ina l d e v o i c i n g and morphophonemic effects o f the c lass i f i e r pre f ixes support the laryngeal spec i f i cat ions o f the f r icat ive consonants, g lotta l stop and [h] in the La ryngea l M e c h a n i s m Hypothes i s . 1.5 Ou t l i ne T h i s thesis examines the phonet ic and phono log i ca l propert ies o f laryngeal d is t inct ions in the consonant inventory o f Tah l tan . Reca l l that there is uncerta inty sur round ing the phonet ic laryngeal spec i f i ca t ion o f p l a in stop consonants in Tah l tan . T o determine the phonet ic laryngeal spec i f i ca t ion o f Tah l tan p la in stop consonants, a phonet ic study, w h i c h examined the acoust ic propert ies o f p la in stops, was conducted. In chapter two , this phonet ic study is presented and 4 discussed. In this chapter, I w i l l argue fo r rev i s ions to the D u a l M e c h a n i s m Hypothes i s ( R i c e 1994) based on the results f r o m the phonet ic study o f p l a i n stop consonants. Spec i f i c a l l y , I argue that the category vo i c ed stop be added to the D u a l M e c h a n i s m Hypothes i s . T h e n in chapters three through five, the c l a ims o f the rev ised D u a l M e c h a n i s m Hypo thes i s , i.e. La r yngea l M e c h a n i s m Hypothes i s , are tested. In chapter three, the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s are tested against observed patterns o f behav io r gove rn ing sy l lab le structure. In this chapter it w i l l be shown that observed patterns o f behav ior gove rn ing sy l l ab le structure support the la ryngea l spec i f icat ions o f the stop consonants in the La ryngea l M e c h a n i s m Hypothes i s . In chapters four and five, the c l a ims o f the D u a l M e c h a n i s m Hypo thes i s are tested against morphophonemic ev idence. M o r e spec i f i ca l l y , chapter four , examines the morphophonemi c processes o f cont inuant v o i c i n g , w h i c h targets stem-init ia l f r icat ives , and word-f ina l devo i c i ng , w h i c h targets word-f ina l f r icat ives, to test the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s . In this chapter it w i l l be shown that morphophonemi c processes o f cont inuant v o i c i n g and word-final d e v o i c i n g support the laryngeal spec i f i ca t ion o f [voice] for the f r icat ive consonants. In chapter five, the morphophonemic effects o f the c lass i f i e r pref ixes (a p re f ix , w h i c h d i rec t l y precedes the verb stem) are examined in order to test the c l a ims o f the L a r yngea l M e c h a n i s m Hypothes i s . In this chapter it w i l l be shown that morphophonemi c effects o f the c lass i f ie r pref ixes support the laryngeal spec i f i cat ions o f the f r icat ive consonants , glottal stop and [h] in the La r yngea l M e c h a n i s m Hypothes i s . In chapter s ix , a s suming the laryngeal spec i f i cat ions under the La ryngea l M e c h a n i s m Hypothes i s , the morphophonemi c process o f cont inuant v o i c i n g , word-final devo i c i ng and the morphophonemic effects o f the c lass i f ie r pref ixes are accounted for in Op t ima l i t y Theo ry ( M c C a r t h y and P r ince 1993). It w i l l be shown that assuming the laryngeal spec i f i cat ions o f the La r yngea l M e c h a n i s m Hypo thes i s , the morphophonemi c process o f cont inuant v o i c i n g , word-f ina l devo i c i ng and the morphophonemi c effects o f the c lass i f i e r pref ixes can accounted fo r in Op t ima l i t y Theo r y ( M c C a r t h y and Pr ince 1993). F i n a l l y , in chapter seven conc l ud ing remarks are made. 5 Chapter Two Phonetic Analysis of Plain Stops Stop consonants are present i n the consonant inventor ies o f a l l k n o w n languages ( M a d d i e s o n 1984). In A thapaskan languages it is c lear stop consonants exh ib i t a three w a y laryngeal contrast. W i t h i n the A thapaskan l iterature, the three series o f stop consonants are t yp i ca l l y descr ibed as vo ice less unaspirated (p la in stops), vo ice less aspirated and g lo t ta l i zed . In Tah l tan , the series o f p l a in stop consonants have been descr ibed incons is tent ly . A l t h o u g h never the focus o f an invest igat ion, Tah l tan p l a in stops have been descr ibed as vo ice less unaspirated ( H a r d w i c k 1984) and as v o i c e d ( Shaw 1981 ; Shaw 1982; 1983; Na te r 1989) . 1 T o determine the phonet ic laryngeal spec i f i ca t ion o f Tah l tan p l a in stop consonants, a phonet ic study, w h i c h examined the acoust ic propert ies o f p l a in stops, was conducted . In this chapter, th is phonet ic study is presented and d iscussed. Th i s chapter is d i v i d e d into s ix sect ions. In sect ion 2.1, the object ive o f the study is presented. In sect ion 2.2, background in fo rmat ion is p rov ided . Spec i f i c a l l y , the Tah l tan consonant inventory is presented, v o i c e d and vo ice less unaspirated stops are d iscussed and the acoust ic propert ies o f v o i c e d and vo ice less unaspirated stops are also d iscussed. In sect ion 2.3, the methods are presented. In sect ion 2.4, the results f r o m this study are presented. In sect ion 2.5, the results are d iscussed , and finally, in sect ion 2.6 c o n c l u d i n g remarks are made. 2.1 Objec t i ve There are spec i f i c acoust ic propert ies, w h i c h differentiate v o i c e d and vo ice less sounds. The goa l o f this study is to examine the acoust ic propert ies o f Tah l tan p l a in stop consonants to determine i f they are v o i c e d , i.e. have the acoust ic propert ies o f v o i c e d sounds, o r vo ice less , i.e. have the acoust ic propert ies o f vo ice less sounds. 2.2 B a c k g r o u n d 2.2.1 Tah l tan Stop Consonants In Tah l tan , stop consonants exh ib i t a three w a y laryngeal contrast. A s p rev ious l y ment ioned , there is uncertainty sur round ing the laryngeal status o f p l a in stop consonants. The p la in stop consonants are ident i f ied i n (1). T h e r ema in ing stop consonants , w h i c h are a lso present 1 In general, voiced and voiceless unaspirated stops are referred to as plain stops. At present, in Tahltan, it is uncertain if the series of plain stops are voiced or voiceless unaspirated. Until the laryngeal specification is determined this series of stops will be referred to simply as 'plain stops'. Once the laryngeal specification is determined this series of stops will be referred to as voiced or voiceless unaspirated. 6 in ( 1 ) , are described as voiceless aspirated and glottalized in Tahltan. This is consistent with the description in the Athapaskan literature. (1 ) Tahltan Stop Consonants2 labial coronal dorsal laryngeal labial alveolar inter-dental alveolar • lateral palatal velar labio-velar uvular glottal a. P t te ts t l ts k kw q ? b. t h teh ts h t i h ts h k h kwh q h c. t' t 9 ' ts ' t i ' t s ' k' q ' a. plain stop b. voiceless aspirated stop c. glottalized stop With the exception of the labial stop consonant [p] and glottal stop, all stop consonants (affricate and non-affricate) participate in the three-way laryngeal contrast discussed above. This is exemplified in (2). (2) Three-way Laryngeal Contrast labial coronal dorsal laryngeal a. P te t ts ti ts k kw q ? b. teh th ts h t i h t s h k h kwh q h c. te' t' ts ' t i ' t s ' k' kw, q ' a. plain stop b. voiceless aspirated stop c. glottalized stop In Tahltan there are two types of stop consonants: affricate and non-affricate. The affricate stops have either a fricated mid-sagittal release or a lateral release. In Tahltan, all affricates are coronal in place of articulation. This is exemplified in (3). (3) Affricate Stop Consonants labial coronal dorsal laryngeal a. P te t ts t i ts k kw q ? b. teh t h ts h t i h ts h k h kwh q h c. te' t' ts ' t i ' t s ' k' kw, q ' a. plain stop b. voiceless aspirated stop c. glottalized stop 2 Unti l the laryngeal specification is determined, the phonetic symbol for voiceless unaspirated stops w i l l be used to represent this series o f stop consonants. 7 Unlike affricate stop consonants, non-affricate stop consonants contrast in place of articulation. For example, in Tahltan, there are labial, coronal, dorsal and laryngeal non-affricate stop consonants. This is exemplified in (4). (4) Non-affricate Stop Consonants labial coronal dorsal laryngeal a. P te t ts rt ts k k w q ? b. t 9 h t h t s h rth tsh k h kwh q h c. t 9 ' t ' t s ' rt' t s ' k' k w ' q ' a. plain stop b. voiceless aspirated stop c. glottalized stop Not all stop consonants will be discussed in this chapter. This chapter will focus on plain non-affricate stop consonants. Specifically, this chapter will focus on the acoustic laryngeal properties of the labial stop consonant [p], the coronal stop consonant [t], and the dorsal stop consonant [k]. These consonants are identified in (5). (5) Consonants Focused on in this Study labial coronal dorsal laryngeal a. P te t ts rt ts k kw q ? b. teh th t s h rth ts h k h kwh q h c. te' f t s ' rt' t s ' k' k w ' q' a. plain stop b. voiceless aspirated stop c. glottalized stop 2.2.2 Voiced and Voiceless Unaspirated Stops As previously mentioned, Tahltan plain stop consonants have been described as voiced and as voiceless unaspirated. The difference between voiced and voiceless unaspirated stop consonants has to do with: (a) vocal fold vibration, and (b) the position of the vocal folds. Voiceless unaspirated stops are the most common type of stop consonant; they are present in 92% of the world's languages (Maddieson 1984). During the production of a voiceless unaspirated stop the vocal folds are apart and the glottis is open. Since the vocal folds are pulled apart there is no vocal fold vibration; instead, air passes directly through the glottis. In certain environments, voiceless unaspirated stops are partially voiced. For example, if a voiceless unaspirated stop is preceded by a voiced segment, i.e. a vowel, sonorant or voiced consonant, there may be a few periods of voicing immediately after the oral closure because the vocal folds are not separated enough by the time the oral closure is reached. The exact amount of voicing depends on the language and the speaker. 8 D u r i n g the product ion o f a vo i c ed stop, the voca l fo lds are brought together, but are not t ight l y c losed . A i r passes between the voca l fo lds , w h i c h causes them to v ibrate. There is var ia t ion in the exact amount o f voca l f o l d v ib ra t ion . T h i s var ia t ion is dependent on the language, the speaker and/or the env i ronment o f the stop consonant. F o r examp le , some languages, such as F rench , have vo i c ed stops, w h i c h require energetic efforts to sustain voca l f o l d v ib ra t ion throughout the stop c losure ( Lade foged and M a d d i e s o n 1996). E v e n in env i ronments w h i c h require s ign i f i cant efforts to sustain voca l f o l d v ibra t ion (i.e. word- in i t i a l l y and f o l l o w i n g vo ice less segments) voca l f o l d v ibra t ion is ma in ta ined . In other languages, vo ca l f o l d v ib ra t ion is dependent on the env i ronment . F o r examp le , f o l l o w i n g v o i c e d segments (i.e. v owe l s , sonorant and v o i c e d consonants) there is voca l f o l d v ib ra t ion throughout the stop c losure . In compa r i son , f o l l o w i n g vo ice less segments there is no voca l f o l d v ib ra t ion . In other words , the voca l fo lds are in a pos i t ion for v o i c i n g , but no voca l f o l d v ibra t ion occurs . In some languages, such as E n g l i s h , the voca l fo lds are in a pos i t ion fo r v o i c i n g , but do not v ibrate throughout the stop c losure ( Lade foged and M a d d i e s o n 1996). E v e n when preceded and f o l l o w e d by vo i c ed segments, v o i c i n g ends soon after the ora l c losure and begins short ly after the stop is released. E n g l i s h stops are referred to as vo i c ed by some researchers because the art iculators, i.e. the voca l fo lds , are in a pos i t ion fo r v o i c i n g . Other researchers refer to E n g l i s h stops as vo ice less unaspirated because acoust i ca l l y they are vo ice less , i.e. there is no voca l f o l d v ib ra t ion . In this thesis, the acoust ic , not the art iculatory propert ies, o f p la in stops are e x a m i n e d ; thus stop consonants w h i c h are l i ke E n g l i s h stops w i l l be referred to as vo ice less unaspirated, not vo i ced . In this phonet ic study, the acoust ic propert ies o f p l a i n stops are examined to determine i f p l a in stops are vo ice less unaspirated or vo i ced . M o r e spec i f i ca l l y , the acoust ic propert ies o f p la in stops w i l l be examined to determine i f p l a in stops are: vo ice less unaspirated in a l l env i ronments ; vo ice less unaspirated, but par t ia l l y vo i c ed f o l l o w i n g v o i c e d sounds (i.e. vowe l s and sonorants) ; v o i c ed in a l l env i ronments ; or vo i ced f o l l o w i n g vo i c ed sounds, but vo ice less f o l l o w i n g vo ice less sounds. 2.2.3 A c o u s t i c Propert ies o f V o i c e d and Vo i c e l e s s Unasp i ra ted Stops A w a v e f o r m is a graphic representation o f a v ibratory event s h o w i n g ampl i tude versus t ime. There are per iod ic and aper iod ic wave fo rms . A per iod ic w a v e f o r m consists o f consists o f a regular pattern o f v ibra t ion that repeats itself. T h e regular pattern o f v ib ra t ion corresponds to the pulses produced by the v ib ra t ing voca l fo lds . V o i c e d sounds, i.e. v owe l s , sonorants and v o i c e d consonants ( i n c lud ing stops), produce per iod ic wave fo rms . Reca l l that in some languages, 9 vo ice less unaspirated stops are part ia l ly vo i c ed i n certa in env i ronments . The part ia l ly vo i c ed por t ion o f the vo ice less unaspirated stop also produces a per iod i c w a v e f o r m . The vo ice less por t ion o f the vo ice less unaspirated stop produces an aper iod ic w a v e f o r m . A n aper iod ic wave fo rm consists o f an i rregular pattern o f v ibra t ion that does not repeat itself. In add i t ion to vo ice less unaspirated stops, other vo ice less stops and vo ice less f r icat ives produces aper iod ic wave fo rms . R e c a l l that in some languages, such as E n g l i s h , the voca l fo lds are in a pos i t ion fo r v o i c i n g , but there is l itt le or no voca l f o l d v ib ra t ion , even when f o l l o w i n g vo i c ed sounds. S ince there is no voca l f o l d v ibra t ion an aper iod ic wave fo rm is p roduced . Sound spectrograms are machine-made graph ic representation o f sounds in terms o f their component f requencies, in w h i c h t ime is shown on the hor izonta l ax i s , f requency on the ver t ica l ax i s , and the intensity o f each f requency at each moment in t ime is shown by the darkness o f each mark ( Lade foged 1993). In a spectrogram, the most obv ious ind ica t ion o f a stop consonant is the stop c losure. F o r example , the stop c losure o f vo ice less stop is absent o f formants and is t ru ly si lent. S im i l a r l y , the stop c losure o f a vo i c ed stop is absent o f formants , however , in contrast to the c losure o f a vo ice less stop, a low-frequency band o f energy referred to as a vo i c e bar character izes the stop c losure o f a vo i c ed stop. In this study o f Tah l tan p la in stops, the acoust ic propert ies d iscussed above (i.e. per iod i c i t y and a vo i ce bar versus aper iod ic i t y and no vo i ce bar) w i l l be used to d is t ingu ish a vo i ced stop f r om a vo ice less unaspirated stop. Other acoust ic propert ies, w h i c h d is t ingu ish vo i ced and vo ice less unaspirated stops are: v o w e l length, w i th longer vowe l s before v o i c e d stops; length o f the stop c losure , w i t h a longer c losure fo r vo ice less unaspirated stops; and strength o f the release burst, w i t h a stronger burst for vo ice less unaspirated stops. These acoust ic propert ies w i l l be cons idered in future phonet ic studies. 2.3 M e t h o d s In this phonet ic study, the data ana lyzed is f r o m two sets o f f ie ldnotes w i t h a c company ing tapes. One set o f f ie ldnotes was co l lec ted by P .A . Shaw in Te legraph Creek , B r i t i sh C o l u m b i a in 1983 . 3 The second set o f f ie ldnotes was co l lec ted by myse l f , in co l l abora t ion w i t h J . D . A ldere te , in Te legraph C reek and Iskut, B r i t i s h C o l u m b i a in June 1999. 4 In tota l , the speech o f f i ve speakers was ana lyzed , one speaker f r om P .A. Shaw ' s f ie ldnotes and four speakers 3 The data was analyzed with the permission of P.A. Shaw. This fieldwork was supported in part by the Linguistic Division of the British Columbia Provincial Museum. 4 This fieldwork was supported in part by a Phillips Fund Grant for Native American Research awarded to John D. Alderete. 10 f r o m the f ie ldnotes co l lec ted b y myse l f , in co l l abora t ion w i t h J . D . A ldere te . O f the f i ve speakers, three are male and t w o are female . T w o o f the speakers are f r o m the Tah l tan c o m m u n i t y o f Iskut and three speakers are f r o m the Tah l tan c o m m u n i t y o f Te legraph Creek . T h e demograph ics o f the Tah l tan speaker d i d not factor into the results. T h e data co l lec ted in Te legraph C reek i n 1983 was des igned to determine general propert ies o f the Tah l tan phono logy and m o r p h o l o g y . In compa r i son , the data co l l ec ted in June 1999 cons is ted o f a 50-word l ist , w h i c h was des igned to determine the laryngeal spec i f i ca t ion o f p la in stop consonants. The w o r d l ist was care fu l l y o rgan ized to ensure that each p la in stop consonant appeared in several near-min imal set env i ronments : post-sibi lant, post-nasal and in tervoca l i ca l l y . In a l l examples , the stop consonant in quest ion is the in i t i a l consonant o f a noun stem, w h i c h is preceded by a possessive p r e f i x . 5 The possess ive p re f i x is the f irst person s ingular , [es], the second person s ingular , [ in] , or the th i rd person s ingular , [me]. T h e data ana lyzed f r o m P .A . Shaw ' s f ie ldnotes were ident ica l i n that the stop consonants , [p], [t] and [k] were noun stem-ini t ia l , and were preceded by a possess ive p re f ix . E a ch speaker was asked to produce t w o tokens o f each w o r d . U s i n g the acoust ic software Signalyze, a w a v e f o r m and a spectrogram were p roduced fo r each token . U s i n g the wave fo rms and spectrograms, two acoust ic measurements were made. F i rs t , the durat ion o f a l l stop consonants was measured. Second, the v o i c e d por t ion , i.e. the per iod ic por t ion , o f each stop consonant was measured. The stop durat ion and v o i c i n g durat ion measurements were taken in the f o l l o w i n g word-internal env i ronments : post-sibi lant and preced ing a v o w e l , post-nasal and preced ing a v o w e l , and in tervoca l i ca l l y . F r o m the recorded measurements, the mean stop durat ion and mean v o i c i n g durat ion were ca lcu lated for each speaker. In add i t ion to the mean , the v o i c i n g percentage o f each stop consonant was ca lcu lated by d i v i d i n g the mean v o i c i n g durat ion by the mean stop durat ion. A l l stop durat ion measurements were taken f r o m the c losure onset to the onset o f v o i c i n g o f the f o l l o w i n g v o w e l . T h i s is exemp l i f i ed in (6). (6) Stop Du ra t i on [es], [m] o r [me] <- stop duration v o w e l c losure onset v o i c i n g onset 5 The prefixed nouns collected, from Tahltan speakers, are presented in Appendix A. 11 In examples where the voicing duration was equivalent to the stop duration, the voicing duration (similar to the stop duration) was measured from closure onset to the onset of the following vowel. This is exemplified in (7). In (7), voicing is indicated with a wavy line. (7) Voicing Duration stop duration ->• voicing duration -> [es], [in] or [me] closure onset vowel voicing onset In examples where the voicing duration was less than the stop duration, the voicing duration was measured from (a) the closure onset to the offset of voicing (at some point within the stop consonant) or (b) from the onset of voicing (at some point within the stop consonant) to the onset of the following vowel. This is exemplified in (8) and (9) respectively. In (8) and (9), voicing is indicated with a wavy line and lack of vocal fold vibration (voicelessness) is indicated with a dashed line. (8) Voicing Duration [es], [in] or [me] stop duration —» voicing duration closure onset vowel voicing onset (9) Voicing Duration [es], [in] or [me] closure onset stop duration voicing duration vowel voicing onset In examples without voicing, i.e. without a periodic waveform, the voicing duration was zero. This is exemplified in (10). Similar to above, a dashed line indicates lack of vocal fold vibration. 12 (10) N o V o i c i n g Dura t i on [es], [in] or [me] c losure onset stop durat ion no voicing duration v o w e l v o i c i n g onset In the f o l l o w i n g sect ion, the results are presented. In this sect ion, stop consonants w i l l be referred to as v o i c e d , par t ia l ly vo i ced or vo ice less . Stop consonants w i l l be referred to as vo i c ed i f the vo i ced durat ion is equiva lent to the stop dura t ion ; par t ia l l y vo i c ed i f the v o i c i n g durat ion is less than the stop durat ion, but greater than ze ro ; or vo ice less i f the v o i c i n g durat ion is ze ro . 