UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A prosodic analysis of Nłek̉epmx reduplication Jimmie, Mandy Na’zinek Beverly Dale 1994

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

Item Metadata


831-ubc_1994-0322.pdf [ 3.31MB ]
JSON: 