6 2.4 Resu l ts In this sect ion the acoust ic measurements, i.e. the stop durat ion , v o i c i n g durat ion and v o i c i n g percentage, are tabulated. Th i s sect ion is d i v i d e d in four sub-sections. In sect ion 2.4.1 through 2.4.3 respect ive ly , the acoust ic measurements , for [p], [t] and [k], in post-sibi lant, post -nasal and in tervoca l i c pos i t ions are presented and d iscussed. In sect ion 2.4.4, a summary o f the results is p rov ided . 2.4.1 Post-Sib i lant Stop Consonants In (11) through (13) respect ive ly , the acoust ic measurements o f [p], [t] and [k] in post-s ib i lant pos i t ion are tabulated. In c o l u m n one o f each table, the speaker is ident i f i ed , in c o l u m n two , the mean stop durat ion is presented, in c o l u m n three, the mean v o i c i n g durat ion is presented, and in c o l u m n four , the v o i c i n g percentage in g i ven . (11) Resul ts o f [p] in Post-Sib i lant Pos i t ion speaker v o i c i n g durat ion stop durat ion v o i c i n g percentage ( V ) * ( V ) * 1 47.26 139.82 33 .80 % 2 48 .59 139.57 30.38 % 3 89.21 127.72 69.85 % 4 119.74 179.40 66.74 % 5 0 146.92 0 % * ( V ) indicates that voicing was measured from the onset o f voicing, within the stop closure, to the onset o f voicing o f the fol lowing vowel. 6 In the results section, stop consonants w i l l be referred to as voiced, voiceless or partially voiced. Based on these results, I w i l l hypothesize a laryngeal specification (i.e. voiced or voiceless unaspirated). This hypothesis w i l l be presented in the discussion section (section 2.5). 13 Fo r speakers one through four , the stop consonant [p] was par t ia l l y vo i c ed i n post-sibi lant pos i t ion . Re ca l l that par t ia l ly vo i ced stops have a v o i c i n g durat ion , w h i c h is less than the stop durat ion , but greater than zero. F o r speakers one and two , there is l itt le var ia t ion in the v o i c i n g percentage. F o r example , 3 1 % was the lowest , w h i l e 3 4 % was the h ighest . 7 S i m i l a r l y , fo r speakers three and four , there is l i t t le var ia t ion in the v o i c i n g percentage. F o r example , 6 7 % was the lowest, wh i l e 7 0 % was the highest. F o r speakers one through four , as ( V ) indicates, the v o i c i n g durat ion was measured f r o m the onset o f v o i c i n g (wh i ch is w i th i n the stop c losure) to the onset o f v o i c i n g o f the f o l l o w i n g v o w e l . In other words , the stop consonant , [p] was vo i ced p r io r to the onset o f v o i c i n g o f the f o l l o w i n g v o w e l . U n l i k e speaker one through four , for speaker f i v e , the stop consonant , [p] was vo ice less in post-sibi lant pos i t ion . In other words , 0 % o f the stop durat ion was vo i ced in post-sibi lant pos i t ion . F igure ( l ) 8 consists o f a wave fo rm and spectrogram o f the w o r d , [esbeze], ' m y k n i f e ' . In (1), the post-sibi lant stop, [p] is ident i f ied w i t h ver t i ca l l ines. In the w a v e f o r m and spectrogram there are ind icat ions , the post-sibi lant stop consonant , [p] is par t ia l ly vo i c ed . In the wave fo rm there is per iod ic i t y through part o f the stop c losure and in the spectrogram there is a vo i ce bar through part o f the stop c losure. The per iod ic i t y and vo i ce bar beg in w i th in the stop c losure and extend to the onset o f v o i c i n g o f the f o l l o w i n g v o w e l . In (12) and (13) respect ive ly , the acoust ic measurements o f [t] and [k] in post-sibi lant pos i t ion are tabulated. (12) Results o f [t] in Post-Sib i lant Pos i t i on speaker v o i c i n g durat ion stop durat ion v o i c i n g percentage 1 0 158.97 0 % 2 0 134.81 0 % 3 0 136.97 0 % 4 0 181.27 0 % 5 0 140.93 0 % 7 In the text, the percentages have been rounded off to the nearest whole number. 8 A l l figures are presented in Appendix B. 14 (13) Resu l ts o f [k] in Post-Sib i lant Pos i t ion speaker v o i c i n g durat ion stop durat ion v o i c i n g percentage 1 0 145.30 0 % 2 0 133.64 0 % 3 0 113.13 0 % 4 0 176.48 0 % 5 0 156.11 0 % Fo r a l l f i ve speakers, the stop consonants , [t] and [k] were vo ice less . In other words , 0 % o f the stop durat ion was vo i ced . F igure (2) consists o f a wave fo rm and spectrogram o f the w o r d [esdihe], ' m y g rouse ' . In (2), the post-sibi lant stop, [t] is ident i f ied w i th ver t ica l l ines. There are ind icat ions that the post-s ib i lant stop, [t] is vo ice less . In the wave fo rm there is no pe r iod i c i t y and in the spectrogram there is no vo i ce bar. F igure (3) consists o f a wave fo rm and spectrogram o f the w o r d [esgendam], ' m y horse ' . In (3), the post-sibi lant stop, [k] is ident i f ied w i t h ver t ica l l ines. There are ind icat ions that the post-sibi lant stop, [k] is vo ice less . In the wave fo rm there is no per iod ic i t y and in the spectrogram there is no vo i ce bar. 2.4.2 Post-Nasal Stop Consonants In (14) through (16), the acoust ic measurements o f [p], [t] and [k] in post-nasal pos i t ion are tabulated. (14) Resul ts o f [p] in Post-Nasal Pos i t ion speaker v o i c i n g durat ion stop durat ion v o i c i n g percentage 1 54.71 54.71 1 0 0 % 2 46 .70 46.70 1 0 0 % 3* 4 173.28 173.28 1 0 0 % 5 35.22 ( n _ J * 113.01 3 1 . 1 7 % ( n _ J * * (n ) indicates that voicing duration was measured form the offset of the preceding nasal, [n] to the offset of voicing within the stop closure. F o r speakers one through four , the stop consonant , [p] was vo i c ed in post-nasal pos i t i on . In other words , 1 0 0 % o f the stop durat ion was vo i c ed . U n l i k e speakers one though four , fo r speaker f i v e , the stop consonant, [p] was part ia l ly vo i c ed . Spec i f i ca l l y , 31 % o f the stop durat ion 9 Unfortunately, for speaker three, there is no data in which the stops, [b], [t] and [k] follow the possessive morpheme [in]; thus for speaker three, the stops were not analyzed in post-nasal position. 15 was vo i c ed . A s (n ) indicates, v o i c i n g began at the stop c losure and ended w i th i n the stop c losure . F igure (4) consists o f a wave fo rm and spectrogram o f the w o r d [ inbede], ' y ou r f o o d ' . In (4), the stop consonant, [p] is ident i f ied w i t h ver t i ca l l ines. In the w a v e f o r m and spectrogram there are ind icat ions that the post-nasal stop consonant, [p] is v o i c e d . F o r the durat ion o f the stop consonant , the w a v e f o r m is per iod ic and there is a vo i ce bar in the spectrogram. In (15) , the acoust ic measurements o f [t] in post-nasal pos i t i on are tabulated. (15) Resu l ts o f [t] i n Post-Nasal Pos i t i on speaker v o i c i n g durat ion stop durat ion v o i c i n g percentage (n )* (n )* 1 34.24 48.73 70.26 % 2 51.16 81.56 62.73 % 3 4 103.15 151.67 68.01 % 5 40 .32 123.68 3 2 . 6 0 % * (n ) indicates that voicing duration was measured form the offset of the preceding nasal, [n] to the offset of voicing within the stop closure. F o r a l l speakers, the stop consonant , [t] was par t ia l l y vo i c ed . F o r speakers one, t w o and four , there was l itt le var ia t ion in the v o i c i n g percentage. F o r example , 6 3 % was the lowest v o i c i n g percentage, w h i l e 7 0 % was the highest. F o r speaker f i ve , 3 2 . 6 0 % o f the stop c losure was vo i c ed . F o r a l l speakers, as (n ) indicates, the v o i c i n g began at the c losure onset and ended du r ing the stop c losure . In (16), the acoust ic measurements o f [k] i n post-nasal pos i t i on are tabulated. (16) Resul ts o f [k] in post-nasal pos i t ion speaker v o i c i n g durat ion stop durat ion v o i c i n g percentage (n )* (n )* 1 42.58 82.26 5 1 . 7 6 % 2 37.74 61.18 6 1 . 6 9 % 3 4 144.50 162.65 88.84 % 5 52.38 136.30 38.43 % * (n ) indicates that voicing duration was measured form the offset of the preceding nasal, [n] to the offset of voicing within the stop closure. S im i l a r to [t], for a l l speakers, the stop consonant [k] was part ia l ly v o i c e d . U n l i k e [t], across speakers there is large var ia t ion in the v o i c i n g percentage. F o r examp le , 3 9 % was the lowest v o i c i n g percentage, w h i l e 8 5 % was the highest v o i c i n g percentage. A l t h o u g h the v o i c i n g percentage var ies greatly, fo r a l l speakers, as (n ) indicates, the v o i c i n g began at the c losure onset and ended w i t h i n the stop c losure. 16 F igure (5) consists o f a w a v e f o r m and a spectrogram o f the w o r d [ ingawe], ' y ou r d r u m ' . The post-nasal consonant, [k] is par t ia l ly vo i c ed . In the w a v e f o r m , there is pe r iod i c i t y through part o f the c losure and in the spectrogram, there is a vo i c e bar through part o f the c losure . N o t i c e that the v o i c i n g begins at the offset o f the preced ing nasal , [n] and ends w i t h i n the stop c losure . 2.4.3 Intervocal ic Stop Consonants In (17) through (19), the acoust ic measurements o f [p], [t] and [k] in an in tervoca l i c env i ronment are tabulated. (17) Resul ts o f [p] In tervoca l i ca l l y speaker v o i c i n g durat ion stop durat ion v o i c i n g percentage ( e _ J * ( e _ J * 1 120.97 137.43 88.02 % 2 135.84 149.27 9 1 . 0 0 % 4 144.85 166.53 86.98 % 5 42.30 164.78 25.67 % * (e ) indicates that voicing duration was measured form the offset of the preceding vowel, [e] to the offset of voicing within the stop closure. Fo r a l l speakers, the stop consonant, [p] was part ia l ly vo i c ed in te rvoca l i ca l l y . F o r speakers, one, two and four , there is l itt le var ia t ion in the v o i c i n g percentage. F o r example , 8 7 % was the lowest v o i c i n g , wh i l e 9 1 % was the highest v o i c i n g percentage. In contrast to speakers one, two and four , the v o i c i n g percentage for speaker f i ve is very l ow , o n l y 2 5 % o f the stop durat ion is vo i ced . F igure (6) consists o f a wave fo rm and a spectrogram o f the w o r d [mebede], ' h i s f o o d ' . The intervoca l ic consonant, [p] is par t ia l ly v o i c e d . In the w a v e f o r m , there is pe r iod i c i t y through part o f the c losure and in the spectrogram, there is a vo i c e bar through part o f the c losure . N o t i c e that the v o i c i n g begins at the offset o f the preced ing v o w e l , [e] and ends w i th i n the stop c losure . In (18) , the acoust ic measurements o f [t] in an in tervoca l i c env i ronment are tabulated. 1 0 Unfortunately, for speaker three, there is no data in which the stops, [b], [t] and [k] follow the possessive morpheme [me]; thus for speaker three, the stops were not analyzed in an intervoclaic environment. 17 (18) Results of [t] Intervocalically speaker voicing duration stop duration voicing percentage ( O * (e_J* 1 96.32 147.79 65.17% 2 88.91 145.05 61.30% 3 4 167.51 268.25 62.45 % 5 77.01 168.39 45.73 % * (e ) indicates that voicing duration was measured form the offset of the preceding vowel, [e] to the offset of voicing within the stop closure. For all speakers, the stop consonant, [t] was partially voiced intervocalically. For speakers, one, two and four, there is little variation in the voicing percentage. For example, 61% was the lowest voicing, while 65% was the highest voicing percentage. In contrast to speakers one, two and four, the voicing percentage for speaker five is low, only 46% of the stop duration is voiced. Figure (7) consists of a waveform and a spectrogram of the word [medegane], 'his sockeye'. The intervocalic consonant, [t] is partially voiced. In the waveform, there is periodicity through part of the closure and in the spectrogram, there is a voice bar through part of the closure. Notice that the voicing begins at the offset of the preceding vowel, [e] and ends within the stop closure. In (19), the acoustic measurements of [k] in an intervocalic environment are tabulated. (19) Results of [k] Intervocalically speaker voicing duration stop duration voicing percentage (e_J* (e_)* 1 38.12 156.24 24.4 % 2 39.51 166.01 23.80 % 3 4 125.47 192.61 65.14% 5 44.40 182.40 24.34 % * ( E ) indicates that voicing duration was measured form the offset of the preceding vowel, [e] to the offset of voicing within the stop closure. For all speakers, the stop consonant, [k] was partially voiced intervocalically. For speakers, one, two and five, there is little variation in the voicing percentage. For example, 23.80% was the lowest voicing, while 24.40% was the highest voicing percentage. In contrast to speakers one, two and four, the voicing percentage for speaker four is very high, 65% of the stop duration is voiced. 18 F igure (8) consists o f a w a v e f o r m and a spectrogram o f the w o r d [megahe], ' h i s rabb i t ' . T h e in tervoca l i c consonant, [k] is par t ia l l y v o i c e d . In the w a v e f o r m , there is pe r iod i c i t y through part o f the c losure and in the spectrogram, there is a vo i c e bar through part o f the c losure . N o t i c e that the v o i c i n g begins at the offset o f the preced ing v o w e l , [e] and ends w i t h i n the stop c losure . 2.4.4 Summary T h e results are summar i zed in (20). In (20) , the v o i c i n g percentage o f the p l a in stops, [p], [t] and [k], in each env i ronment is presented. (20) Summary o f Resu l ts speaker post-sibi lant post-nasal in tervoca ic b d g b d g b d g 1 3 4 % 0 % 0 % 100 % 7 0 % 5 2 % 8 8 % 6 5 % 2 4 % 2 31 % 0 % 0 % 100 % 63 % 6 2 % 91 % 61 % 2 4 % 3 7 0 % 0 % 0 % 4 6 5 % 0 % 0 % 100 % 6 8 % 8 9 % 8 7 % 6 2 % 6 5 % 5 0 % 0 % 0 % 31 % 3 2 % 3 9 % 2 6 % 4 6 % 2 4 % T h e results show that there are several d is t inct d i f ferences between [p], [t] and [k] in post-sibi lant, post-nasal and in tervoca l i c pos i t ions . F o r example , in post-sibi lant pos i t ion , the stop consonant, [p] was the on l y consonant to be par t ia l l y vo i c ed . T h e p l a in stops, [t] and [k] were consistent ly vo ice less in this pos i t i on . In post-nasal pos i t i on , the stop consonant , [p] was the on l y consonant to be f u l l y vo i ced . The p l a in stops, [t] and [k] were cons is tent ly par t ia l l y v o i c e d in this pos i t ion . In tervoca l i ca l l y , the stops, [p], [t] and [k] were a l l par t ia l ly v o i c e d . Howeve r , fo r speakers one, t w o and four , the stop, [p] has the highest v o i c i n g percentage. In fact, this is the case in a l l env i ronments , [p] a lways has the highest v o i c i n g percentage. 2.5 D i s cuss i on Reca l l that in Tah l tan , p l a in stop consonants have been descr ibed as vo ice less unaspirated ( H a r d w i c k 1984) and as v o i c e d ( Shaw 1981 ; Shaw 1982; Shaw 1983; Na te r 1989). T o determine a laryngeal spec i f i ca t ion the acoust ic propert ies o f Tah l tan p la in stop consonants were examined . B a s e d on the phonet ic results, presented in the prev ious sect ion, I hypothes ize that the lab ia l stop consonant is vo i c ed , i.e. [b], w h i l e the r ema in ing p la in stop consonants are vo ice less 19 unasp i ra ted . " T h i s hypothes is , in con junc t ion w i t h the fact there are aspirated and g lo t ta l ized stops in Tah l tan , imp l ies that Tah l tan stops exh ib i t four dist inct laryngeal ar t icu lat ions: v o i c e d , vo ice less unaspirated, vo ice less aspirated and g lo t ta l i zed . The Tah l tan stop consonants are presented in (21). (21) Tahl tan Stop Consonants labial coronal dorsal laryngeal labial dental inter-dental alveolar lateral palatal velar labio-velar uvular glottal a. b b. t to ts rt ts k k w q ? c. th teh tsh rth ts h k h kwh q h d. t' t9' ts' rt' ts' k' q' a. vo i c ed stop b. vo ice less unaspirated stop c. vo ice less aspirated stop d. g lo t ta l ized stop In the remainder o f this sect ion, ev idence , w h i c h supports this hypothes is , w i l l be presented and d iscussed. Spec i f i ca l l y , phonet ic , cross-l inguist ic and aerodynamic ev idence w i l l be presented and d iscussed. A l s o in this sect ion, I w i l l argue that the current hypothes is prov ides ev idence for rev is ions to the D u a l M e c h a n i s m Hypo thes i s (R i ce 1994) (chapter one-section 1.3). Reca l l that the D u a l M e c h a n i s m Hypothes i s is a hypothes is , w h i c h concerns the laryngeal spec i f i ca t ion o f stop and fr icat ive consonants in A thapaskan languages. The phonet ic results, presented in the prev ious sect ion, support the hypothes is that the p la in stop consonant, [b] is v o i c e d , wh i l e the r ema in ing p la in stop consonants are vo ice less unaspirated. In post-sibi lant, post-nasal and in tervoca l i c pos i t ions , the stop consonant , [b] was vo i c ed or par t ia l ly v o i c e d . 1 2 The fact that [b] was par t ia l l y vo i ced in post-sibi lant pos i t ion was " In this phonetic study, the acoustic properties of the affricate, labio-velar and uvular stops were not examined. Based on the phonetic properties of the stops, [d] and [k], I hypothesize that all stops, with the exception of [b], are voiceless unaspirated. To verify this hypothesis future phonetic studies will have to be conducted. 1 2 The results for speaker five were not consistent with the results of other speakers. Future phonetic studies will have to conducted to determine if the results of speaker five are specific to this speaker or specific to a group of speakers. Since there is no evidence supporting one hypothesis over the other, the results for speaker five were not taken into consideration when constructing the present hypothesis. 20 c ruc ia l in hypothes iz ing that the stop, [b] is vo i c ed . In post-nasal and in tervoca l i c pos i t ions , the stop, [b] is preceded by vo i c ed sounds, but in post-sibi lant pos i t ion , [b] is preceded by a vo ice less sound . The fact that [b] is vo iced/part ia l ly vo i c ed in post-nasal and in tervoca l i c pos i t ions has a phonet ic exp lanat ion , i.e. [b] is vo i c ed f o l l o w i n g vo i c ed sounds. H o w e v e r , the fact that [b] is par t ia l ly vo i c ed in post-sibi lant pos i t ion does not have a phonet ic exp lana t ion ; thus I hypothes ize that the stop consonant [b] is vo i c ed . In compa r i son , [d] and [k] were vo ice less in post-sibi lant pos i t ion and were part ia l ly vo i c ed in post-nasal and intervoca l ic pos i t ions . In other words , [t] and [k] were part ia l ly vo i ced when preceded by vo i c ed sounds, and were vo ice less when preceded by the vo ice less sounds. Therefore , I hypothes ize that the stops [t] and [k] are vo ice less unaspirated. T o account for the fact that [t] and [k] are par t ia l ly vo i c ed in post-nasal and in tervoca l i c pos i t ions , I hypothes ize that [t] and [k] are phonet i ca l l y vo i c ed when preceded by v o i c e d sounds. M a d d i e s o n (1984) shows that c ross- l ingu is t i ca l l y there is a re la t ionship between v o i c i n g and place o f ar t icu lat ion. Th i s re lat ionship is exemp l i f i ed in (22). In (22) , the number o f languages w i th vo i c ed and vo ice less stops produced at b i l ab i a l , a lveo lar and ve lar places o f ar t icu lat ion are presented. (22) F requency o f P l a i n Stops by P lace o f A r t i cu l a t i on ( M a d d i e s o n 1984) b i l ab ia l a l veo lar dorsa l p la in vo ice less 263 290 283 p la in vo i ced 199 195 175 T h e results show that across languages, the most c o m m o n vo ice less stop is the a lveo lar stop, [t], c l ose l y f o l l o w e d by dorsa l stop, [k] and f i na l l y , the labia l stop, [p]. In compar i son , the most c o m m o n vo i c ed stop is lab ia l stop [b], c lose l y f o l l o w e d by a lveo lar stop, [t] and f i na l l y , the dorsa l stop, [k]. T h e fact that the lab ia l stop [b] is the most c o m m o n vo i c ed stop is consistent w i t h the hypothesis that the lab ia l stop consonant, [b] is v o i c e d in Tah l tan . M a d d i e s o n (1984) goes on to show that the re lat ionship exemp l i f i ed in (22) is a lso ref lected in the consonant inventor ies o f languages. In 24 languages, the a lveo lar stop [t] and the dorsal stop [k] are present, but the lab ia l stop [p] is not. In 18 o f the 24 languages, there is a series o f vo i c ed stops, w h i c h inc ludes the lab ia l stop consonant, [b]. W i t h regards to vo i c ed stops, in 21 languages, the lab ia l stop consonant [b] is present, but the dorsal stop [k] is not. In 15 o f the 21 languages, the lab ia l stop [b] and a lveo lar stop [t] are present and in 6 o f these languages, the lab ia l stop [b] is the on l y vo i c ed consonant. Severa l o f these facts are consistent w i th the hypothes is fo r Tah l tan p la in 1 3 Across speakers there was variation in voicing percentage (see (11) in section 2.4.1); future phonetic studies will have to be conducted to determine if this variation is speaker-related or dialect-related. 21 stops, w h i c h states that the labia l stop consonant [b] is v o i c e d , w h i l e the r ema in ing p la in stop consonants are vo ice less unaspirated. F i rs t , under this hypothes is , and s im i l a r to the facts above , the vo ice less a lveolar stop consonant [t] and vo ice less dorsal stop [k] are present, but the vo ice less lab ia l stop [p] is not. Second , under this hypothes is , and s im i l a r to the facts above, the vo i c ed labia l stop consonant [b] is present, but the vo i c ed a lveo lar and dorsa l stops, [t] and [k], are not. T h e hypothes is that the stop consonant [b] is v o i c e d , w h i l e the r ema in ing stop consonants are vo ice less unaspirated is a lso supported by ae rodynamic ev idence ( K i n g s t o n 1996). D u r i n g the product ion o f vo i c ed stops, the voca l fo lds are v ib ra t ing . In order to ma in ta in voca l f o l d v ib ra t ion , air must f l o w between the voca l fo lds f r o m the sub-glottal cav i t y to the supra-glottal cav i ty . In order for air to f l o w between these cav i t ies , the air pressure in the supra-glottal cav i ty must be greater than the air pressure in the sub-glottal cav i ty . O n e w a y to s l o w d o w n the increase in air pressure in the supra-glottal cav i t y is to have greater vo l ume in the ora l cav i ty . Thus , a stop consonant, w h i c h is produced at the front o f the mouth is more l i ke l y to rema in vo i c ed throughout the stop c losure because the vo lume o f the ora l cav i t y is greater. In compar i son , a stop consonant produced at the back o f the mouth is less l i k e l y to remain vo i c ed throughout the stop c losure , because the v o l u m e o f the oral cav i ty is smal ler . In other words , a stop consonant w i t h a lab ia l p lace o f ar t icu la t ion, i.e. [b], is most l i k e l y to be v o i c e d , w h i l e a stop consonant w i t h a dorsal p lace o f ar t icu la t ion, i.e. [k], is least l i k e l y to be vo i c ed . T h i s art iculatory ev idence supports the hypothes is that [b] is v o i c e d , wh i l e the r ema in ing stops are vo ice less unaspirated (i.e. [b], w h i c h is p roduced at the front o f the mouth , is vo i c ed , wh i l e [t] and [k], w h i c h are produced further back in the mouth , are vo ice less unaspirated). The hypothes is , present in this sect ion, p rov ides ev idence for rev is ions to the D u a l M e c h a n i s m Hypothes i s (R i ce 1994). Re ca l l that the D u a l M e c h a n i s m Hypothes i s is a hypothes is , w h i c h concerns the laryngeal spec i f i ca t ion o f stop and f r icat ive consonants in A thapaskan languages. T h e Dua l M e c h a n i s m Hypothes i s proposed by R i c e (1994) is presented in (23). T h e x is used to indicate w h i c h feature(s) each series o f consonants is spec i f i ed for . (23) D u a l M e c h a n i s m Hypothes i s (R i ce 1994) voiceless voiceless voiceless voiced unaspirated aspirated fricative fricative Spread glottis X Voice X Stop X X 22 R e c a l l that i n the A thapaskan l iterature, stop consonants are descr ibed as vo ice less unaspirated, vo ice less aspirated and g lo t ta l i zed , w h i l e f r icat ives are descr ibed as vo i c ed and vo ice less . The D u a l M e c h a n i s m Hypothes i s proposed by R i c e (1994) is consistent w i t h these descr ipt ions . In this chapter, I hypothes ize that the lab ia l stop consonant, [b] is v o i c e d , wh i l e the r ema in ing p la in stop consonants are vo ice less unaspirated. U n d e r the D u a l M e c h a n i s m Hypothes i s proposed by R i c e (1994) there is no category v o i c e d stop. B a sed on the present hypothes is , I argue that the category vo i c ed stop be added to the D u a l M e c h a n i s m Hypothes i s to account fo r the fact [b] is vo i ced . The La r yngea l M e c h a n i s m Hypo thes i s is presented in (24). (24) L a r yngea l M e c h a n i s m Hypothes i s ( rev ised f r o m R i c e 1994) voiced voiceless voiceless glottalized voiceless voiced glottal h stop aspirated stops unaspirated stops stops fricatives fricatives stop stop X X X X voice X X SG X X CG X X In th is thesis I assume, f o l l o w i n g C l ements (1985) and Sagey (1986) that representations are h ie ra rch ica l l y structured. In (25), the h ierarch ica l representations under the La r yngea l M e c h a n i s m Hypothes i s are l a id out. (25) L a r yngea l M e c h a n i s m Hypothes i s Structural Representat ions [b] I Root / \ [stop] Laryngeal I [voice] Voiceless aspirated stop Root / \ [stop] Laryngeal I [SG] Voiceless fricatives Root Voiceless unaspirated stop Root / \ [stop] Laryngeal Glottalized stop Root / \ [stop] Laryngeal [CG] Voiced fricatives Root Laryngeal I [voice] 23 Glottal stop Root Root Laryngeal Laryngeal [CG] [SG] 2.6 C o n c l u s i o n The goal o f this chapter has been to examine the acoust ic propert ies o f Tah l tan p la in stops in order to determine a laryngeal spec i f i ca t ion . B a sed on phonet ic results, in con junc t ion w i th cross-l inguist ic , aerodynamic and h is tor ica l ev idence , I have hypothes ized that the stop consonant , [b] is vo i ced and that the r ema in ing p la in stop consonants are vo ice less unaspirated. 24 Chapter Three Syllable Structure Constraint In th is chapter, the c l a ims o f the La r yngea l M e c h a n i s m are tested against observed patterns o f behav ior gove rn ing sy l lab le structure. Spec i f i c a l l y , this chapter w i l l focus on the stop and f r icat ive consonants , w h i c h surface in sy l lab le- in i t ia l and sy l lab le-f ina l pos i t i on , i.e. in onset and coda pos i t ions , in order to test the c l a ims o f the L a r yngea l M e c h a n i s m Hypothes i s . In sect ion 3.1, the La r yngea l M e c h a n i s m is restated, l n sect ion 3.2, background in fo rmat ion is p rov ided . Spec i f i ca l l y , the Tah l tan consonant inventory is presented, the Tah l tan or thographic consonant inventory is presented, and a lso there is in fo rmat ion about Tah l tan sy l lab le structure. In sect ion 3.3, data, w h i c h exemp l i f i e s the d is t r ibut ion o f consonants in sy l lab le- in i t ia l and sy l lab le-f ina l pos i t ions , is presented. In sect ion 3.4, a sy l lab le structure constraint , w h i c h accounts for the d is t r ibut ion o f consonants , is hypothes ized . In this sect ion, it w i l l be shown that the patterns o f behav io r gove rn ing sy l l ab le structure, in con junc t ion w i t h the sy l l ab le structure constraint, support the c l a ims o f the L a r yngea l M e c h a n i s m Hypothes i s . F i n a l l y , in sect ion 3.5, c o n c l u d i n g remarks are made. 3.1 Theoret i ca l A s s u m p t i o n s Reca l l that the goa l o f this chapter is to test the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s against observed patterns o f behav ior gove rn ing sy l l ab le structure. T h e La ryngea l M e c h a n i s m Hypothes i s was introduced in chapter two . R e c a l l that based on the phonet ic study o f p la in stop consonants, the D u a l M e c h a n i s m Hypo thes i s (R i ce 1994) was rev ised. Spec i f i c a l l y , the category o f v o i c e d stop was added to the Hypo thes i s . The La r yngea l M e c h a n i s m Hypo thes i s is presented in (1). (1) La ryngea l M e c h a n i s m Hypothes i s ( rev ised f r o m R i c e 1994) b voiceless aspirated stops voiceless unaspirated stops glottalized stops voiceless fricatives voiced fricatives glottal stop h stop X X X X voice X X SG X X CG X X 25 3.2 B a c k g r o u n d 3.2.1 Tah l tan Consonant Inventory T h i s chapter w i l l focus on the consonants, w h i c h surface in sy l lab le- in i t ia l and sy l lable-f ina l pos i t ions , in order to test the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s . Re ca l l that the Tah l tan consonant inventory consists o f four laryngeal series o f stop consonants , t w o laryngeal series o f f r icat ive consonants and a s ingle series o f sonorant consonants. The Tah l tan consonant inventory is presented in (2). (2) Tah l tan Consonant Inventory labial coronal dorsal laryngeal labial alveolar inter-dental alveolar lateral palatal velar labio-velar uvular glottal a. b b. t te ts ti ts k k w q ? c. t h teh ts h t i h . ts h k h kwh q h d. f te' ts' ti' tS' k' k « . q' e. e s i s (~e) X x w X h f. 6 z 1 i (~ y) Y Y w g- m n w a. voiced stop e. voiceless fricative b. voiceless unaspirated stop f. voiced fricative c. voiceless aspirated stop g. sonorant d. glottalized stop 3.2.2 Or thograph ic Conven t i ons In the A thapaskan literature spec i f i c or thographic convent ions have been establ ished. F o r example , voiceless aspirated stops, i.e. [t1], are t yp i c a l l y represented w i t h the phonet ic symbo l s for voiceless unaspirated stops, i.e. [tj. Voiceless unaspirated stops, i.e. [tj, are represented w i th the phonet ic s ymbo l s for voiced stops, i.e. fdj. In sy l lab le-f ina l pos i t i on , voiceless unaspirated stops, i.e. ft], are represented w i t h the phonet ic s y m b o l fo r voiceless unaspirated stops, i.e. [tj, not the or thographic s y m b o l , i.e. fdj. In chapter two , I hypothes ized that the lab ia l stop consonant , i.e. [b], is vo i c ed . In the orthography, the phonet ic s ymbo l for voiced stops, is used to represent this stop consonant , i.e. [b]. These or thographic convent ions are summar ized i n (3). F o l l o w i n g (3), in (4), the Tah l tan or thographic consonant inventory is presented. These orthographic convent ions are f o l l o w e d in this thesis. 26 (3) Orthographic Conventions phonetic orthographic syllable-initial b b t d th t syllable-final t t (4) Tahltan Orthographic Consonant Inventory labial coronal dorsal laryngeal labial alveolar inter-dental alveolar lateral palatal velar labio-velar uvular glottal a. b b. d dS dz dl dz 9 9W G ? c. t te ts ti ts k k w q d. t' te' ts' t i ' ts' k' kw, q' e. e s i s (~ 9) X x w X h f. 6 z 1 z (~y) Y Y w K g- m n w a. voiced stop e. voiceless fricative b. voiceless unaspirated stop f. voiced fricative c. voiceless aspirated stop g. sonorant d. glottalized stop 3.2.3 Tahltan Syllable Structure To better understand Tahltan syllable structure, it is important to have a basic understanding of the Tahltan verb and noun; thus before discussing Tahltan syllable structure, the Tahltan verb and noun will be briefly introduced. The Tahltan verb consists of a verb stem, which is composed of a root plus aspectual suffix; and prefix morphemes.1 The verb stem and verbal prefixes all fall into identifiable classes, all of which are identified in (5). (5) Tahltan Verb (Hardwick 1984) object postposition adverb distributive incorporated direct object unspecified/ stem non-sing, subject 1 2 3 4 5 6 7 derivational conjugation mode subject classifier verb stem 8 9 10 11 12 The aspectual suffix varies in phonological shape. This variation depends upon mode and the phonological shape of the root, i.e. i f the root is open or closed (Hargus 1988). 27 A n example o f a Tah l tan verb is presented in (6). In (6), the numbers , under the morpheme b reakdown , cor respond to the number associated w i t h the verbal pref ixes in (5). A l s o , in (6) the verb stem is ident i f ied in square brackets. (6) Tah l tan V e r b morpheme b reakdown Tah l tan example E n g l i s h g loss 3 8 8 11 12 da de ne h [dos] da de ne h [dos] he's boiling O In Tah l tan , a noun m i n i m a l l y consists o f a noun stem, w h i c h is composed o f a root p lus a stem format i ve su f f ix . U n l i k e the Tah l tan verb , noun stems are not preceded by a series o f pref ixes . Instead, a noun may cons is t o f a stem a lone, may be a f f i xed or may be part o f a c o m p o u n d . The di f ferent types o f nouns are e x e m p l i f i e d in (7). The f irst example in (7) is an independent noun stem. In the second examp le , the noun stem [dih] is a f f i xed . M o r e spec i f i ca l l y , the noun stem is preceded by the f i rst person s ingular possess ive pre f ix , [es], and is f o l l o w e d by the possessive su f f i x , [e]. T h e th i rd example is a c o m p o u n d , w h i c h consists o f two noun stems. (7) N o u n s Independent N o u n Stem d ih grouse A f f i x e d N o u n Stem es d ih e my grouse C o m p o u n d tutsedle [tu] + [tsedle] creek [water] + [small] N o w that the Tah l tan verb and noun have been b r i e f l y in t roduced, Tah l tan sy l lab le structure w i l l be d iscussed. In Tah l tan , sy l lab les are p r ima r i l y open , i.e. C V or C V : . Open sy l lab les are present i n the examples in (8). In the f irst example , the verba l pref ixes a l l have the sy l lab le f o rm C V . In the second example , the verb stem, [0e] has an open sy l l ab le f o rm C V . In the th i rd and fourth examples respect ive ly , the noun stems have the open sy l lab les fo rms C V and C V : . In the f ina l example , the possessive p re f ix , [me] has the open sy l lab le f o rm C V . In (8), a l l open sy l lab les are presented in square brackets. 28 (8) O p e n Sy l l ab les morpheme b reakdown 3 8 8 11 12 da de ne h dos 7 12 he h 0e Tah l tan example E n g l i s h g loss [da] [de] ne h dos he's boiling O he h [Ge] [tu] [6e:] [me] [di] [he] they 're tanning O water belt his grouse C l o s e d sy l lab les are a lso present in Tah l tan , they have the f o l l o w i n g fo rms : C V C or V C . In Tah l tan , c losed sy l lab les are found in two env i ronments : word-f ina l l y and d i rec t l y p reced ing the stem. C l o s e d sy l lab les are present in the examples in (9) and (10). The examples in (9) have c losed sy l lab les in word-f ina l pos i t ion . In the f irst example , the verb stem has the c losed sy l lab le f o r m C V C , and in the second example , the noun stem has the c losed sy l lab le f o r m C V C . (9) Word-f ina l C l o s e d Sy l l ab les Tah l tan example E n g l i s h g loss morpheme b reakdown 3 8 8 11 12 da de ne h [dos] da de ne h [dos] he's boiling O [bes] knife T h e examples in (10) have c losed sy l lab les d i rec t l y p reced ing the stem. T h e f irst and second examples respect ive ly have the c losed sy l lab le fo rms C V C and V C di rect ly p reced ing the verb stem. The th i rd and fourth examples have the c losed sy l l ab le fo rms C V C and V C d i rec t l y p reced ing the noun stem. In (10), the c losed sy l lab les are ident i f ied in square brackets. (10) C l o s e d Sy l l ab les d i rec t l y preced ing the Stem morpheme b reakdown 3 8 8 11 12 da de ne h dos 11 12 h 0e Tah l tan example da de [neh] dos [eh] 0e [dah] bede [es] bede E n g l i s h g loss he's boiling O he's tanning O (i.e. hide) our food my food 29 3.3 D a t a Reca l l that this chapter w i l l focus on the stop and fr icat ive consonants , w h i c h surface in sy l lab le- in i t ia l and sy l lab le-f ina l pos i t ions , in order to test the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s . In this sect ion, data, w h i c h exemp l i f i e s the d is t r ibut ion o f consonants in sy l lable-in i t i a l and sy l la lbe-f ina l pos i t ions , are presented. 3.3.1 Consonants in Sy l lable-Ini t ia l Pos i t ion A l l stop consonants, i.e. vo i c ed , vo ice less unaspirated, vo ice less aspirated, g lo t ta l i zed and glot ta l stop, surface sy l lab le- in i t ia l pos i t ion . T h i s is exemp l i f i ed in (11) . (11) Stop Consonants in Sy l lable-In i t ia l Pos i t i on vo i c ed [b]e:s knife [b]ede food vo ice less unaspirated [d]ih grouse [dl]une mouse [d5] iya necklace [9]ah rabbit vo ice less aspirated [t]ene road [t6]e: rock [ts]ic nose [ ts]a:k 'odle rain bucket [k]e: foot glot ta l ized [ t ' l iyA hook [t ']o:g wart [t0' ]alu:d8e eyebrows [tO']ati moss/diapers [k ' ]ug paper glottal stop [?]ei beaver dam S i m i l a r to stop consonants, both series o f f r i cat ives , i.e. vo i c ed and vo ice less , surface in sy l lab le- in i t ia l pos i t ion . Th i s is exemp l i f i ed in (12). 30 (12) F r i ca t i ves in Sy l lable-In i t ia l Pos i t i on vo ice less [I]uwe fish [9]e: belt [s]e:g spit [9]ina slave [x je l trap [h ]odzih caribou vo i ced [6]a0 snow [y]Anje geese 3.3.2 Consonants in Sy l l ab le-F ina l Pos i t i on Vo i c e l e s s unaspirated stops and glotta l stop surface in sy l lab le-f ina l pos i t ion . Th i s is exemp l i f i ed in (13). (13) Stop Consonants in Sy l lab le-f ina l Pos i t i on vo i ce less unaspirated dustenre[t] ciYe:[t] t9'a[ti] kek'i[t9] de0dlin tse[ts] t'o:[k] te'i[k] bottle pillow moss/diaper box green wood wart tobacco glotta l stop ela[?] bark T h e vo i ced stop consonant, [b] does not surface in sy l lab le-f ina l pos i t ion , nor do the vo ice less aspirated and g lo t ta l ized stops. Vo i c e l e s s f r icat ives , but not vo i c ed f r icat ives surface in sy l lab le-f ina l pos i t ion . E x a m p l e s w i t h vo ice less f r icat ives i n sy l lab le-f ina l pos i t ion are presented in (14). (14) F r i ca t i ve Consonants i n Sy l lab le-f ina l Pos i t ion vo ice less unaspirated te[8] cane tA[8] arrow mi:[I] snare du[s] cat t i i [c ] grease ku[x] rice di[h] grouse 31 3.3.3 Summary In (15), there is a summary of the stop and fricative consonants, which surface in syllable-initial and syllable-final positions in Tahltan. (15) Syllable-initial and Syllable-final Consonants voiced stop voiceless unaspirated stop voiceless aspirated stop glottalized stops voiced fricative voiceless fricatives m [h] syllable-initial X X X X X X X X syllable-final X X X X 3.4 Syllable Structure Constraint As shown above, voiceless unaspirated stops, voiceless fricatives, glottal stop and [h] surface in syllable-final position. Voiced stops, voiceless aspirated stops, glottalized stops and voiced fricatives do not surface in syllable-final position. To account for these facts, I hypothesize that [-sonorant] consonants, i.e. stops and fricatives, which have a primary place specifications and a secondary laryngeal specification, i.e. [voice], [SG] or [CG], do not surface in syllable-final position. This hypothesis is expressed with the following syllable structure constraint. (16) Syllable Structure Constraint * [-sonorant] / \ [place] [laryngeal]J o Note that in the specification of glottal stop and [h] laryngeal is the primary place specification; therefore they will not be subject to this constraint. Recall that the goal of this chapter is to test the claims of the Laryngeal Mechanism Hypothesis against observed patterns of behavior governing syllable structure. Under the Laryngeal Mechanism Hypothesis, the stop consonant, [b] is specified for the laryngeal feature [voice]; thus the hypothesis above can account for the fact that [b] does not surface in syllable-final position. In other words, [b] does not surface in syllable-final position because it has the laryngeal specification [voice], as the hypothesis states, [- sonorant] consonants, which have a laryngeal specification do not surface in syllable-final position. If [b] were unspecified for the laryngeal specification [voice], i.e. were grouped with the voiceless unaspirated stops, there would be no explanation for why [b] does not surface in syllable-final position. Thus the fact that [b] does not surface in syllable-final position supports the Laryngeal Mechanism Hypothesis. 32 3.5 Conclusion The goal of this chapter is to test the claims of the Laryngeal against observed patterns of behavior governing syllable structure. Under the Laryngeal Mechanism Hypothesis, the stop consonant, [b] is specified for the laryngeal feature [voice]. The fact that [b] is specified for the feature [voice] provides an explanation for the asymmetry of what stops occur in syllable final position. 33 Chapter Four Classifiers The morphophonemi c effects o f the c lass i f ie r pref ixes te l l us a lot about the laryngeal spec i f i ca t ion o f f r icat ive and stop consonants. T h i s chapter w i l l examine the m o r p h o p h o n e m i c effects o f the c lass i f ie r pref ixes in order to test the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s . Re ca l l that the La ryngea l M e c h a n i s m Hypothes i s concerns the laryngeal spec i f i ca t ion o f stop and fr icat ive consonants. T h i s chapter is d i v i d e d into ten sect ions. In sect ion 4 .1 , the c lass i f ie r pref ixes are ident i f i ed . In this sect ion the importance o f the c lass i f i e r pref ixes in re lat ion to laryngeal spec i f i cat ions is a lso d iscussed. In sect ion 4.2, the funct ion o f the c lass i f i e r pref ixes is b r i e f l y d iscussed . T h e c lass i f i e r pref ixes interact w i t h the f o l l o w i n g verb stem-init ia l f r icat ives and the preced ing subject pref ixes . T o ident i fy the pos i t ion o f the c lass i f i e r morphemes in re lat ion to the verb stem and subject pre f ixes , the Tah l tan verb is presented in sect ion 4.3. In sect ion 4.4, the La ryngea l M e c h a n i s m is restated and br i e f l y d iscussed . In sect ions 4.6 through 4.9 the morphophonemics o f each c lass i f i e r p re f ix is exemp l i f i ed and an ana lys is for each c lass i f ie r p re f ix is presented and d iscussed. F i n a l l y , i n sect ion 4.10 c o n c l u d i n g remarks are made. 4.1 C l a s s i f i e r P re f ixes and La r yngea l Spec i f i ca t ion In A thapaskan languages four c lass i f iers have been ident i f i ed . In Tah l tan these c lass i f iers are d-, 1-, 0 - and h-. T h e laryngeal status o f verb stem-init ia l f r icat ives is d i rect ly dependent on the h-, 1- and 0 - c lass i f ie rs . W h e n the h- c lass i f i e r is present verb stem-init ia l f r icat ives are consistent ly vo ice less . W h e n the 1- c lass i f i e r is present verb stem-init ia l f r icat ives are consistent ly v o i c e d . W h e n the 0 - c lass i f ie r is present f r icat ives are vo i c ed f o l l o w i n g a vo i ced segment and vo ice less f o l l o w i n g a vo ice less segment, i.e. v o i c i n g ass imi l a t ion . The d-classif ier a lso interacts w i th stem-init ial f r icat ives . W h e n the d-classif ier is present the feature [stop] associates to the verb stem-init ia l f r icat ives . In add i t ion to f r icat ives , the d c lass i f i e r interacts w i t h glottal stop. W i t h the except ion o f the d- c lass i f i e r interact ing w i t h glotta l stop, c lass i f iers do not affect stem-init ia l stop and sonorant consonants . 4.2 Func t ion A t present, in Tahl tan the funct ion o f the c lass i f ie r pref ixes has not been studied in deta i l . The func t ion o f c lass i f iers has been studied in other A thapaskan languages; however , the funct ions are st i l l not comple te l y understood. W h a t is understood, w i t h regard to c lass i f ie r 34 funct ions , does appear to be fa i r l y consistent across languages. In th is sect ion the func t ion o f the c lass i f iers w i l l be summar i zed . W h e n poss ib le Tah l t an examples w i l l be used to e x e m p l i f y the c lass i f ie r funct ions . In A thapaskan languages, one o f the four c lass i f iers consistent ly co-occurs w i t h the verb . Such c lass i f iers have no dist inct func t ion , but are s i m p l y l ex i c a l l y spec i f i ed . In add i t ion to be ing l ex i c a l l y spec i f i ed , the d- and h- c lass i f ie rs have der ivat iona l funct ions . T h e h- c lass i f i e r der ives transi t ive or causat ive verbs f r om intransi t ive verbs. Const ruc t ions i n w h i c h the d- c lass i f i e r is present inc lude re f l ex ives , rec iproca ls and passives. The f o l l o w i n g Tah l tan data shows that in re f l ex ive construct ions the d- c lass i f i e r is cons is tent ly present ( H a r d w i c k 1984). (1) d- c lass i f i e r in re f l ex ive construct ions morpheme b reakdown Tah l tan example E n g l i s h g loss 9-10 11-12 y e n s h [x w ase ] y i h [xw9se] I tickled him 6 8 9-10 11 12-?ede de ye n s d [x w ase ] ?ede de s [g w ase] I tickled myself The d- c lass i f i e r a lso occurs in intransit ive construct ions , w h i c h have a p re f ix ind i ca t ing a customary ac t ion (i.e. ' She went aga in ' versus ' She went ' ) o r an ac t ion " b a c k " (i.e. ' She went back ' versus ' She went ' ) . Future research w i l l determine to what extent the func t ion o f the c lass i f i e r pref ixes in Tah l tan is the same as other A thapaskan languages. 4.3 Tah l tan V e r b The Tah l tan verb consists o f a verb stem, w h i c h is composed o f a root p lus aspectual su f f i x ; and p re f ix morphemes . 1 The verb stem and verba l pref ixes a l l f a l l into ident i f iab le classes, a l l o f w h i c h are ident i f ied in (2). (2) Tah l tan verb ( H a r d w i c k 1984) object postposition adverb distributive incorporated direct object unspecified/ stem non-sing subject 1 2 3 4 5 6 7 derivational conjugation mode subject classifier verb stem 8 9 10 11 12 The aspectual suffix varies in phonological shape. This variation depends upon mode and the phonological shape o f the root, i.e. i f the root is open or closed (Hargus 1988). 35 Th i s chapter is concerned w i t h the morphophonemic s o f the c lass i f ie r pref ixes , w h i c h precede the verb stem and f o l l o w the subject morphemes . The effects o f the c lass i f i e r pref ixes can be seen on the f o l l o w i n g verb stem-init ia l consonant or on a preced ing subject morpheme (pos i t ion 11). The subject morphemes are pos i ted to have the f o l l o w i n g fo rms in the unde r l y i ng representation. (3) Subject morphemes ( H a r d w i c k 1984) f irst person s ingular s-second person s ingular n-th i rd person s ingular unmarked f irst person p lura l 0 i d- 2 second person p lura l ah-thi rd person p lura l unmarked The th i rd person p lura l subject morpheme is unmarked in pos i t ion 11 preced ing the c lass i f i e r pref ixes . Instead, the th i rd person p lura l subject morpheme , [he] surfaces in pos i t ion 7. 4.4 Theoret ica l A s s u m p t i o n s R e c a l l that the goa l o f this chapter is to examine the morphophonemi c effects o f the c lass i f ie r pref ixes in order to test the c l a ims o f the L a r yngea l M e c h a n i s m Hypo thes i s . The Laryngea l M e c h a n i s m Hypothes i s is restated in (4). (4) La ryngea l M e c h a n i s m Hypothes i s ( rev ised f r o m R i c e 1994) b voiceless voiceless glottalized voiceless voiced glottal h aspirated unaspirated stops fricatives fricatives stop stops stops stop X X X X voice X X SG X X CG X X 4.5 Hypothes i s In the analys is o f the morphophonemi c effects c lass i f i e r pref ixes , I assume f o l l o w i n g R i c e (1989) , verb stem-init ia l f r icat ives are unmarked in l ex i ca l representation for laryngeal spec i f i ca t ion . Fr icat ives : (a) rema in unmarked for laryngeal spec i f i ca t ion or (b) surface w i t h the laryngeal spec i f i ca t ion [voice] . F r i ca t ives unspec i f i ed fo r laryngeal spec i f i ca t ion are vo ice less 2 The final consonant [d] of the first person singular never surfaces as an independent segment. The final consonant [d] is analyzed as a floating feature[stop] in this analysis (section 4.8). For convenience, the final consonant [d] of the first person plural will be written with a [d], not a floating [stop] feature. 36 under the La r yngea l M e c h a n i s m Hypothes i s . In compa r i son , f r icat ives marked for the laryngeal spec i f i ca t ion [voice] are vo i c ed under the L a r yngea l M e c h a n i s m Hypothes i s . In the analys is o f the morphophonemi c effects o f the c lass i f ie r pref ixes it w i l l become c lear that f r icat ives w i t h the spec i f i ca t ion [voice] acquire this laryngeal spec i f i ca t ion f r o m a preced ing c lass i f i e r pre f ix . 4.6 h- c lass i f i e r 4.6.1 M o r p h o p h o n e m i c s o f the h- c lass i f i e r In Tah l t an , the h- c lass i f i e r comes f r o m the Proto-Athapaskan c lass i f ie r *i. In some A thapaskan languages, i n c l ud ing Tah l tan , the h- c lass i f i e r is phonet i ca l l y rea l ized as [h]; thus it is referred to as the h- c lass i f ier . In Tah l tan , the h- c lass i f ie r is un ique fo r several reasons: a) it is the on l y c lass i f i e r rea l ized phonet i ca l l y as an independent segmental s lot ; b) it is the o n l y c lass i f i e r to interact w i th the subject morpheme ; and c) it is the c lass i f ie r cons is tent ly f o l l o w e d by verb stem-init ia l vo ice less f r icat ives. A s shown in (5), the h- c lass i f i e r is rea l i zed phonet i ca l l y as [h] in the th i rd person s ingular and th i rd person p lu ra l . In a l l the examples in table (5) the h- c lass i f i e r d i rec t l y precedes the verb stem, w h i c h is ident i f ied in square brackets. (5) T h i r d person s ingular and th i rd person p l u r a l 3 morpheme b r e a k d o w n 4 Tah l tan example E n g l i s h g loss 11 12 h [9e] e h [0e] he's tanning O (i.e. hide) 7 12 he h [0e] he h [0e] .they're tanning O (i.e. hide) 3 The data in this chapter is cited with permission from Patricia A . Shaw's fieldnotes collected in Telegraph Creek, British Columbia in 1981-82, and from a Telegraph Creek speaker resident o f Dawson Creek, British Columbia, in 1983. This fieldwork was supported in part by the Linguistics Div is ion o f British Columbia Provincial Museum. The generalizations and analysis in this chapter is based on this set o f fieldnotes. 4 The numbers correspond to the verbal prefixes identified in (2). Two prefix morphemes often surface as a single morpheme; this interaction is indicated with a hyphen (-). 5 The word-initial vowel, [e] is an epenthetic vowel. 37 3 11 12 de h [xui] 3 7 12 de h e h [xui] 11 12 h [X W 9S] 7 12 h e h [xwas] de h [xui] hi de h [xui] e h [xwase] he h [xw9se] he's throwing O they 're throwing O he's tickling O they 're tickling O The h- classifier interacts with the preceding second person singular subject morpheme, [n]. When the h- classifier is present, it coalesces with the second person singular morpheme, [n], surfacing as [n]. (6) Second person singular morpheme breakdown Tahltan example English gloss 11 12 n h [9e] 3 11 12 de n h [xui] 11 12 n h [xwas] in [9e] de n [xui] in [xwase] you're tanning O (i.e. hide) you 're throwing O you 're tickling O In the first person singular, the h- classifier shows no overt effect. This is exemplified in (7). (7) First person singular morpheme breakdown Tahltan example 11 12 s h [9e] 3 11 12 de s h [xui] 11 12 s h [xwas] e9 [9e] de s [xui] es [xwase] English gloss I'm tanning O (i.e. hide) I'm throwing O I'm tickling O 6 Before [n] the epenthetic vowel, [e] surfaces as [i]. 7 The first person singular, [s] surfaces phonetically as [s], [s] or [9]. Similarly, the initial consonant of the first person plural, [9] surfaces phonetically as [s], [s] or [9]. This is a result of coronal harmony, see Shaw 1991 for discussion and analysis of Tahltan consonant harmony. 38 The second person plural subject morpheme is [ah-]; thus i f is difficult to tell if: a) the h-classifier has no overt effect or b) if the h- classifier is realized phonetically and the subject morpheme [h] is deleted. (8) Second person plural morpheme breakdown Tahltan example English gloss 11 12 ah h [0e] 3 11 12 de n h [xui] 11 12 n h [xwas] ah [0e] de n [xui] in [xwase] you 're (pi) tanning O you 're throwing O you 're tickling O Particularly significant is the fact that in the first person plural, the d-effect, a process discussed in section (4.8), occurs and the h- classifier shows no overt effect. (9) First person plural morpheme breakdown Tahltan example English gloss 11 12 0id h [0e] 3 11 12 de 0id h [xui] 11 12 0id h [xwas] 0i [dSe] de 0i [gui] si [gwase] we 're tanning O we 're throwing O we 're tickling O 39 The character ist ics o f the h- c lass i f i e r (descr ibed above) are e x e m p l i f i e d in the comple te verba l paradigms in (10). (10) Phono log i c a l effects o f the h- c lass i f i e r morpheme b reakdown Tah l tan example 11 12 s h [Ge] 11 12 n h [Ge] 11 12 h [Ge] 11 12 Gid h [Ge] 11 12 ah h [Ge] 7 12 he h [Ge] 3 11 12 de s h [xui ] 3 11 12 de n h [xu i ] 3 11 12 de h [xui ] 3 11 12 de Gid h [xui ] 3 11 12 de ah h [xui ] 3 7 12 de he h [xui ] 11 12 s h [ x w as ] 11 12 n h [x w as ] 11 12 h [x w as ] 11 12 Gid h [ x w as ] 11 12 ah h [x w as ] 7 12 he h [x w as ] eG [Ge] i n [Ge] e h [Ge] Gi [d5e] ah [Ge] he h [Ge] de s [xu i ] de n [xui ] de h [xui ] de Gi [gui] d ah [xu i ] h i de h [xui ] es [xw9se] i n [x w ase ] e h [x w ase ] si [g w ase] ah [x w ase ] he h [x w ase ] E n g l i s h g loss I'm tanning O (i.e. hide) you 're tanning O he's tanning O we 're tanning O you 're (pi) tanning O they 're tanning O I'm throwing O you 're throwing O he's throwing O we 're throwing O you 're (pi) throwing O they 're throwing O I'm tickling O you 're tickling O he's tickling O we 're tickling O you 're (pi) tickling O they 're tickling O 40 The paradigms above g ive imper fec t i ve fo rms . In contrast, in the perfect ive the first person s ingular subject, [s] is deleted and the h c lass i f i e r surfaces phonet i ca l l y . T h i s is e x e m p l i f i e d in (11). (11) Phono log i c a l effects o f the h- c lass i f ie r 3 8 8 11 12 imper fec t ive da de ne s h [dos] 3 8 8 11 12 da de ne s h [dos] perfect ive imper fec t ive perfect ive 11 12 s h [gani] 9-10 11 12 0e n s h [gan] da de ne s [dos] da de ne h [dos] es [gani] Oi h [gan] I'm boiling O I boiled O I'm drying O I dried O imper fec t ive perfect ive 11 12 s h [x w ase ] 9-10 11 12 ye n s h [x w ase ] es [x w ase ] y i h [x w ase ] I'm tickling him I tickled him 4.6.2 A n a l y s i s o f the h- c lass i f ie r Re ca l l that the goa l o f this chapter is to examine the morphophonemi c effects o f the c lass i f ie r pref ixes to test the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s . In this sect ion it w i l l be shown that the La r yngea l M e c h a n i s m Hypo thes i s can account fo r the m o r p h o p h o n e m i c effects o f the h- c lass i f ier . M o r e spec i f i ca l l y it w i l l become c lear that the laryngeal feature [SG] is act ive w i t h i n the phono logy . F o r S lave, R i c e (1986, 1989b) proposes that a l l c lass i f iers are f loa t ing features. F o l l o w i n g R i c e , I assume the h- c lass i f ie r in Tah l tan is a f loa t ing [ SG ] feature, w h i c h precedes the verb stem and f o l l ows the subject morpheme. (12) h- c lass i f i e r s [xui ] [ SG ] 11 12 I w i l l beg in by s h o w i n g h o w the present analys is o f the h- c lass i f i e r accounts for the interact ion between the h- c lass i f ie r and the subject morphemes in Tah l tan . In the th i rd person s ingular and th i rd person p lura l the h- c lass i f ie r , i.e. the f loa t ing feature [ SG ] , surfaces as an independent segment [h]. The reason the h- c lass i f i e r surfaces as an independent segment has to do w i t h the Tahl tan sy l lab le structure. In Tah l tan sy l lab les are p r ima r i l y open , i.e. C V or C W . 41 There are two env i ronments where sy l lab les are c losed : word-f ina l l y , i.e. the ul t imate sy l l ab le , and in the pos i t ion d i rec t l y p reced ing the verb stem, i.e. the penult imate sy l l ab le . T h e latter pos i t i on , the coda pos i t ion o f the penult imate sy l l ab le , is where the h- c lass i f i e r is rea l ized phonet i ca l l y . In the th i rd person s ingular , w h i c h is m o r p h o l o g i c a l l y unmarked , and th i rd person p lu ra l , w h i c h surfaces in pos i t ion seven, there is no compet i t i on f r o m any other pre f ixes fo r segmental rea l izat ion in the coda pos i t i on , so the f loa t ing [SG] feature surfaces as an independent segment [h]. Th i s is exemp l i f i ed in (13) and (14). (13) T h i r d person s ingular de [xui ] de h [xui ] [SG] _> | [SG] 3 11 12 3 11 12 (14) T h i r d person p lura l de he [xui ] de he h [xui ] [SG] _ » | [SG] 3 7 11 12 3 7 11 12 In the case o f other p ronomina l subjects in the pa rad igm, there is compet i t i on for what gets rea l ized in the pre-stem coda pos i t i on . In the second person s ingular , the penul t imate c o d a pos i t ion is occup ied by the subject morpheme, [n]. In the second person s ingular , the h- c lass i f i e r associates to subject morpheme, [n]. The result is the vo ice less nasal [n]. T h i s is e x e m p l i f i e d in ( 1 5 ) . (15) Second person s ingular de n [xui ] de n [xui ] [ SG ] ° \ [ SG ] 3 11 12 3 11 12 In the f i rst person s ingular and second person p lura l the h- c lass i f i e r shows no overt effect. It is poss ib le that (a) the h- c lass i f ie r remains unparsed; (b) the h- c lass i f i e r deletes; or (c) the h- c lass i f ie r associates vacuous ly . Poss ib i l i t y (c) w o u l d f o l l o w f rom the hypothes is that f r icat ives are spec i f i ed for the feature [ SG ] . Poss ib i l i t y (c) is exemp l i f i ed in (16) and (17). 42 (16) F i rs t person s ingular de s [xui ] de s [xui ] [SG] _> \ [SG] 3 11 12 3 11 12 (17) Second person p lura l de ah [xu i ] de ah [xui ] [SG] _> \ [ SG ] 3 11 12 3 11 12 H i s to r i c a l ev idence supports the fact f r icat ives are spec i f i ed for the feature [ SG ] . Reca l l that the h- c lass i f i e r comes f r o m Proto-Athapaskan * i . T h i s h is tor ica l/d iachronic change can be ana l yzed as a process o f debucca l i za t ion , i.e. loss o f the ora l cav i t y ar t icu lat ion [lateral], l eav ing on l y the laryngeal spec i f i ca t ion [ SG ] . Th i s further supports the hypothes is that f r icat ives are spec i f i ed fo r the feature [ SG ] . (18) * i h A _» | [lar] [lat] [lar] I I [ SG ] [ SG ] Howeve r , the hypothes is that f r icat ives are spec i f i ed fo r the feature [ SG ] is not consistent w i t h the La r yngea l M e c h a n i s m Hypothes i s . U n d e r the La r yngea l M e c h a n i s m Hypo thes i s f r icat ives are not spec i f i ed for the feature [ SG ] . A l s o , the hypothes is that f r icat ives are spec i f i ed for the feature [SG] is prob lemat ic for the ana lys is o f cont inuant v o i c i n g , chapter five, and the ana lys is o f the d-effect, sect ion 4.8. Thus I assume h i s to r i ca l l y that f r icat ives were spec i f i ed for the feature [ SG ] , but synchron i ca l l y the feature [ SG ] is not act ive as a d is t inc t i ve spec i f i ca t ion on fr icat ives in the phono logy . W h e n the h- c lass i f ie r is present, verb stem-init ia l f r i cat ives are cons is tent ly vo ice less . It is poss ib le that (a) the h- c lass i f i e r remains unparsed or (b) the h- c lass i f i e r associates vacuous l y . I f poss ib i l i t y (b) is assumed, once aga in , it has to be assumed that f r icat ives are spec i f i ed fo r the feature [ SG ] . A s p rev ious l y ment ioned , th is is : (a) not consistent w i th the La r yngea l M e c h a n i s m Hypothes i s and (b) p rob lemat i c for the analys is o f cont inuant v o i c i n g , chapter 5, and the d-effect, sect ion 4.8. In add i t ion , i f poss ib i l i t y (b) is assumed, the f o l l o w i n g quest ion ar ises: w h y doesn ' t 43 the feature [SG] associate to vo ice less unaspirated stops? U n d e r the La r yngea l M e c h a n i s m Hypothes i s vo ice less unaspirated stops are unspec i f i ed for laryngeal spec i f i ca t ion . I f the h-c lass i f i e r associates to verb stem-init ia l consonants , one w o u l d expect vo ice less aspirated stops to surface as vo ice less unaspirated stops. 8 F o r examp le : (19) * n [gani] n [kani ] [SG] \ / [SG] 11 12 11 12 Poss ib i l i t y (b) is prob lemat ic fo r a number o f reasons; thus I assume the h- c lass i f i e r does not associate to verb stem-init ial consonants, i.e. remains unparsed. 4.6.3 Summary In th is analys is the h- c lass i f ie r is anay lzed as a f loa t ing [SG] feature. In the th i rd person s ingular and p lu ra l , the h- c lass i f i e r is rea l ized phone t i ca l l y as [h] in the penul t imate coda pos i t ion . In the second person s ingular , the h- c lass i f i e r associates to the f ina l consonant , [n] o f the subject morpheme, devo i c i ng it to [n]. In the f i rst person s ingular and second person p lu ra l , the h- c lass i f ie r associates vacuous l y to the subject morpheme. It never associates to a ver stem-in i t ia l segment to its r ight. 4.7 1- c lass i f ie r 4.7.1 M o r p h o p h o n e m i c effects o f the 1- c lass i f i e r In A thapaskan languages the 1- c lass i f i e r is hypothes ized to be der i ved f r om the c lass i f i e r To solve this problem one could assume plain stop consonants are specified for the feature [voice], and that the presence of the feature [voice] would disallow the feature [SG] to associate due to the co-occurrence constraint *[SG]/[voice]. For example: n gani in gani [SG] [voi] -> [SG] [voil 11 12 11 12 One issue to consider, with regards to this solution, is that the phonological and phonetic specifications are not consistent. Recall that in the phonetic study, discussed in chapter two, plain stops were voiceless unaspirated, not voiced. Therefore I hypothesize stop consonants are not specified for the feature [voice]. 44 comb ina t i on * d - + * i - (d + h) (Stanley 1969) . 9 In Tah l t an it is not apparent the 1- c lass i f i e r is der i ved f r o m this comb ina t i on ; however i n some A thapaskan languages (i.e. S lave , Sekan i ) it is c lear l y apparent that this is the case. Re ca l l that the d- and h- c lass i f ie rs have der ivat iona l funct ions . In S lave , when the d- c lass i f i e r is der i va t iona l l y added to an h-classif ier verb the effect is that o f the 1- c lass i f ier . F o r example : Obj. + de + d + h + [sa]; [deza] ( R i c e 1989: pg . 458) . In Tah l tan the 1- c lass i f ie r never surfaces as an independent segment. O n e can te l l the 1-c lass i f i e r is present by its phono log i ca l effects on verb stem-init ial f r icat ives . T h e fr icat ives affected by the 1- c lass i f ie r are ident i f ied in the Tah l tan consonant inventory in (20) . (20) F r ica t ives affected by the I- c lass i f i e r labial coronal dorsal laryngeal labial alveolar inter-dental alveolar lateral palatal velar labio-velar uvular glottal a. b b. t te ts t i ts k k w q ? c. th teh ts h tih t s h k h kwh q h d. t' te' ts' t i ' ts ' k' k w ' q ' e. e s i s (~9) X x w X h f. 5 z 1 z (~ y) Y Y w K g- m n w a. voiced stop e. voiceless fricative b. voiceless unaspirated stop f. voiced fricative c. voiceless aspirated stop g. sonorant d. glottalized stop The vo i c ed counterparts o f the lateral consonant , [I], i.e. [1], and the palatal consonant , [§]> ' e - [y]> a r e n o t f r icat ives phonet ica l l y . H o w e v e r , in A thapaskan languages the lateral consonant [1] and the palatal consonant [y] f o r m a natural class w i th the f r icat ive consonants and thus are grouped w i th the f r icat ives in the consonant inventor ies . 9 The fact that the 1- classifier derived from the *d- + * i - sequence, provides evidence that plain stops are voiced. *d + * i _> 1 [stop] l iar ] [lat] [lar][lat] [voi] [voi] The phonetic evidence, present in chapter x, does not support this synchronically. Recall that the voiceless unaspirated stop was voiceless unaspirated. To account for the fact 1- derives from *d- + * i - (h), I assume historically plain stops were specified for the feature [voice], but synchronically the feature [voice] is not active as a distinctive specification on stops in the phonology 45 W h e n the 1- c lass i f i e r is present, verb stem-init ia l f r icat ives are v o i c e d , except in the f irst person p lura l where the d-effect (d iscussed in sect ion 4.8) occurs . U n l i k e the h- c lass i f ie r , the 1-c lass i f i e r does not interact w i t h preced ing subject morphemes . Par t i cu la r l y s ign i f i cant is the fact that the h- c lass i f ie r surfaces as an independent segment in the th i rd person s ingular and th i rd person p lu r a l ; under the same c i rcumstances the 1- c lass i f i e r does not surface as an independent segment. In the verba l paradigms in (21), the character ist ics o f the 1- c lass i f i e r are e x e m p l i f i e d . (21) Phono log i c a l effects o f the 1- c l a s s i f i e r 1 0 morpheme breakdown Tah l tan example E n g l i s h G l o s s 8 8 11 12 ne i s 1 [0it] n i 0 [Sit] I got hot 8 8 11 12 ne i n 1 [0it] n i n [Sit] You (sg) got hot 8 8 11 12 ne i 1 [Git] n i [Sit] He got hot 8 8 11 12 ne i 0 id 1 [Git] ne Gi [dSit] We got hot 7 8 8 12 he ne i 1 [Git] he n i [Sit] They got hot 3 11 12 ka s 1 [sei] ka s [zei ] I hollered 3 11 12 ka en 1 [sei] ka n [zei] You (sg) hollered 3 11 12 ka 1 [sei] ka [zei] He hollered 3 11 12 k a G i d 1 [sei] ka s i [dzei] We hollered 3 11 12 ka ah 1 [sei] ka da h [zei] You (pi) hollered 7 3 11 12 he ka s 1 [sei] ka he [zei] They hollered 1 0 Unfortunately in (21), there are no verbs with a stem-initial [i]. In other morphophonemic processes which target fricatives, [i] is affected (the d-effect, continuant voicing), since this is the case I strongly believe this is an accidental gap. However in order to be absolutely certain, more data needs to be collected. 46 11 12 s h [ca?e] e s [ya?e] I got sick 11 12 n h [ca?e] e n [ya?e] You (sg) got sick 11 12 h [ca?e] e [ya?e] He got sick 11 12 B id h [ca?e] e s i [dza?e] We got sick 11 12 ah h [ca?e] ah [ya?e] You (pi) got sick 7 12 he h [ca?e] e he [ya?e] They got sick 3 9 10 11 12 na 0e n s 1 [xi] na s [ r i ] I saved O A t present, it is unclear i f the 1- c lass i f i e r affects verb stem-init ia l stop consonants . The uncertainty has to do w i th the fact that the 1- c lass i f i e r is never rea l ized as an independent segment; thus i f a verb has a stop in stem-init ia l pos i t i on one cannot te l l i f the 1- c lass i f i e r is present o r not. Second ly , the 1- c lass i f i e r has no apparent der ivat iona l f unc t i on ; thus one cannot refer to a spec i f i c const ruct ion and k n o w fo r certa in the 1- c lass i f i e r is present. H o w e v e r , the process o f v o i c i n g ass imi l a t ion (sect ion 4.9) and cont inuant v o i c i n g (chapter five) do not affect stop consonants ; thus I hypothes ize the 1- c lass i f i e r does not affect stop consonants. 4.7.2 A n a l y s i s o f the 1- c lass i f i e r Re ca l l that the goa l o f this chapter is to examine the morphophonemi c effects o f the c lass i f ie r pref ixes to test the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s . In this sect ion it w i l l be shown that the La ryngea l M e c h a n i s m Hypo thes i s can account fo r the morphophonemi c effects o f the 1- c lass i f ier . M o r e spec i f i ca l l y it w i l l become c lear that the laryngeal feature [voice] is phono log i c a l l y act ive in the f r icat ive c lass. F o l l o w i n g H a r d w i c k (1984) on Tah l tan , Ha rgus (1985) on the c lose l y related language Sekan i and R i c e (1989) on S lave , I hypothes ize that the 1- c lass i f i e r is a f loa t ing feature [voice] . (22) 1- c lass i f i e r k a s [sei] [voice] 3 11 12 47 Throughout the verba l pa rad igm, the 1- c lass i f i e r associates to verb stem-init ia l f r icat ives . In other words , when the 1- c lass i f i e r is present, verb stem-init ia l f r icat ives acquire the laryngeal spec i f i ca t ion [voice] . T h i s is exemp l i f i ed in (23). (23) k a s [sei] k a s [zei] [voice] —>• / [voice] 3 11 12 3 11 12 U n l i k e the h- c lass i f ie r , the 1- c lass i f ie r has no effect on the subject morphemes . I f the 1-c lass i f ie r affected the subject morphemes , one w o u l d expect the f i rst person s ingular subject morpheme [s] to surface as [z]. N o t e further that there is no regressive ass imi l a t ion o f vo i ce either: the vo ice less [s] and vo i ced [z] are cont iguous in the output str ing, fo r example [kasze i ] , I hollered. 4.7.3 Summary W h e n the 1- c lass i f ier , ana lyzed as a feature [voice] , is present, it associates to verb stem-in i t ia l cont inuants ; thus f r icat ives are cons is tent ly spec i f i ed fo r the laryngeal feature [voice] f o l l o w i n g the 1- c lass i f ier . A t present, it is unclear i f stop consonants are affected by the 1-c lass i f ier . The 1- c lass i f ie r , un l ike the h- c lass i f ie r , has no overt effect on subject morphemes . 4.8 d- c lass i f ie r 4.8.1 M o r p h o p h o n e m i c effects o f the d- c lass i f i e r In Tah l tan the d- c lass i f i e r never surfaces as an independent segment. H o w e v e r , in other A thapaskan languages (i.e. H u p a , K o y u k o n ) the d- c lass i f i e r surfaces as a segment; thus it is referred to as the d- c lass i f ier . In Tah l tan , the d- c lass i f iers rea l izat ion w i t h a ?-initial stem is [ f ] , w h i c h is what establishes its [coronal] va lue . O n e can te l l the d- c lass i f ie r is present by its phono log i ca l effect on verb stem-init ia l f r icat ives and glotta l stop. The consonants af fected by the d- c lass i f ie r are ident i f ied in the Tahl tan consonant inventory in (24). 48 (24) Consonants affected by the d- c lass i f i e r labial coronal dorsal laryngeal labial alveolar inter-dental alveolar lateral palatal velar labio-velar uvular glottal a. b b. t te ts ti ts k kw q ? c. t h teh ts h t i h ts h kh kwh qh d. t' te' ts' ti' ts ' k' kw' q' e. e s i s (~S) X x w X h f. 8 z 1 z (~ y) Y Y w K g- m n w a. voiced stop e. voiceless fricative b. voiceless unaspirated stop f. voiced fricative c. voiceless aspirated stop g. sonorant d. glottalized stop In the A thapaskan literature the interact ion between the d- c lass i f ie r and verb stem-init ia l consonants is referred to as the d-effect ( H o w r e n 1971). A s seen in (25.a) the consonant that surfaces as a result o f the d-effect has the stop qua l i ty o f the [d] and the p lace ar t icu lat ion o f the f r icat ive consonant; (25.b) shows its app l i ca t ion 'on a ?-stem; and (25.c) represents its fa i lure to effect any surface change on any other consonant. (25) D-effect alternations ( H a r d w i c k 1984) c lass i f ie r stem-init ia l output consonant a. d + i H> d l d + 6 _> dS d + s _> d z d + x g d + x w _> g w b. d + ? t' c. d + C -> C The alternations in (25) are e x e m p l i f i e d in (26) and (27). In (26) , the d-effect is t r iggered by the presence o f the d- c lass i f ier . In the lef tmost c o l u m n o f table (26) a Tah l tan verb is presented, two examples are g i ven w i t h each verb . T h e first example in each set shows that i f the 0 - or h- c lass i f ie r is present, then the d-effect does not occur . The second example in each set is in the f irst person s ingular re f l ex ive (myse l f ) . The re f l ex ive p re f ix [?ede] (pos i t ion 9) 49 cons is tent ly co-occurs w i t h the d c lass i f ie r ; therefore the d-effect does occur . T o c lear l y indicate where the phono log i ca l change is t ak ing p lace , the stem is ident i f ied i n brackets in each example . (26) E x a m p l e s o f the d-effect t r iggered b y the d- c lass i f i e r morpheme b reakdown Tah l tan example E n g l i s h g loss 9-10 11-12 0e n s h [xin] Ge h [xin] I killed O 6 9-10 11 12-?ede Ge n s d [xin] ?ede se s [gin] I killed myself 9-10 11-12 y e n s h [x w ase ] y i h [x w ase ] I tickled him 6 8 9-10 11 12-?ede de ye n s d [x w ase ] ?ede de s [g w ase] I tickled myself 9-10 11 12 G e n s 0 [?un] Gi [?un] I shot O 6 9-10 11 12-?ede Ge n s d [?un] ?ede s [t 'un] I shot myself In add i t ion to the d- c lass i f ie r , the final consonant o f the first person p lura l ([Gid]) (pos i t ion 11) interacts w i t h verb stem-init ia l f r icat ives . T h i s interact ion between the final consonant o f the first person p lura l ([Gid]) (pos i t ion 11) and verb stem-init ia l f r icat ives is a lso subsumed under the d-effect in the A thapaskan l iterature. The consonants that surface as a result o f these interact ions are ident ica l to those in (25). In (27) the d-effect is tr iggered b y the final consonant o f the first person p lura l [Gid]. S i m i l a r to (26), two examples are p rov ided w i t h each verb . The first example o f each set is in the first person s ingular ; thus the d-effect is not present. The second example is in the first person p lu r a l ; therefore the d-effect is present. In these examples the h-, 0 - or 1- c lass i f ie rs is present; these c lass i f iers show no overt effect in the presence o f the d-effect. 50 (27) E x a m p l e s o f the d-i morpheme b reakdown effect t r iggered by the first person p lu ra l [0id] Tah l tan example E n g l i s h g loss 8 8 9-10 11 12 u d e y e 0 s h [iet] ?u de h [let] I melted it 8 8 9-10 11 12 u de ye 0 9 id h [iet] ?u de 9 i [diet] We melted it 11-12 e s h [0e] e h [0e] I'm tanning it 11-12 e Gid h [0e] 0 i [doe] We 're tanning it 8 11 12 de s h [xu i ] de s [xui ] I'm throwing it 8 11 12 de Gid h [xui ] de 0 i [gui] We 're throwing it 9-10 11-12 0 e n s 0 [ i ih ] 0 i [ l ih] I tasted it 9-10 11-12 G e n 0 i d 0 [ i ih ] 0 i [dl ih] We tasted it 8 11 12-de s 0 [seh] de s [s:eh] I'm spitting 8 11 12 de Gid 0 [seh] de s i [dzeh] We 're spitting 8 9-10 11 12 de 0 n s 0 [xui ] de s [xui ] I scraped it 8 9-10 11 12 de 0 n s 0 [xui ] de Gi [gui] We scraped it 1 2 3 9 10 11 12 ?e ya tA Ge n s 0 [?otG] ?e r a tA Gi [?otG] I washed it 1 2 3 9 10 11 12 ?e ya tA Ge n Gid 0 [?otG] ? e r a tA Gi [t 'otG] We washed It 3 11 12 ka s 1 [sei] k a s [zei] I hollered 3 11 12 ka 0 id 1 [sei] k a s i [dzei] We hollered 4.8.2 A n a l y s i s o f the d- c lass i f i e r R e c a l l that the goa l o f this chapter is to examine the morphophonemi c effects o f the c lass i f i e r pref ixes o t st the c l a ims o f th  La r ynge l M e c h a n i s m Hypothes i s . In th is sect ion it 51 w i l l be shown that the La r yngea l M e c h a n i s m Hypo thes i s can account for the m o r p h o p h o n e m i c effects o f the d- c lass i f ier . F o l l o w i n g Shaw (1991) on Tah l tan , I assume the d- c lass i f i e r is a feature [stop], w h i c h precedes the verb stem and f o l l o w s the subject. (28) d- c lass i f ie r ?ede s [sei] [stop] 6 11 12 W h e n the d- c lass i f ie r is present, the feature [stop] associates to the verb stem-init ia l f r icat ives . The output is a consonant spec i f i ed for the feature [stop] and no laryngeal spec i f i ca t ion , i.e. a vo ice less unaspirated stop consonant . (29) ?ede s [sei] ?ede s [dzei ] [ s t o p ] " * [stop] 6 11 12 6 11 12 The d- c lass i f i e r a lso interacts w i th glotta l stop. F o l l o w i n g Shaw (1991) , I assume consonants unspec i f i ed fo r p lace features rece ive the p lace feature [coronal] as a default oral p lace spec i f i ca t ion . The markedness assumpt ions o f Shaw ' s analys is w o u l d be interpreted in the Op t ima l i t y Theo ry f r amework adopted here in terms o f the p lace markedness h ierarchy * L a b i a l , * Dorsa l » * C o r o n a l » ""Pharyngeal (see Shaw 1991 for mot i va t ion o f this f r om Tah l tan consonant harmony) . (30) ?ede s [?un] ?ede s [ t 'e i ] [stop] [lar] -> / A | [stop] [lar] [pi] [CG] | | [CG] [cor] 6 11 12 6 11 12 In the d iscuss ion o f the h- c lass i f ie r it was suggested that there is a poss ib i l i t y that f r icat ives are spec i f i ed for the feature [ SG ] . In this d i scuss ion it was a lso ment ioned this was prob lemat i c fo r the analys is o f the d-effect. I f f r i cat ives are spec i f i ed fo r the feature [ SG ] , one w o u l d expect an aspirated stop to surface as a result o f the d-effect, not a vo ice less unaspirated stop. F o r examp le : 52 (31) ?ede s [sei] * ?ede s [tsei] /\ [stop] [SG] 6 11 [stop] [ SG ] 12 6 11 12 T h i s lack o f aspirat ion on der ived stop consonants supports the hypothes is that f r icat ives are not spec i f i ed fo r the laryngeal feature [ SG ] . Th i s is consistent w i t h the La r yngea l M e c h a n i s m Hypothes i s . The d- c lass i f i e r does not interact w i t h the preced ing subject morpheme. I f it d i d , one w o u l d expect the f irst person s ingular subject morpheme , [es] to surface as [edz]. S im i l a r to the 1-c lass i f ier , the d- c lass i f i e r never surfaces as an independent segment. R e c a l l that the th i rd person s ingular and th i rd person p lura l are both m o r p h o l o g i c a l l y unmarked i n pos i t ion 11. T h e h-class i f ie r surfaces as an independent segment, [h] in these pos i t ions ; the d- c lass i f i e r under the same c i rcumstances does not surface as an independent segment. 4.8.3 Summary W h e n the d- c lass i f ier , represented as a feature [stop], is present the [stop] feature comb ines w i t h the place features o f the f o l l o w i n g fricative consonant ; the result is a vo ice less unaspirated stop. The d- c lass i f i e r a lso interacts w i t h glottal stop. The result is [ f ] , a consonant w i t h the feature [stop] f r o m the d- c lass i f ier , the feature [ C G ] f r om the glotta l stop and the p lace feature [coronal] f r om the rea l izat ion o f the least marked ora l cav i t y P lace ar t icu la t ion. 4.9 0 - c lass i f i e r The 0 - c lass i f i e r never surfaces as an overt segment. W h e n the 0 - c lass i f i e r is present stem-init ia l f r icat ives are vo ice less f o l l o w i n g a vo ice less segment, and are vo i ced f o l l o w i n g a vo i c ed segment ( vo i c i ng ass imi la t ion ) . T h i s process o f v o i c i n g ass imi l a t ion is exemp l i f i ed in (32). N o t i c e that the morphophonemic effects o f the 0 - c lass i f i e r are un l i ke the morphophonemi c effects o f the 1- c lass i f ie r . W h e n the 1- c lass i f ie r is present, stem-init ia l f r icat ives are consistent ly vo i c ed . In contrast, w h e n the 0 - c lass i f ie r is present, f r i cat ives are on l y vo i c ed f o l l o w i n g vo i c ed segments. S i m i l a r to the h- and 1- c lass i f ie r , the d-effect occurs in the first person p lu ra l . 53 (32) V o i c i n g ass imi la t ion morpheme b reakdown Tahl tan example E n g l i s h g loss 3 11 12 da s 0 [tei] da s [ ie i ] I'm holding pi O 3 11 12 da n 0 [te*] da n [ lei ] you 're holding pi O 3 11 12 da 0 [tei] da ya [ lei ] he's holding pi O 3 11 12 da Gid 0 [ ie i ] da Gi [dle i ] we 're holding pi 0 3 11 12 da ah 0 [tei] d ah [ ie i ] you 're (pi) holding pi 0 3 7 12 da he 0 [tei] da ha ya [ le i ] they 're holding pi O 8 11 12 de s 0 [seh] de s [seh] I'm spitting 8 11 12 de n 0 [seh] di n [zeh] you 're spitting 8 11 12 de 0 [seh] de [zeh] he's spitting 8 11 12 de Gid 0 [seh] de si [dzeh] we 're spitting 8 11 12 de ah 0 [seh] d ah [seh] you 're (pi) spitting 7 8 12 he de 0 [seh] ha de [seh] they 're spitting 3 11 12 na s 0 [Get] na G [Get] I'm standing 3 11 12 na n 0 [Get] na n [Set] you 're standing 3 11 12 na 0 [Get] na [6et] he's standing 3 11 12 na Gid 0 [Get] na Gi [doet] we 're standing 3 11 12 n a a h 0 [Get] n ah [Get] you 're (pi) standing 3 7 12 na he 0 [Get] na he [Set] they 're standing In the perfect ive, when the 0 - c lass i f ie r is present, the first person s ingular subject deleted. Th i s is exemp l i f i ed in (33). (33) P h o n o l o g i c a l effects o f the 0 - c lass i f i e r imper fec t ive ie ne s [tsis] I'm folding perfect ive ie n i [tsis] I folded imper fec t ive de s [seh] I'm spitting perfect ive d i [zeh] I spat 4.9.1 A n a l y s i s o f the 0 - c lass i f i e r R e c a l l that the goa l o f this chapter is to examine the morphophonemi c effects o f the c lass i f i e r pref ixes to test the c l a ims o f the L a r y n g e a l M e c h a n i s m Hypothes i s . In this sect ion it w i l l be shown that the La r yngea l M e c h a n i s m Hypo thes i s can account for the m o r p h o p h o n e m i c effects o f the 0 - c lass i f ier . W h e n the 0 - c lass i f i e r is present, verb stem-init ia l f r icat ives are vo i c ed in the second person s ingular , th i rd person s ingular and th i rd person p lu ra l . I hypothes ize that stem-init ia l f r icat ives acquire the laryngeal spec i f i ca t ion [voice] f r o m a preced ing v o w e l or nasal consonant. T h i s requires the add i t iona l hypothes is that the feature [voice] is phono log i c a l l y act ive on vowe l s and nasals. (34) Second person s ingular de n 0 [seh] de n I - > I [voice] [voice] 8 11 12 8 11 12 (35) T h i r d person s ingular de 0 [seh] I [voice] 8 11 12 de 0 [zeh] [ vo i ce ] ' 8 11 12 (36) T h i r d person p lura l he de 0 [seh] [voice] 8 12 he de [voice] 7 8 12 In the f i rst person s ingular and second person p lu ra l , verb stem-init ia l f r icat ives do not rece ive a laryngeal spec i f i ca t ion [voice] ; thus these f r icat ives surface as vo ice less . 55 (37) F i r s t person s ingular de s 0 [seh] de s 0 [seh] 8 11 12 . 8 11 12 (38) Second person p lura l de ah 0 [seh] de ah 0 [seh] -> 8 11 12 8 11 12 4.9.2 Summary T h e morphophonemic effects o f the 0 - c lass i f i e r can be accounted for under the La r yngea l M e c h a n i s m Hypothes i s . The behav ior o f the 0 - c lass i f ie r requires the add i t iona l hypothes is that the feature [voice] is phono log i c a l l y act ive on vowe l s and nasals. 4.10 C o n c l u s i o n T h e La ryngea l M e c h a n i s m Hypo thes i s accounts fo r the laryngeal behav ior o f the morphophonemic effects o f the h-, 1-, d- and 0 - c lass i f ie rs . Spec i f i c a l l y , the h- c lass i f i e r accounts fo r the hypothes is the laryngeal feature [SG] is act ive w i t h i n the phono logy . The 1- and 0 - c lass i f iers account for the hypothesis the laryngeal feature [voice] is phono log i c a l l y act ive in the fricative c lass. T h e 0 - c lass i f ie r requires the add i t iona l hypothes is that the feature [voice] is phono log i ca l l y act ive on vowe l s and nasals. 56 Chapter Five Noun Stem-Initial and Stem-Final Voicing Alternations Cont inuant v o i c i n g is a f ami l i a r morphophonemi c process in A thapaskan languages ( K a r i 1973, C o o k 1984, Hargus 1988, R i c e 1988, 1991 and others). W h e n the process o f cont inuant v o i c i n g is present, noun stem-init ia l vo ice less f r icat ives surface as vo i c ed f r i ca t ives . 1 In A thapaskan languages, there is a lso a process o f word-f ina l devo i c i ng . W h e n the process o f word-f ina l devo i c i ng is present noun stem-final vo i c ed f r icat ives surface as vo ice less f r i cat ives . T h i s chapter w i l l examine data from Tah l tan , w h i c h exhib i ts the process o f cont inuant v o i c i n g and word-f ina l devo i c i ng , in order to test the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s . Th i s chapter is d i v i ded into four sect ions. In sect ion 5.1, the La r yngea l M e c h a n i s m Hypothes i s is restated. In sect ion 5.2, the morphophonemi c process o f cont inuant v o i c i n g is exemp l i f i ed and an analys is fo r the process o f cont inuant v o i c i n g is presented. In sect ion 5.3, the phono log i ca l process o f word-f ina l devo i c i ng is exemp l i f i ed and an analys is fo r the process o f word-f ina l devo i c i ng is presented. F i na l l y , in sect ion 5.4 c o n c l u d i n g remarks are made. 5.1 Theoret ica l A s s u m p t i o n s Reca l l that the goa l o f this chapter is to examine the processes o f cont inuant v o i c i n g and word-f ina l devo i c i ng in order to test the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s . T h e La ryngea l M e c h a n i s m Hypothes i s is restated in (1). (1) La ryngea l M e c h a n i s m Hypothes i s ( rev ised f r o m R i c e 1994) b voiceless aspirated stops voiceless unaspirated stops glottalized stops voiceless fricatives voiced fricatives glottal stop h stop X X X X voice X X SG X X C G X X 5.2 Cont inuant V o i c i n g T h e process o f cont inuant v o i c i n g is present in many A thapaskan languages (e.g. N a v a j o , Sarcee, S lave , Sekan i and others). W h e n cont inuant v o i c i n g is present, vo ice less f r icat ives surface as vo i c ed f r icat ives. The process o f cont inuant v o i c i n g does not affect a l l f r i cat ive 1 In Athapaskan languages, the process o f continuant voic ing also affects preposition stem-initial fricatives. A t present, in Tahltan, there are no examples which exhibit the continuant voic ing o f prepositions stem-initial fricatives; thus this thesis w i l l focus on the continuant voicing o f noun stem-initial fricatives. 57 consonants. Instead, it affects noun stem-initial fricatives, but only when the noun stem is preceded by derivational morphology. The consonants affected by the process of continuant voicing, i.e. the voiceless fricatives, are identified in the Tahltan consonant inventory in (2). The consonants, which surface as a result of continuant voicing, i.e. the voiced fricatives, are also identified in (2). (2) Consonants Affected by Continuant Voicing labial coronal dorsal laryngeal labial alveolar inter-dental alveolar lateral palatal velar labio-velar uvular glottal a. b b. t te ts ti ts k k w q ? c. th teh tsh t i h ts h k h k w h q h d. t' te' ts' t i ' ts' k' q' e. o e , s i s (~9) X x w X h f. e 6 z 1 z (~ y) Y Y w %• m n w a. voiced stop e. voiceless fricative b. voiceless unaspirated stop f. voiced fricative c. voiceless aspirated stop , g. sonorant d. glottalized stop Notice that the process of continuant voicing does not affect the fricative [h]. First, [h] does not have a voiced counterpart, and secondly, [h] is specified for the feature [SG], and in this analysis the feature [SG] does not co-occur with the feature [voice]. 5.2.1 Data When a noun stem is preceded by the possessive morpheme, the process of continuant voicing is present.2 In this section, possessed nouns, which exhibit the process of continuant voicing, are presented, but prior to this the Tahltan possessive morphemes are identified. In Tahltan, the possessive morphemes consist of a prefix and the suffix [e].3 The possessive morphemes are presented in (3). 2 In Athapaskan languages, the process of continuant voicing also affects compounds. Unfortunately, at present, in Tahltan, there is only one example, which exhibits continuant voicing in compounds; thus continuant voicing in compounds will not be discussed in this thesis. Example: [tGiyu] hair [tGi] head +[xu] hair. 3 In the possessive construction, the noun stem is not always followed by the possessive suffix [e]. For example, noun stems which end in a vowel are not followed by the possessive suffix. 58 (3) Tahltan possessive morphemes (Hardwick 1984) possessive prefix possessive suffix first person singular second person singular es-en--e -e third person singular first person plural me-dah-hu-hu--e -e second person plural third person plural -e -e The first person singular and the first person plural have final consonants, which are voiceless, i.e. [s] and [h] respectively, while the remaining prefixes end in a vowel or a nasal consonant. This is significant because the process o f continuant voicing still occurs even though a voiced segment does not precede the noun stem-initial fricative. In other words, the fact the process o f continuant voicing occurs following a voiceless consonant is evidence that this is not a phonetic process. The process o f continuant voicing is exemplified in (4). In the leftmost column of (4) are independent noun stems. Notice that the stem-initial fricatives are voiceless. In columns two and three, the noun stem is preceded by a possessive prefix; thus the process o f continuant voicing is present. In column two, the noun stem is preceded by the possessive prefix [es], first person singular. In column three, the noun stem is preceded by the possessive prefix, [me], third person singular. 4 Notice that the process o f continuant voicing is present following the voiceless fricative [s] and following the vowel [e]. (4) Continuant Voic ing independent noun stem first person singular third person singular English gloss led es led e me led e smoke iu:d es lud e me lu:d e scab iuwe es luwe me luwe fish Be: eG 6e: me 3e: belt se:g es zeg e me ze:g e spit sei es zei e me zei e hook cm es yine me yin e song cina es yina me yina slave xei es yel e me yel e trap xei es yel e me yel e packsack xos es yoz e me yoz e thorn 4 The process of continuant voicing is present following all possessive prefixes. The first person singular and third person singular are exemplified in (4) in order to show that the process of continuant voicing occurs following voiceless consonants and following voiced segments. 59 The process of continuant voicing does not affect a second class of noun stem-initial fricatives. The independent noun stem and the possessed noun both have voiced stem-initial fricatives. This is exemplified in (5). (5) Stem-initial Voiced Fricatives independent noun stem first person singular third person singular English gloss 8a0 e0 5a0 e me 8a0 e snow yAnje es yAnje me y A n J e geese The process of continuant voicing does not affect noun stem-initial stop consonants. When a possessive prefix is added to an independent noun stem, the stem-initial stop consonant does not change. This is exemplified in (6). (6) Noun Stem-initial Stop Consonants independent noun stem first person s ingular th i rd person s ingular Eng l i s h gloss be:s es be:z e me be:z e knife bede es bede me bede food dih es d ih e me d ih e grouse dec in es dec in e me dec in e stick dlune es dlune me dlune mouse d6 iya es d f i i ya 'e me dd i ya ' e necklace gah es gah e me gah e rabbit gaw es gaw e me gaw e drum tene es tene me tene road tu:nade01in es tu:enadeGlin me tu:enade01in waterfall t9e: e0 t0e: me t0e: rock tsic es tsic me tsic. nose ca:k 'od le es ca :k ' od l e rain bucket ke: es ke? me ke? shoe/foot ku:g es ku:g e me ku:g e trunk t'iyA es t'iyA me t'iyA hook t 'o:g es t 'o:g e me t 'o:g e wart t0 'alu:d5e e0 t0 'alu:dde eyebrows tG 'art es tO 'a t i me t 0 ' a t i moss/diapers k ' ug es k ' u g e me k ' ug e paper 5.2.2 Analysis In this section, previous analysis of continuant voicing will be presented, followed by the present analysis of continuant voicing. The process of continuant voicing has been analyzed as a process of continuant devoicing (Kari 1973, Cook 1984). For Navajo, Kari (1973) hypothesizes that stem-initial fricatives are underlyingly specified for the laryngeal feature [voice]. To account for voiceless stem-initial 60 f r icat ives in word- in i t ia l pos i t i on , K a r i (1973) assumes noun stem-init ia l f r icat ives lose the spec i f i ca t ion [voice] when in word- in i t ia l pos i t i on . C o o k (1984) hypothes izes the same fo r the A thapaskan language Sarcee. R i c e (1988 , 1989, and 1991) disagrees w i t h K a r i (1973) and C o o k (1984) . R i c e (1988, 1989, and 1991) prov ides strong ev idence in favor o f the cont inuant v o i c i n g hypothes is . F i rs t , a d e v o i c i n g hypothes is does not account for independent noun stems w i t h v o i c e d f r icat ives , i.e. the data in (5), restated in (7). (7) Stem-Init ial V o i c e d Fr i ca t i ves independent noun stem f irst person s ingular th i rd person s ingular E n g l i s h g loss 3 a 0 e 0 3 a 0 e me 3 a 0 e snow yAnje es Y A n J e m e Y A n J e geese Second , when a noun stem is incorporated into the verb c o m p l e x , noun stem-init ia l f r icat ives are vo ice less . Th i s is e x e m p l i f i e d in the f o l l o w i n g example . R i c e analyzes data f r om S lave . H o w e v e r , the f o l l o w i n g example is f r o m Tah l tan shows that this same genera l izat ion ho lds in Tah l tan as w e l l . (8) Incorporated N o u n Stem independent noun stem E n g l i s h gloss incorporated noun stem E n g l i s h g loss x a hair me[xa]?esits I combed O (a dog) The example is (x) is prob lemat i c fo r the d e v o i c i n g hypothes is . Reca l l that K a r i (1973) and C o o k (1984) hypothes ize that stem-init ia l f r icat ives are under l y ing l y spec i f i ed for the feature [vo ice] , and that they loss the spec i f i ca t ion [voice] in word- in i t i a l pos i t ion . In (8), the noun stem-init ia l f r i cat ive is vo ice less , however the noun stem-init ia l f r i cat ive is not in word- in i t i a l pos i t ion . T o account fo r the process o f cont inuant v o i c i n g , R i c e (1988, 1989, and 1991) hypothes izes that stem-init ia l f r icat ives are unmarked for the laryngeal spec i f i ca t ion [voice] . R i c e (1988 , 1989 and 1991) hypothes izes that a l i n k i n g morpheme, w h i c h consists o f the feature [voice] , is inserted when a noun stem is preceded by a spec i f i c set o f s morphemes , such as the possessive morpheme. R i c e (1988, 1989 and 1991) assumes that the l i n k i n g morpheme, w h i c h consists o f the feature [voice] , associates to noun stem-init ia l f r icat ives , resu l t ing in a vo i c ed f r icat ive . Th i s is exemp l i f i ed in (9). 61 (9) Cont inuant V o i c i n g (R i ce 1988, 1989 and 1991) independent noun stem possessed noun stem led es led e I [voice] Reca l l that there are noun stem-init ia l f r i cat ives , w h i c h are vo i c ed in the independent noun stem and in the possessive const ruct ion (see (5) o r (7)). R i c e (1989) hypothes izes that these f r icat ives are spec i f ied under l y ing l y fo r the feature [voice] . A s a result, they do not part ic ipate in the process o f cont inuant v o i c i n g . T h i s is exemp l i f i ed in (10). (10) F r ica t ives U n d e r l y i n g l y Spec i f i ed for the Feature [voice] independent noun stem possessed noun 6a0 es 6a9 e I I [voice] [voice] Reca l l that the goa l o f this chapter was to examine the process o f cont inuant v o i c i n g in order to test the c l a ims o f the La ryngea l M e c h a n i s m Hypo thes i s . The cont inuant d e v o i c i n g hypothes is proposed by K a r i (1973) and C o o k (1984) , and the cont inuant v o i c i n g hypothes is proposed by R i c e (1988, 1989 and 1991) both support the La r yngea l M e c h a n i s m Hypothes i s . In other words , both analyses support the hypothes is that the laryngeal feature [voice] is phono log i ca l l y act ive w i t h i n the f r icat ive c lass. A s w e w i l l see, the present ana lys is a lso supports the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s . Spec i f i ca l l y , the present ana lys is , l ike the prev ious ana lys is , supports the c l a i m that the feature [voice] is phono log i c a l l y act ive w i t h i n the f r icat ive c lass. F o l l o w i n g R i c e (1988, 1989 and 1991), I assume this is a process o f cont inuant v o i c i n g , not a process o f cont inuant devo i c i ng as K a r i (1973) and C o o k (1984) have hypothes ized . A l s o f o l l o w i n g R i c e (1988, 1989 and 1991), I hypothes ize stem-init ia l f r icat ives are unspec i f ied for the laryngeal spec i f i ca t ion [voice] . U n d e r the L a r yngea l M e c h a n i s m Hypothes i s f r icat ives unspec i f ied fo r laryngeal spec i f i ca t ion are vo ice less . T o account fo r the process o f cont inuant v o i c i n g , i.e. to account fo r the fact noun stem-init ia l f r icat ives are vo i c ed in the possessive const ruct ion , I assume noun stem-init ia l f r icat ives rece ive a laryngeal spec i f i ca t ion [voice] f r o m a " s t em j o i n e r " or " l i n k i n g m o r p h e m e " , w h i c h consists o f the feature [voice] . I hypothes ize that the stem j o i n e r [voice] is present in certain n o m i n a l der ivat iona l construct ions , such as the possess ive construct ion. W h e n the stem j o i n e r [voice] is present, it associates to the noun stem-in i t ia l f r icat ives . Th i s is exemp l i f i ed in (11). 62 (11) Cont inuant v o i c i n g independent noun stem possessed noun k d es led e I [voice] T o account fo r the non-alternating f r icat ives , i.e. stem-init ia l f r icat ives , w h i c h are vo i c ed in the independent noun stem and in the possessive const ruc t ion , I assume, f o l l o w i n g R i c e (1989) , this c lass o f f r icat ives are spec i f i ed under l y ing l y fo r the laryngeal spec i f i ca t ion [voice] . Th i s is exemp l i f i ed in (12). (12) F r ica t ives U n d e r l y i n g l y Spec i f i ed fo r the Feature [voice] independent noun stem possessed noun 6a6 E S 6a9 e I I [voice] [voice] 5.2.3 Summary In this sect ion, the morphophonemi c process o f cont inuant v o i c i n g has been examined . E x a m i n a t i o n o f this process has shown that the laryngeal feature [voice] is p h o n o l o g i c a l l y act ive w i th i n the f r icat ive c lass. Th i s supports the c l a ims o f the La r yngea l M e c h a n i s m Hypothes i s . 5.3 Word-F ina l D e v o i c i n g The process o f word-f ina l d e v o i c i n g is present in some A thapaskan languages (e.g. K o y u k o n ) . The process o f word-f ina l d e v o i c i n g affects noun stem-final f r icat ives. W h e n the process o f word-f ina l devo i c i ng is present, vo i c ed f r icat ives surface as vo ice less f r icat ives . The consonants affected by the process o f word-f ina l d e v o i c i n g , i.e. the vo i c ed f r i cat ives , are ident i f ied in the Tah l tan consonant inventory in (13). The consonants, w h i c h surface as a result o f word-f ina l devo i c i ng , i.e. the vo ice less f r i cat ives , are a lso ident i f ied in the consonant inventory in (13). 63 (13) Consonants Affected by Word-final Devoicing labial coronal dorsal laryngeal labial alveolar inter-dental alveolar lateral palatal velar labio-velar uvular glottal a. b b. t te ts ti ts k k w q ? c. th t 9 h tsh ti*1 ts h k h k w h q h d. f t 9 ' ts' t i ' ts' k' k w ' q' e. 9 s i s ( ~ 9 ) X xw X h f. 6 z 1 z (~ y) Y Y w K g- m n w a. voiced stop e. voiceless fricative b. voiceless unaspirated stop f. voiced fricative c. voiceless aspirated stop g. sonorant d. glottalized stop 5.3.1 Data The process of word-final devoicing is exemplified in (14). Recall that the possessive morpheme consists of a prefix and the possessive suffix, [e]. When a noun stem is followed by the possessive suffix, [e], stem-final fricatives are voiced. When a noun stem is independent, stem-final fricatives are voiceless. It is crucial to note that the stem-final fricatives in the possessive construction are not word-final. In comparison, the stem-final fricatives in the independent noun stem are word-final. (14) Word-final Devoicing independent noun stem first person singular third person singular English gloss mi:i es mid e me mil e snare ke:ti'ui es ke.-ti'ul e me ke:ti'ul e shoelaces t i 'ui es ti 'ul e me ti 'ul e rope be:s es be:z e me be:z e knife tiic es tiey e me tiey e grease There is a second class of fricatives, which are not affected by the process of word-final devoicing. The independent noun stem and the possessed noun, both have voiceless stem-final fricatives. This is exemplified in (15). 64 (15) Stem-final V o i c e l e s s Fr ica t ives independent noun stem first person s ingular E n g l i s h g loss teG eG teG e cane tAs es tAs e arrow ts 'as e s t s ' a s e set line sei es se i e gaf kux es kux e rice dus es dus e cat ge0 eG ge9 e king salmon t'es es t'es e charcoal 5.3.2 A n a l y s i s In this sect ion, prev ious ana lys is o f word-f ina l d e v o i c i n g w i l l be presented, f o l l o w e d by the present analys is o f cont inuant v o i c i n g . In the Tah l tan l iterature, the process o f word-f ina l devo i c i ng has been descr ibed as a process o f v o i c i n g (Nater 1989). U n d e r the v o i c i n g hypothes is , stem-final f r icat ives , w h i c h are vo ice less in the independent noun stem become v o i c e d in the possessive const ruc t ion . H o w e v e r there is data w i t h non-alternating f r icat ives (see (15)), i.e. the data w i th stem-final vo ice less f r i cat ives , w h i c h are vo ice less in the independent noun stem and in the possessed noun . These fo rms are prob lemat i c for the v o i c i n g hypothes is . In other words , the v o i c i n g hypothes is cannot account fo r the fact f r icat ives remain vo ice less in the possess ive const ruct ion . Thus I assume the devo i c i ng hypothes is . S im i l a r to stem-init ia l f r icat ives , I hypothes ize that the stem-final f r icat ives in fo rms l ike those in (15) are unspec i f ied fo r the laryngeal spec i f i ca t ion [voice] . Re ca l l that under the La r yngea l M e c h a n i s m Hypothes i s , f r icat ives unspec i f i ed fo r laryngeal spec i f i ca t ion are vo ice less . T h i s therefore represents the unmarked case. Re ca l l that there were a select group o f stem-init ia l f r icat ives , w h i c h were under l y ing l y spec i f i ed for the laryngeal spec i f i ca t ion [voice] . I hypothes ize that, s im i l a r to stem-init ia l f r i cat ives , a select group o f stem-final f r icat ives represented by the data in (14) are spec i f i ed fo r the feature [voice] . I hypothes ize that this c lass o f f r icat ives part ic ipate in the process o f word-final devo i c i ng . In other words , stem-final f r icat ives , w h i c h are under l y ing l y unspec i f i ed fo r the laryngeal feature [voice] , lose the laryngeal spec i f i ca t ion [voice] in word-f ina l pos i t ion . T h i s is exemp l i f i ed in (16). N o t i c e that in the possess ive const ruc t ion , the stem-final f r i cat ive is not word- f ina l ; thus the laryngeal spec i f i ca t ion [voice] is not lost. In the noun stem, the stem-final f r i cat ive is word- f ina l ; thus the laryngeal [voice] is lost. 65 (16) Word-f ina l D e v o i c i n g possessed noun independent noun stem es mil e mii I [voice] 5.3.3 Stem-Fina l Stop Consonants In the Tah l tan l iterature, noun stem-final stop consonants are descr ibed as vo ice less when word-f ina l and vo i c ed when f o l l o w e d by the possess ive su f f i x [e] ( H a r d w i c k 1984; Na te r 1989). Re ca l l that Na te r (1989) hypothes izes that stem-final f r icat ives part ic ipate in a process o f v o i c i n g , not a process o f devo i c i ng . Na te r (1989) hypothes izes that, in add i t ion to f r icat ive consonants , stop consonants part ic ipate the process o f v o i c i n g , i.e. stem-final stops, w h i c h are vo ice less in the independent noun stem become v o i c e d in the possess ive const ruc t ion . A s p rev ious l y ment ioned , I assume a d e v o i c i n g hypothes is , not a v o i c i n g hypothes is . Second , I hypothes ize that stem-final stop consonants do not part icipate in the process o f word-f ina l devo i c i ng . In other words , I hypothes ize that stem-final stop consonants are phono log i c a l l y vo ice less in word-f ina l pos i t ion and in the possessive construct ion. Instead, I hypothes ize that stem-final stops are phonet i ca l l y , not phono log i ca l l y , v o i c e d in stem-final pos i t ion o f the possess ive const ruct ion . (17) S tem-Fina l Stop Consonants independent noun stem first person s ingular E n g l i s h g loss dustenret es dus tenred e bottle ca:kina?it es ca:kina?id e rain jacket cire : t es c i r e : d e pillow tG'Axdake:t e0 t0'Axdake:d e pipe tG 'a t i e0 tO 'adl e moss/diaper kek ' i te es kek ' idd e box de9d l in tsets es deOdlin tsedz e green wood t 'o:k es t'o:g e wart tG'ik e0 tG 'ig e tobacco k 'uk es kug e trunk Stops in stem-final pos i t ion i n the possess ive construct ion surface in in tervoca l i c pos i tons and, as was seen in chapter t w o are subject to a phonet ic process o f part ia l v o i c i n g . 66 5.3.4 Summary In th is sect ion, the process o f word-f ina l d e v o i c i n g has been examined . Exam ina t i on o f this process has shown that the laryngeal feature [voice] is phono log i c a l l y act ive w i t h i n the f r icat ive c lass. T h i s supports the c l a ims o f the L a r yngea l M e c h a n i s m Hypothes i s . 5.4 C o n c l u s i o n T w o processes have been examined in th is chapter, cont inuant v o i c i n g and word-f ina l devo i c i ng . W h e n the process o f cont inuant v o i c i n g is present, noun stem-init ia l vo ice less f r icat ives surface as vo i c ed f r icat ives. W h e n the process o f word-f ina l devo i c i ng is present, noun stem-final vo i c ed f r icat ives surface as vo ice less f r icat ives . E x a m i n a t i o n o f both processes supports the c l a ims o f the La r yngea l M e c h a n i s m Hypo thes i s . Spec i f i c a l l y , the processes support the hypothes is that the laryngeal feature [voice] is phono log i c a l l y act ive w i t h i n the f r icat ive c lass. 67 Chapter Six Morphophonemic Processes in Optimality Theory Thus far, the morphophonemic effects of the classifier prefixes (chapter four), the morphophonemic processes of continuant voicing and the process of word-final devoicing (chapter five) have been examined. In this chapter, these processes will be accounted for in Optimality Theory (McCarthy and Prince 1993). This chapter is divided into five sections. In section 6.1, the theoretical assumptions are discussed. In section 6.2, the morphophonemic effects of the classifier prefixes, the process of continuant voicing and the process word-final devoicing are summarized. In section 6.3, the constraints and constraint ranking are presented. In section 6.4, the morphophonemic processes are accounted for using Optimality Theory. Finally, in section 6.5 concluding remarks are made. 6.1 Theoretical Assumptions Recall that this thesis assumes the laryngeal specification of the Laryngeal Mechanism Hypothesis. The Laryngeal Mechanism Hypothesis is restated in (1). (1) Laryngeal Mechanism Hypothesis (revised from Rice 1994) b voiceless voiceless glottalized voiceless voiced glottal h aspirated unaspirated stops fricatives fricatives stop stops stops stop X X X X voice X X SG X X C G X X In this thesis I assume, following Clements (1985) and Sagey (1986), phonological representations are hierarchically structured. In (2), the hierarchical representations under the Laryngeal Mechanism Hypothesis are laid out. Note that all segments with oral/pharyngeal cavity constriction are also assumed to have Place specification (Labial, Coronal, Dorsal, Pharyngeal); as the present focus is the laryngeal specification of the major classes, Place is not included in these diagrams. 68 (2) Dual Mechanism Hypothesis Structural Representations [b] I Root / \ [stop] Laryngeal I [voice] Voiceless aspirated stop Root / \ [stop] Laryngeal I [SG] Voiceless fricatives Root Glottal stop I Root I Laryngeal I [CG] Voiceless unaspirated stop Root / \ [stop] Laryngeal Glottalized stop Root / \ [stop] Laryngeal [CG] Voiced fricatives Root I Laryngeal I [voice] [h] I Root I Laryngeal I [SG] 6.2 Summary of Morphophonemic Processes In this section, the morphophonemic effects of the classifier prefixes, the process of continuant voicing and the process word-final devoicing are summarized. The morphophonemic effects of the classifier prefixes are summarized in (3). Recall that there are four classifier prefixes: h-, 1-, d- and 0 - . The classifier prefixes immediately precede the verb stem and follow the subject prefixes. 69 (3) Summary o f the M o r p h o p h o n e m i c E f fec ts o f the C l a ss i f i e r P re f ixes analys is target output 1- - f loa t ing feature [voice] - verb stem-init ia l f r i cat ives - vo i c ed f r icat ives d- - f l oa t ing feature [stop] - verb stem-init ia l f r icat ives - verb stem-init ia l g lotta l stop - vo ice less unaspirated stops - g lo t ta l ized co rona l stop [ f ] h- - a f loa t ing feature [SG] - second person s ingular subject morpheme [n] -fo] - in the th i rd person s ingular and p lura l the h- c lass i f i e r surfaces as an independent segment [h] 0 - - v o i c i n g ass im i l a t i on ; the feature [voice] spreads f r o m a subject morpheme - verb stem-init ia l f r i cat ives - vo i c ed f r icat ives In (4), the morphophonemic process o f cont inuant v o i c i n g and the process o f word-f ina l d e v o i c i n g are summar ized . (4) S u m m a r y o f Cont inuant V o i c i n g and Word- f ina l D e v o i c i n g process analys is target output cont inuant v o i c i n g - a f loa t ing feature [voice] is a stem j o i n e r - noun stem-init ia l f r icat ives - vo i c ed f r i ca t ive word-f ina l d e v o i c i n g - a parsed feature [voice] de l inks - noun stem-final f r icat ives in word-f ina l pos i t ion - vo ice less f r i ca t ive 6.3 Constra ints and Const ra in t R a n k i n g Op t ima l i t y Theo ry ( M c C a r t h y and P r ince 1993) is a non-der ivat ional theory. W i t h i n Op t ima l i t y Theo ry a set o f candidates are evaluated acco rd ing to series o f ranked and v io l ab l e constraints. The op t ima l candidate, i.e. the output f o r m , is the candidate, w h i c h best satisf ies or m i n i m a l l y v io lates the g rammar ' s constraint h ierarchy. In th is ana lys is o f Tah l tan , the f o l l o w i n g constraints are c ruc i a l : G r o u n d e d Cond i t i ons : G r o u n d e d cond i t ions (A r change l i and Pu l l e yb l ank 1994) require that certain features co-occur and that others do not. The G r o u n d e d Cond i t i ons hypothes is states that feature c o -occurrence i nvoked in natural language must be phonet i ca l l y mot iva ted . A c c o r d i n g to the G r o u n d e d Cond i t i ons hypothes is , feature co-occurrence w h i c h is not phonet i ca l l y mot iva ted may not be invoked in natural language. In this analys is o f Tah l tan , the G r o u n d e d Cond i t i ons , w h i c h are c ruc i a l , f o l l o w f rom the second c l a i m . The c ruc i a l G r o u n d e d Cond i t i ons are as f o l l o w s : 70 * [ C G ] / [ V O I C E ] : A segment spec i f i ed fo r the feature [ C G ] cannot be spec i f i ed for the feature [voice] . U n d e r the La r yngea l M e c h a n i s m Hypothes i s , g lotta l stop and g lot ta l ized stop are spec i f i ed fo r the feature [ C G ] . D u r i n g the p roduc t ion o f glottal stop and g lo t ta l ized stops the voca l fo lds are pressed together t ight ly . S i m i l a r l y , du r ing the p roduc t ion o f a v o i c e d stop the voca l fo lds are brought together. U n l i k e glotta l stop and g lot ta l ized stops, du r ing the product ion o f a vo i c ed consonants, the voca l fo lds are not brought together t igh t l y ; instead they are brought together loose ly , so that air is able to pass between them. V o c a l fo lds cannot s imul taneous ly be in the pos i t ion for g lotta l stop/glottal ized stops and v o i c e d sounds, i.e. there is no glotta l stop or g lo t ta l ized stop, w h i c h ise vo i c ed ( Lade foged and M a d d i e s o n 1996). Th i s fact is expressed phono log i c a l l y w i t h the grounded cond i t i on * [CG]/ [vo ice ] . * [ S G ] / [ V O I C E ] : A feature spec i f i ed for the feature [SG] cannot be spec i f i ed for the feature [voice] . The consonant [h] and aspirated stops are spec i f i ed for the feature [ SG ] . D u r i n g the p roduc t ion o f [h] and aspirated stops the voca l fo lds are pu l l ed apart, a l l o w i n g a i r to pass through the glott is . In compar i son , dur ing the p roduc t ion o f a vo i c ed segment the voca l fo lds are brought together loose ly . V o c a l fo lds , w h i c h are s imul taneous ly in the pos i t ion for [h]/ aspirated stops and vo i c ed consonants are not poss ib le . T h i s fact is expressed phono log i c a l l y w i t h the grounded cond i t i on * [SG]/[voice] . * [ S T O P ] / [ D O R S A L ] / [ V O I C E ] : A segment cannot be spec i f i ed for the features [stop], [dorsal] and [voice] . * [ S T O P ] / [ C O R O N A L ] / [ V O I C E ] : A segment cannot be spec i f i ed for the features [stop], [coronal] and [voice] . * [ S T O P ] / [ L A B I A L ] / [ V O I C E ] : A segment cannot be spec i f i ed fo r the features [stop], [ labial] and [voice] . 71 These three grounded cond i t ions f o l l o w f r o m phonet ic ev idence presented i n chapter two . R e c a l l that the b i l ab ia l stop consonant [b] was v o i c e d i n a l l the word-internal env i ronments invest igated: post-sibi lant and preced ing a v o w e l , post-nasal and preced ing a v o w e l , and in te rvoca l i ca l l y . In compar i son , [d] and [g] were vo ice less unaspirated in post-sibi lant pos i t i on . O n l y when surrounded by vo i c ed segments were [d] and [g] f u l l y v o i c e d or par t ia l l y vo i c ed . R e c a l l that these acoust ic results were not surpr is ing . C o m p a r e d to [d] and [g], [b] is the easiest stop consonant to vo i ce . D u r i n g the product ion o f [b] there is a large space above the glott is . F o r a re la t ive ly l ong per iod o f t ime a i r can f l o w f r o m the lungs, through the glott is and into this re la t ive ly large space. A s l ong as the a i r is f l o w i n g the voca l fo lds are kept v ib ra t ing . U n l i k e [b], the space above the glott is du r ing the p roduc t ion o f [d] and [g] is quite sma l l . Because the cav i t y is so sma l l , v o i c i n g can on l y be mainta ined fo r a short per iod o f t ime (Oha l a and R i o d a n 1979; K i n g s t o n 1996). These phonet ic facts are expressed phono log i c a l l y w i t h the f o l l o w i n g g rounded cond i t ions : *[stop]/[dorsal]/[voice], *[stop]/[coronal]/[voice] and *[stop]/[labial]/[voice]. In add i t ion to these constraints, the f o l l o w i n g constraint h ierarchy f o l l o w s f r o m these phonet ic facts: (5) Const ra in t H i e ra r chy *[stop]/[dorsal]/[voice] » *[stop]/[coronal]/[voice] » *[stop]/[labial]/[voice] Th i s constraint h ierarchy is consistent w i t h the phonet ic fact that dorsa l stops are the least l i k e l y to be vo i c ed , w h i l e lab ia l stops are the most l i k e l y to be vo i c ed . Fa i th fu lness Constra ints : Fa i thfu lness constraints require ident i ty between an input f o r m and an output f o r m . In this ana lys is , the f o l l o w i n g Fa i th fu lness constraints are c ruc i a l : M A X IO: A n y segment in the input must have a correspondent in the output. T h i s M a x constraint prevents the delet ion o f any segment, ensur ing that the output is as fa i th fu l as poss ib le to the input. M A X IO P A T H : A n y path between a segment and any feature in the input must have a correspondent path i n the output. D E P IO P A T H : A n y path between a segment and any feature in the output must have a correspondent path in the input. 72 These M a x and D e p constraints prevent the de let ion ( M a x ) or insert ion (Dep ) o f l i nks to any features, ensur ing that the output is as fa i th fu l as poss ib le to the input. A l i g n m e n t Constra ints : In this analys is o f Tah l tan , Featura l A l i g n m e n t constraints ( A k i n l a b i 1996) are c ruc i a l . Featural A l i g n m e n t constraints require that the edge o f a g rammat i ca l category be a l igned w i th an edge o f a p rosod ic feature. The Featural A l i g n m e n t constraints required in th is ana lys is are as f o l l o w s : A L I G N ( [ V O I C E ] , L E F T , V E R B S T E M , L E F T ) : A l i g n the left edge o f the feature [voice] w i th the left edge o f the stem. Reca l l that in this thesis, the 1- c lass i f i e r is ana l yzed as a f loa t ing feature [vo ice] , w h i c h associates to verb stem-init ia l f r icat ives and glottal stop. T h i s A l i g n m e n t constraint ensures that when the feature [vo ice] , i.e. the 1- c lass i f ier , is parsed, it a l igns on l y w i t h the left edge o f the verb stem, i.e. the stem-init ia l consonant. A L I G N ( [ S T O P ] F E A T U R E , L E F T , V E R B S T E M , L E F T ) : A l i g n the left edge o f the feature [stop] w i t h the left edge o f the stem. Reca l l that in this thesis, the d- c lass i f i e r is ana lyzed as a f loa t ing feature [stop], w h i c h associates to verb stem-init ial f r icat ives and glotta l stop. T h i s A l i g n m e n t constraint ensures that when the feature [stop], i.e. the d- c lass i f ie r , is parsed, it a l igns on l y w i t h the left edge o f the verb stem, i.e. the stem-init ia l consonant. A L I G N ( [ SG ] , R I G H T , V E R B S T E M , L E F T ) : A l i g n the r ight edge o f the feature [SG] w i t h the left edge o f the stem. Reca l l that in this thesis, the h- c lass i f i e r is ana lyzed as a f loa t ing feature [ SG ] . U n l i k e the 1-, d-and 0 - , w h i c h affect verb stem-init ial consonants, the h- c lass i f i e r affects the p reced ing subject morpheme, and does not affect verb stem-init ia l consonants. T h i s a l ignment constraint ensures that when the feature [ SG ] , i.e. h- c lass i f ie r , is parsed, the r ight edge o f the feature [ SG ] is a l igned 73 w i t h the left edge o f the verb stem. In other words , the h- c lass i f i e r is associated w i t h the subject morpheme, w h i c h precedes the verb stem, not w i t h the verb-stem in i t i a l consonant. A L I G N ( [ V O I C E ] , L E F T , N O U N S T E M , L E F T ) : A l i g n the left edge o f the feature [voice] w i t h the left edge o f the stem. Reca l l that in this thesis, to account for the process o f cont inuant v o i c i n g , I assume a stem jo ine r , w h i c h consists o f the feature [voice] , associates to noun stem-init ia l f r icat ives. T h i s a l ignment constraint ensures that when the feature [vo ice] , i.e. stem j o i ne r , is parsed, it a l igns o n l y w i th the left edge o f the noun stem, i.e. the noun stem-init ia l consonant . P A R S E [ V O I C E ] : The feature [voice] must associate to a segment. P A R S E [ S T O P ] : The feature [stop] must associate to a segment. P A R S E [ SG ] : The feature [ SG ] must associate to a segment. T h e Parse constraints ensure that when the features [vo ice] , [stop] and [SG] are present, they l i nk to a segment. Sy l l ab le Structure Constr iant : * [-Sonorant] I [ L a r y n g e a l l j j : A segment at the r ight edge o f a sy l lab le cannot be spec i f i ed for the features [-sonorant] and [voice] . A b b r e v i a t i o n : *[-s] [v] R e c a l l that f r icat ives and stops are vo ice less in word-f ina l pos i t ion . Th i s constraint ensures that f r icat ive and stop consonants are not spec i f i ed fo r the feature [voice] in word-f ina l pos i t ion . In th is analys is o f Tah l tan , it is c ruc i a l that the constraints ( ident i f ied above) are ranked as f o l l o w s : 1 1 These constraint rankings will be motivated in the following section. 74 (6) Const ra in t R a n k i n g * [CG]/[voice] : * [SG]/[voice] : * complex * [-son] : * [stop]/ *[stop]/ : Max : Align ([stop], ! ! coda/onset 1 : [voice]/ [voice]/ i IO i left, verb [larla j [dorsal] [coronal] : : stem, left) Align ([voice], j Align ([SG], | Parse ; Parse Max IO ; Parse Dep IO ; *[stop]/ left, verb stem, • right, verb | [stop] j [SG] Path ; [voice] Path 1 [voice]/ left) ; stem, left) i [labial] 6.4 M o r p h o p h o n e m i c Processes in Op t ima l i t y Theo ry T h i s sect ion is d i v i ded in f i ve parts. In sect ions 6.4.1 through 6.4.3, the morphophonemi c effects o f the c lass i f ie r pref ixes are accounted fo r us ing O p t i m a l i t y Theory . In sect ion 6.4.4 and 6.4.5 respect ive ly , the processes o f cont inuant v o i c i n g and word-f ina l d e v o i c i n g are accounted fo r us ing Op t ima l i t y Theory . 6.4.1 1-classif ier In this thesis, the 1- c lass i f ie r is ana l yzed as a f loa t ing feature [vo ice] , w h i c h associates to verb stem-init ia l f r icat ives . T o account fo r the morphophonemi c effects o f the 1- c lass i f i e r in Op t ima l i t y Theo ry the f o l l o w i n g constraints and constraint r ank ing are c ruc i a l : A L I G N ( [ V O I C E ] , L E F T , P A R S E » D E P IO V E R B S T E M , L E F T ) ; [ V O I C E ] P A T H Tab leau ( 1 ) demonstrates h o w these constraints account fo r the m o r p h o p h o n m i c effects o f the 1-c lass i f ier . Tab leau ( 1 ) M o r p h o p h o n e m i c s o f the 1- c lass i f i e r in Op t ima l i t y T h e o r y 2 kasfzei] I hollered ka s [sei] [v] align (verb stem, left, [voice], left) Parse [voice] Dep IO Path a. ka s [^ei] [v] **! b. ka s [sei] [v] *! c. ka s [sei] [v/ *! ** d. ka s [zei] —> t * 2 In all tableaux, only the verb stem, which is presented in square brackets, i.e. [], is underlyingly specified for feature specification. 75 In candidate (a), the feature [voice] is parsed to the verb stem-init ia l f r i cat ive , [z] and to the verb stem-internal v o w e l , [e]; thus D e p IO Path is v io la ted tw i ce . In candidate (b), the feature [voice] is not parsed; thus the Parse [voice] constraint is v io la ted and this candidate is e l im ina ted . In candidate (c), the feature [voice] is parsed to the verb stem-internal v o w e l , [e]; thus Dep IO Path is v io la ted once. T h e A l i g n ( [vo ice] , left, verb stem, left) constraint is a lso v io la ted by candidate (c) because the feature [voice] is a l igned w i t h the left edge o f the verb stem-internal v o w e l , [e], not w i t h the verb stem-init ia l consonant . In the op t ima l candidate (d), the feature [voice] is a l igned w i t h the verb-stem in i t ia l f r i ca t i ve ; thus D e p IO Path is v io la ted once. The op t ima l candidate satisfies a l l other constraints. C o m p a r i s o n o f candidate (b) w i t h the op t ima l candidate shows that it is c ruc ia l to rank D e p I O Path b e l o w Parse [vo ice] . 6.4.2 d- c lass i f i e r In th is thesis, the d- c lass i f i e r is ana lyzed as a f loa t ing feature [stop], w h i c h associates to verb stem-init ia l f r icat ives . Reca l l that this interact ion is referred to as the d-effect in the A thapaskan l iterature. The consonants, w h i c h surface, as a result o f the d-effect are vo ice less unaspirated stops. In order to account for the d-effect in Op t ima l i t y Theo r y the f o l l o w i n g constraints and constraint r ank ing are c ruc i a l : A L I G N ( [ S T O P ] , L E F T , P A R S E » D E P IO V E R B S T E M , L E F T ) ; [ S T O P ] P A T H Tab leau (2) demonstrates h o w these constraints account fo r the d-effect in O p t i m a l i t y Theory . Tab leau (2) D-effect kas i [dze i ] I tickled myself ka si [sei] [s] align (verb stem, left, [stop], left) Parse [stop] Dep 10 path a. kasi [dzedl] [sj""*~ * * l b. ka si [sei] [s] *! c. kadzi[dzei] *! ** d. kasi [cjzei] [s] -> * In candidate (a), the feature [stop] is a l igned to the verb stem-init ia l and verb stem-final consonant ; thus D e p IO Path is v io la ted tw i ce . D e p I O Path is the on l y constraint v io la ted by this candidate. In candidate (b), the feature [stop] is not parsed; thus the constraints Parse [stop] is 76 violated and this candidate is eliminated. In candidate (c), the feature [stop] is parsed to the verb stem-initial consonant and to the preceding subject morpheme; thus Dep IO Path is violated twice. In this candidate the feature [stop] is aligned to the left edge of the subject morpheme, not the left edge of the verb stem; thus the Align ([stop], left, verb stem, left) constraint is also violated. Similar to candidate (a), the optimal candidate (d), only violates Dep IO Path. Unlike candidate (a), the optimal candidate (d), only violates Dep IO Path once. Comparison of candidate (b), with the optimal candidate (d), shows crucial ranking between Parse [stop] and Dep IO Path. 6.4.3 h- classifier In this thesis, the h- classifier is analyzed as a floating feature [SG]. Recall that in this analysis of the h- classifier, verb stem-initial consonants are not affected by the h-classifier. Instead, the h- classifier interacts with the preceding subject morpheme. In the third person singular and third person plural, the h- classifier is realized phonetically as an independent segment, [h]. In the second person singular, the h- classifier associates to the associates to the second person singular subject morpheme, [n], which results in the voiceless nasal, [n]. In the first person singular and second person plural, I hypothesize that the h- classifier associates vacuously (i.e. shows not overt effect) to the preceding subject morpheme. In order to account for the morphophonemic effects of the h- classifier, the following constraints and constraint ranking are crucial: * C O M P L E X CODA, M A X A L I G N ([SG], PARSE » DEP IO C O M P L E X ONSET; IO; RIGHT, V E R B STEM, LEFT); [SG] PATH Tableaux (3) through (5) demonstrate how these constraints account for the morphophonemic effects of the h- classifier. Recall that in the third person singular, the h- classifier is realized phonetically as the independent segment, [h]. This is exemplified in the following tableau. Similarly, in the third person plural, the h- classifier is realized as an independent segment, [h]. An Optimality account of the third person plural is identical to tableau (3). 77 Tableau (3) Morphophonemic Effects of the h- classifier in Optimality Theory e h [0e] he's tanning O e [8e] [sg] * complex coda/onset MaxIO align (verb stem, left, [ S G ] , right) Parse [ S G ] Dep IO Path a. e [OB] [sg] *! b. e h[0e] [sg] —> * In candidate (a), the feature [SG] is not parsed; as a result Parse [SG] is violated. In the optimal candidate (b), the h- classifier is realized phonetically as an independent segment, [h]; thus Dep IO Path is violated once. This is the only constraint violated by this candidate. Comparison of candidate (a) with the optimal candidate shows that, once again, it is crucial to rank Parse [SG] above Dep IO Path. In the second person singular, the h- classifier associates to the second person singular subject morpheme, [n], which results in the voiceless nasal, [n]. This is exemplified in the following tableau. Tableau (4) Morphophonemic Effects of the h- classifier in Optimality Theory in [0e] you 're tanning O in [8E] [sg] * complex coda/onset MaxIO align (verb stem, left, [ S G ] , right) Parse [ S G ] Dep IO Path a. in h [9e] [sg] *! * b. ih[0e] [sg] *! * c. in [9e] [sg] *! d. in h [8e] [sg] *! ** e. in [9e] [sg] -> * In candidate (a), the h- classifier is phonetically realized, the subject morpheme is present and verb stem-initial consonant is also present; thus the *Complex Coda/Complex Onset constraint is violated. Dep IO Path is also violated once by this candidate. In candidate (b), the h- classifier is realized phonetically; thus Dep IO Path is violated once. In this candidate the subject morpheme is not present; thus the Max IO constraint is also violated. In candidate (c), the feature [SG] is not parsed; hence Parse [SG] is violated. This is the only constraint violated by this candidate. In candidate (d), the h- classifier is phonetically realized. The feature [SG] is also 78 parsed to the second person s ingular subject morpheme [n] ; therefore D e p IO Path is v io la ted tw ice . The * C o m p l e x Coda/Onset constraint is a lso v io la ted by this candidate because the subject morpheme, h- c lass i f i e r and verb stem-init ia l consonant are a l l present. In the op t ima l candidate (e), the feature [SG] is parsed to the second person s ingular subject morpheme , [n]; as a result the D e p 1 0 Path constraint is v io la ted . T h i s is the on l y constraint v io la ted by this candidate. C o m p a r i s o n o f candidate (c), w i t h the op t ima l candidate shows that it is c ruc i a l to rank Dep IO Path b e l o w Parse [ SG ] . In the f irst person s ingular , the h- c lass i f i e r associates vacuous l y to the subject morpheme. T h i s is exemp l i f i ed in tableau (5). S i m i l a r l y , in the second person p lu ra l , the h-c lass i f i e r associates vacuous l y to the preced ing subject morpheme. A n Op t ima l i t y account o f the second person p lura l is ident ica l to tableau (5). Tab leau (5) M o r p h o p h o n e m i c E f fects o f the h- c lass i f i e r in Op t ima l i t y Theo ry e0 [0e] I'm tanning O e e [9e] [sg] * complex coda/onset Max IO align (verb stem, left, [SG], right) Parse [SG] Dep IO Path a. e 9 h [9e] [sg] *! * b. 8 h, [9e] [sg] *! C. 8 9 [08] [sg] *! d. e 9 [9e] [sg'] *! e. e ^  [9e] [sg] —> * In candidate (a), the h- c lass i f ie r is rea l ized phonet ica l l y . In this candidate, the subject morpheme, the h- c lass i f i e r and the verb stem-init ia l consonant are present; thus the " "Complex Coda/Onset constraint is v io la ted . D e p IO Path is a lso v io la ted once by this candidate. In candidate (b), the h- c lass i f ie r is phonet i ca l l y rea l i zed . In this candidate, the subject morpheme is not present; as a result M a x 10 is v io la ted . T h i s is the on l y constraint v io la ted by this candidate. In candidate (c), the feature [SG] is not parsed; thus the Parse [SG] constraint is v io la ted . In candidate (d), the feature [SG] is parsed to the verb stem-init ia l f r i ca t i ve . T h i s is a v io l a t i on o f the A l i g n ( [ SG ] , r ight , verb stem, left) constraint. D e p 10 Path is also v io la ted once by this candidate. In the op t ima l candidate (e), the feature [SG] is parsed vacuous ly to the subject morpheme, [0]; thus D e p IO Path is v io la ted once. T h i s is the on l y constraint v io la ted by this 79 candidate. C o m p a r i s o n o f candidate (c), w i t h the op t ima l candidate (e), shows c ruc i a l r ank ing between Parse [SG] and D e p IO Path . 6.4.4 Cont inuant V o i c i n g R e c a l l that the process o f cont inuant v o i c i n g occurs w h e n the noun stem is preceded by a possess ive pre f ix . W h e n the process o f cont inuant v o i c i n g is present, vo ice less f r icat ives surface as vo i c ed f r icat ives. T o account for the process o f cont inuant v o i c i n g , I assume a stem j o i ne r , w h i c h consists o f the feature [voice] , is present i n the possess ive const ruct ion . W h e n present, the stem jo ine r , i.e. feature [voice] , associates to noun stem-init ia l f r icat ives . T o account fo r the morphophonemi c process o f cont inuant v o i c i n g i n O p t i m a l i t y Theo ry the f o l l o w i n g constraints and constraint r ank ing are c ruc i a l : A L I G N ( [ V O I C E ] , L E F T , P A R S E » D E P I O N O U N S T E M , L E F T ) ; [ V O I C E ] P A T H Tab leau (6) demonstrates h o w these constraints can account fo r the process o f cont inuant v o i c i n g . Tab leau (6) Cont inuant V o i c i n g i n Op t ima l i t y Theo r y es [led] e my smoke es [ieijl] e [s] [v] align (noun stem, left, [voice], left) Parse [voice] Dep IO Path a. es [iecj] e |f[s] [v] *! b. es [leij] e [s] [v] c. es [led] e / [s] [v] *! * d. es [ledl e l i / [s] [v] —> * In candidate (a), the feature [voice] is not parsed; hence the constraint Parse [voice] is v io la ted . In candidate (b), the feature [voice] is parsed to the noun stem-init ia l f r i cat ive and to the noun stem-internal v o w e l , [e]; as a result D e p IO Path is v io la ted tw i ce . In candidate (c), the feature [voice] is parsed to the noun stem-internal v o w e l , [e]; therefore D e p IO Path is v io la ted once. In this candidate, the feature [voice] is not a l igned w i t h the left edge o f the noun stem; thus 80 the A l i g n ( [vo ice] , left, noun stem, left) constraint is a lso v io la ted . In the opt ima l candidate (d), the feature [voice] is parsed to the noun stem-init ia l f r i ca t i ve ; thus D e p IO Path is v io la ted once. T h e op t ima l candidate satisf ies a l l other constraints. C o m p a r i s o n o f candidate (a) w i t h the op t ima l candidate shows that it is c ruc ia l to rank Dep IO Path b e l o w Parse [voice] . R e ca l l that the process o f cont inuant v o i c i n g does not affect h-init ial noun stems. In order to account for th is , the f o l l o w i n g grounded cond i t i on is c ruc i a l : * [SG]/ [voice] . Tab l eaux (7) w i l l demonstrate h o w this constraint prevents the feature [voice] f r o m assoc ia t ing to h-initial noun stems. Tab leau (7) Cont inuant V o i c i n g in Op t ima l i t y Theo ry es [hodzih] e my caribou es hodzihe [sg] [v] * [SG] Parse [voice] Dep IO Path a. es [hodzih]e / f s g ] [v] *! * b. es [hodzih]e [sg] [v] -> * In candidate (a), the feature [voice] is parsed to the noun stem-ini t ia l , [h]. U n d e r l y i n g l y [h] is spec i f ied fo r the feature [ SG ] ; thus the *[SG]/[voice] constraint is v io la ted . D e p IO Path is a lso v io la ted once by this candidate. In the op t ima l candidate (b), the feature [voice] is not parsed; thus Parse [voice] is v io la ted . T h i s is the o n l y constraint v io la ted by the opt ima l candidate. C o m p a r i s o n o f candidate (a) w i t h the op t ima l candidate shows that it is c ruc ia l to rank *[SG]/[voice] above Parse [voice] . Reca l l that the process o f cont inuant v o i c i n g does not affect noun stem-init ia l aspirated stops, vo ice less unaspirated stops, g lo t ta l ized stops and glottal s top. 3 In order to account for th is , the f o l l o w i n g constraints are c ruc i a l : * [ C G ] / * [ SG ]/ * [ S T O P ] / * [ S T O P ] / » P A R S E [ V O I C E ] ; [ V O I C E ] ; [ D O R S A L ] / [ C O R O N A L ] / [ V O I C E ] [ V O I C E ] ; [ V O I C E ] ; Tab l eaux (8) through (10) w i l l demonstrate h o w these constraints prevent the feature [voice] f r o m assoc ia t ing to noun stem-init ia l stop consonants. 3 Recall that underlyingly, the labial stop, [b] is specified for the feature [voice]. I hypothesize that the feature [voice] associates vacuously to the stop consonant [b]. See tableau (11) for Optimality analysis. 81 In tableau (8), the feature [voice] does not affect the noun stem-initial aspirated stop consonant. Tableau (8) Continuant Voicing in Optimality Theory E S [tu:] e my water es [tu:] e [si[sg] M * [CG]/[voice] * [SG]/[voice] * [stop]/[voice]/ [dorsal] * [stop]/[voice]/ [coronal] Parse [voice] a. es [du:] e / [ s f t s g ] [v] *! b. es [tu:] e [sftsg] [v] —> * In candidate (a), the feature [voice] is parsed to the noun stem-initial stop, [t]. Underlyingly [t] is specified for the features [stop] and [SG]; thus the *[SG]/[voice] constraint is violated. Dep IO Path is also violated once by this candidate. In the optimal candidate (b), the feature [voice] is not parsed; thus Parse [voice] is violated. This is the only constraint violated by the optimal candidate. Comparison of candidate (a) with the optimal candidate shows that it is crucial to rank *[SG]/[voice] above Parse [voice]. In tableau (9), the feature [voice] does not affect the noun stem-initial glottalized stop consonant. Tableau (9) Continuant Voicing in Optimality Theory E S [k'ug] E my paper es [k'ug] e [sTfcg] M * [CG]/[voice] * [SG]/[voice] * [stop]/[voice]/ [dorsal] * [stop]/[voice]/ [coronal] Parse [voice] a. es [cfug] e / [sHcg] [v] *! b. es [k'ug] e [sftcg] [v] -> * In candidate (a), the feature [voice] is parsed to the noun stem-initial stop, [k']. Underlyingly [k'] is specified for the features [stop] and [CG]; thus the *[CG]/[voice] constraint is violated. Dep IO Path is also violated once by this candidate. In the optimal candidate (b), the feature [voice] is not parsed; thus Parse [voice] is violated. This is the only constraint violated by 82 the opt ima l candidate. C o m p a r i s o n o f candidate (a) w i t h the opt ima l candidate shows that it is c ruc ia l to rank * [CG]/ [vo ice ] above Parse [voice] . In tableau (10), the noun stem-init ia l consonant is the stop consonant [d]. Reca l l that the stop consonant , [d] is vo ice less unaspirated; under the La r yngea l M e c h a n i s m Hypothes i s vo ice less unaspirated stops are unspec i f i ed fo r la ryngea l features. T o ensure that the stem j o i ne r , i.e. feature [vo ice] , does not associate to the vo ice less unaspirated stop [d], the grounded cond i t i on *[stop]/[coronal/[voice] is c ruc ia l . T h i s is exemp l i f i ed in tableau (10). Tab leau (10) Cont inuant V o i c i n g in Op t ima l i t y Theo r y es [dih] e my grouse es [dih] e [s] [cor] Iv] * [CG]/[voice] * [SG]/[voice] * [stop]/[voice]/ [dorsal] * [stop]/[voice]/ [coronal] Parse [voice] a. es [dih] e / [sf [cor] [v] *! b. es [dih] e [sftcor] [v] -> * In candidate (a), the feature [voice] is parsed to the stem-init ia l stop consonant [d]. In add i t ion to the feature [voice] the stem-init ia l stop [d] is spec i f i ed fo r the feature [stop] and [coronal ] ; therefore the *[stop]/[voice]/[coronal] constraint is v io la ted . In the opt ima l candidate (b), the feature [voice] remains un-parsed; thus Parse [voice] is v io la ted . T h i s is the on l y constraint v io la ted by this candidate. C o m p a r i s o n o f candidate (a) w i t h the op t ima l candidate shows that it is c ruc ia l to rank Parse [voice] b e l o w *[stop]/[voice]/[coronal]. In tableau (11), the noun stem-init ial consonant is the b i l ab ia l stop consonant , [b]. Re ca l l that under the La ryngea l M e c h a n i s m Hypothes i s the b i l ab ia l stop consonant , [b] is spec i f i ed for the features [stop] and [voice] . Th i s laryngeal feature spec i f i ca t ion f o l l o w s f r o m phonet ic ev idence presented in chapter two . R e c a l l that the stop consonant , [b] was v o i c e d , w h i l e the stop consonants , [d] and [g] were hypothes ized to be vo ice less unaspirated. T o ensure that the feature [voice] is parsed to the stop consonant, [b] the f o l l o w i n g constraints and constraint r ank ing are c ruc i a l : P A R S E [ V O I C E ] ; M A X IO P A T H » * [ S T O P ] / [ L A B I A L ] / [ V O I C E ] V 83 Tab leau (11) Cont inuant V o i c i n g in Op t ima l i t y Theo r y es [be:s] e my knife es [be:z] e & [v] [v] . Max IO Path Parse [voice] * [stop]/[voice]/ [labial] a. es [be:z] e [sTM [v] *! * b. es [pe:z] e [s] [v] *! * b. es [be:z] e / [ s f i v ] [v] —> * In candidate (a), the feature [voice] is not parsed; thus Parse [voice] is v io la ted . T h e grounded cond i t ion *[stop]/[labial]/[voice] is also v io la ted by this candidate because the noun stem-init ia l stop consonant , [b] is spec i f i ed for a l l three o f these features. In candidate (b), the feature [vo ice] , w h i c h is associated to [b] in the input, is not present in the output; thus the M a x IO Path constraint is v io la ted . S ince the feature [voice] is not associated to the stop, [b], the *[stop]/[labial]/[voice] constraints is not v io la ted . In this candidate, the feature [voice] is not parsed in this candidate; thus the Parse [voice] constraint is v io la ted . In the op t ima l candidate (c), the feature [voice] is parsed vacuous l y to the verb stem-init ia l stop consonant [b]; thus the Parse [voice] constraint in not v io la ted . T w o constraints are v io la ted by the op t ima l candidate : D e p IO Path and *[stop]/[voice]/[labial]. C o m p a r i s o n o f candidate (a) w i t h the op t ima l candidate shows that it is c ruc ia l to rank D e p IO Path b e l o w M a x IO Path. C o m p a r i s o n o f candidate (b) w i t h the op t ima l candidate shows that it is c ruc ia l to rank D e p IO Path and * [stop]/[voice]/[labial] b e l o w M a x IO and Parse [voice] . 6.4.5 Word-F ina l D e v o i c i n g R e c a l l that, in this thesis, I hypothes ize t w o classes o f stem-final f r i ca t ives : one, w h i c h are unspec i f ied for laryngeal spec i f i ca t ion in word-f ina l pos i t i on , and a second c lass, w h i c h are spec i f i ed for the feature [voice] . Reca l l that the second class o f f r icat ives part ic ipated in the process word-f ina l devo i c i ng . Spec i f i c a l l y , I assume f r icat ives , w h i c h are unde r l y i ng l y spec i f i ed for the feature [voice] , lose this spec i f i ca t ion in word-f ina l pos i t ion . T o account fo r the process o f word-f ina l d e v o i c i n g in Op t ima l i t y Theo r y the f o l l o w i n g constraints are c ru c i a l : * [-Sonorant] [Laryngeal ] » M A X IO la ; P A T H 84 Tab leaux (12) and (13) demonstrate h o w the process o f word-f ina l d e v o i c i n g can be accounted for i n Op t ima l i t y Theory . Tab leau (12) Word-f ina l devo i c i ng in O p t i m a l i t y Theo r y [m i i ] snare [mil] [v] * [-sonorant] [laryngeal], a Max IO Path a. [mil] [ V ] *! b. [mil] —> * In candidate (a), feature [voice] remains parsed to the stem-final f r i ca t i ve , [1]; thus * [ vo i ce ] ] a constraint is v io la ted . The op t ima l candidate (b) v io lates the M a x IO Path constraint because the feature [voice] , parsed to the stem-final f r icat ive i n the input, is not present in the output. C o m p a r i s o n o f candidate (a) w i t h the op t ima l candidate shows that it is c ruc ia l to rank M a x IO Path b e l o w *[voice] ]o\ S i m i l a r to f r icat ives , stop consonants are vo ice less in word-f ina l pos i t i on . The f o l l o w i n g tableau demonstrates h o w this fact is accounted fo r in Op t ima l i t y theory. Tab leau (13) Word-f ina l devo i c i ng in O p t i m a l i t y Theo r y [ku:k] trunk [king] [s] [dor] * [-sonorant] [laryngeal] a Max IO Path a. [ku:g] [s] [d] [v] *! b. [ku:g] [s] [dor] —> In candidate (a), a feature [voice] has been added and is parsed to the word-f ina l stop consonant [g]; thus D e p IO Path is v io la ted . The * [ vo i ce ] ] a constraint is v io la ted by this candidate because the v o i c e d word-f ina l stop is in a c o d a pos i t ion . T h e op t ima l candidate satisf ies a l l constraints. 6.5 C o n c l u s i o n In this chapter, the morphophonemic effects o f the c lass i f i e r pref ixes , the process o f cont inuant v o i c i n g and the process o f word-f ina l d e v o i c i n g were accounted fo r w i t h i n Op t ima l i t y Theory . 85 Chapter Seven Conclusion This thesis has examined the phonetic and phonological properties of laryngeal distinctions in the consonant inventory of Tahltan. Specifically, the phonetic acoustic properties of plain stop consonants have been examined. Based on the phonetic findings, the Dual Mechanism Hypothesis (Rice 1994), which is a hypothesis that concerns the laryngeal specification of stop and fricative consonants, was revised. The revisions included the addition of the category voiced stop, and extended the specifications of laryngeal mechanisms to include [CG]. (1) Laryngeal Mechanism Hypothesis (revised from Rice 1994) voiced voiceless voiceless glottalized voiceless voiced glottal h stop aspirated stops unaspirated stops stops fricatives fricatives stop stop X X X X voice X X SG X X CG X X In addition to the phonetic evidence, in this thesis it has been shown that patterns of behavior governing syllable structure, the morphophonemic processes continuant voicing and word-final devoicing, and the morphophonemic effects of the classifier prefixes support the claims of the Laryngeal Mechanism Hypothesis. Specifically, it was shown that observed patterns of behavior governing syllable structure support the laryngeal specifications of the stop consonants, while the morphophonemic processes of continuant voicing and word-final devoicing and morphophonemic effects of the classifier prefixes support the laryngeal specifications of the fricative consonants, glottal stop and [h]. Finally, it has been demonstrated that assuming the laryngeal specifications of the Laryngeal Mechanism Hypothesis, the morphophonemic processes of continuant voicing and word-final devoicing, and the morphophonemic effects of the classifier prefixes could be accounted for with an Optimality Theoretic analysis (McCarthy and Prince 1993). 86 Bibliography Aldere te , John and T a n y a B o b . 1999. F ie ldnotes and Tapes o f Language Consu l tants at Te legraph C reek and Iskut. M s . , U n i v e r s i t y o f B r i t i sh C o l u m b i a . A k i n l a b i , A k i n b i y i . 1996. Featural a f f ixa t ion . Journal ofLinguistics 32 : 239-289. A k i n l a b i , A k i n b i y i . 1997. Ka l aha r i v o w e l ha rmony . The Linguistic Review 14: 97-138. A r c h a n g e l i , D i a n a and Doug l a s Pu l l e yb l ank . 1994. Grounded Phonology. C a m b r i d g e : M I T Press. Carter , C o l i n , Pa t r i ck C a r l i c k , E d i t h C a r l i c k , Susan Denn i s , A g n e s Hunter , M a b e l Denn i s and Fredd ie Quock . . 1994. Tahltan Children's Illustrated Dictionary. M s . , Tah l tan T r i ba l C o u n c i l . Carter , C o l i n . Pa t r i ck C a r l i c k , A n g e l a D e n n i s , R e g i n a L o u i e , Susie Tashoots and M y r a B l a c k b u r n , F redd ie Q u o c k and E d i t h C a r l i c k . 1994. Basic Tahltan Conversation Lessons. M s . , Tah l tan T r i b a l C o u n c i l . C l ements , G . N . 1985. The geometry o f phono log i ca l features. Phonology Yearbook 2: 225-252 . C o o k , E-D. 1984. Sarcee Grammar. V a n c o u v e r : Un i ve r s i t y o f B r i t i sh C o l u m b i a Press. G o r d o n , Ma t thew . 1993. The Phonetic Structures ofHupa. Un i ve r s i t y o f C a l i f o r n i a , L o s A n g e l e s W o r k i n g Papers. Ha rgus , Sharon. 1988. The Lexical Phonology of Sekani. N e w Y o r k : Ga r l and . H a r d w i c k , Margaret . 1984. Tahltan Morphology and Phonology. M A Thes i s , Un i ve r s i t y o f Toronto . H o w r e n , Robert . 1971. A fo rma l i za t ion o f the A thapaskan d-effect. International Journal of American Linguistics 39 : 96-114. K a r i , James. 1973. Navajo Verb Prefix Phonology. Doc to ra l d issertat ion, Un i ve r s i t y o f N e w M e x i c o . K i n g s t o n , John. 1996. Introduction to Phonetic Theory. M s . , Un i ve r s i t y o f Massachuset ts , Amhe r s t . K l a t t , D e n n i s H . 1975. V o i c e onset t ime, f r i ca t ion and aspirat ion in word- in i t ia l consonant c lusters. Journal of Speech and Hearing Research 18: 686-706. L ade foged , Peter. 1993. A Course in Phonetic. N e w Y o r k : Harcour t B race Jovan i ch . L ade foged , Peter and Ian M a d d i e s o n . 1996. The Sounds of the World's Languages. O x f o r d : B l a c k w e l l s . 87 Lamontagne , G r e g and K e r e n R i c e . 1994. An Optimality Theoretic Account of Athapaskan D-effects. M s . , Un i ve r s i t y o f Toronto . M a d d i e s o n , Ian. 1984a. Patterns of Sound. C a m b r i d g e : C a m b r i d g e Un i v e r s i t y Press. M c C a r t h y , John and A l a n P r ince . 1993. Gene ra l i zed a l ignment . In Gee t B o o i j & Jaap van M a r i e (eds), Yearbook of Morphology (pp. 79-153). Dordrecht : K l u w e r . M c C a r t h y , John and A l a n P r ince . 1995. Fa i th fu lness and redupl icat ive ident i ty . Papers in Op t ima l i t y Theo ry , University of Massachusetts Occasional Paper 18: 249-384. M c D o n o u g h , Joyce M . 1990. Topics in the Phonology and Morphology of the Navajo Verb. Doc to ra l d issertat ion, Un i ve r s i t y o f Massachuset ts . M c D o n o u g h , Joyce M and Peter L ade foged . 1993. Navajo Stops. Un i ve r s i t y o f C a l i f o r n i a , L o s Ange l e s W o r k i n g Papers. Nater , H a n k F. 1989. Some comments on the phono logy o f Tah l tan . International Journal of American Linguistics: 25-42. Oha l a , J.J and C . R io rdan . 1979. Pass ive voca l tract enlargement du r ing v o i c e d stops. In J . J . W o l f and D . H . K a t t (eds), Speech communication papers (pp.89-92). N e w Y o r k : A c o u s t i c a l Soc ie ty o f A m e r i c a . Pu l l e yb l ank , Doug l a s . 1996. Neu t ra l vowe l s in O p t i m a l i t y Theo ry : a compar i son o f Y o r u b a and W o l o f . Canadian Journal of Linguistics 4 1 : 295-347. Rando ja , T i n a . 1990. The Phonology and Morphology of Halfway River Beaver. Doc to r a l d issertat ion, Unv i e r s i t y o f Ot tawa. R i c e , K e r e n . 1988. Cont inuant v o i c i n g in S lave (Nor thern A thapaskan ) : the c y c l i c app l i ca t ion o f default rules. In M i c h a e l H a m m o n d & M i c h a e l N o o n a n (eds), Theoretical Morphology (pp. 371-388). N e w Y o r k : A c a d e m i c Press. R i c e , K e r e n . 1991. P rosod ic const i tuency in Hare (Athapaskan) : ev idence fo r the foot. Lingua 82, 201-245. R i c e , K e r e n . 1989b. A Grammar of Slave. B e r l i n : M o u t o n de Gruy ter . R i c e , K e r e n . 1994. La ryngea l features in A thapaskan languages. Phonology 1 1 : 107-147. C a m b r i d g e Un i ve r s i t y Press. Sagey, E l i zabe th . 1986. The Representation of Features and Relations in Nonlinear Phonology. Docto ra l dissertat ion, M I T . Shaw, Pa t r i c ia . 1981. F ie ldnotes and Tapes o f Language Consu l tants at Te legraph C reek . M s . , Un i ve r s i t y o f B r i t i sh C o l u m b i a . Shaw, Pat r i c ia . 1982. F ie ldnotes and Tapes o f Language Consu l tants at Te legraph Creek . M s . , Un i ve r s i t y o f B r i t i sh C o l u m b i a . 8 8 Shaw, Patr ic ia . 1983. F ie ldnotes and Tapes o f Language Consu l tants at Te legraph Creek . M s Un i ve r s i t y o f B r i t i sh C o l u m b i a . Shaw, Pat r ic ia . 1991. Consonant ha rmony systems: the spec ia l status o f corona l harmony . In E d Ca ro l e Paradis and Jean-Francois Prunet (eds), Phonetics and Phonology, Volume 2, The Special Status of Coronals: Internal and external evidence (pp. 125-157). San D i e g o : A c a d e m i c Press Inc. Stanley, R i c h a r d . 1969. Navajo Phonology. Doc to r a l d issertat ion, M I T . V a u x , Bert . 1998. T h e laryngeal spec i f i ca t ion o f f r icat ives . Linguistic Inquiry 29: 497-511. Appendix A In (1) through (3), the p re f ixed nouns , w h i c h were co l l ec ted f o r m Tah l tan speakers, and w h i c h were ana lyzed fo r the present phonet ic study, are presented. In (1), the p l a in stops, [p], [t] and [k] f o l l o w the s ib i lant [s], w h i c h is part o f the f i rst person possess ive morpheme [es]. (1) Post-Sibi lant P l a i n Stop Consonants es V (stem-init ia l ly ) es[p]eze my knife es[p]ade my mittens es[p]atk£ my moccasins es[p]ede my food es[t]ihe my grouse Es[t]e9tee5e my squirrel Es[t]egane my sockeye es[t]ebehe my sheep es[t]ustenvet my bottle Es[k]ahe my rabbit es[k]awe my drum Es[k]endam my horse Es[k]uhs my ant ES [k ]E0E my king salmon In (2), the p la in stop consonants [p], [t] and [k] f o l l o w the nasal [n], w h i c h is part o f the second person possessive morpheme [ in] . (2) Post-Nasal P l a i n Stop Consonants in V (stem-init ia l ly ) in[p]eze your knife in[p]ade your mittens in[p]atk£ you moccasins in[p]£de your food in[t]ihE your grouse in[t]e9t6E6e your squirrel in[t]£ganE your sockeye in[t]£b>£h£ your sheep in[t]ust£nvet your bottle in[k]ah£ your rabbit in[k]awe your drum in[k]£ndam your horse in[k]uhe your ant in[k]£9E your king salmon 90 In (3), the plain stops [p], [t] and [k] are intervocalic. In all examples, the plain stop consonants, follow the third person possessive morpheme [me] 'his' and precede a stem-internal vowel. (3) Intervocalic Plains Stop Consonants v_v me[p]eze his knife me[p]ade his mittens me[p]atke his moccasins me[p]ede his food me[t]ihe his grouse me[t]e9t9ede his squirrel me[t]egane his sockeye me[t]ebehe his sheep meftlustenyet his bottle me[kahe his rabbit me[k]awe his drum me[k]endame his horse me[k]uhe his ant me[k]e9e his king salmon 91 Figure (1) Appendix B Post-sibilant [p]; [esba:ke] 'my moccasins' Figure (3) Post-sibilant [k]; [esgendame] 'my horse' Figure (4) Post-nasal [p]; [inba:tke] 'your moccasins' ' ! ' " " « ' H . . . . | | I : , m -i-93 Figure (5) Post-nasal [k]; [ingawe] 'your drum' Figure (6) Intervocalic [p]; [mebede] 'his food' -32Q0---2300--1400--500 - 1 4 6 0 )\. [! l l 8 0 - 1 1 0 0 - 2 3 8 0 -, ; ' • • i m , , , , , mm 4 0 0 5 0 0 m s 94 Figure (7) Intervocalic [t]; [medegane] 'his sockeye' 95 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0089100/manifest

Comment

Related Items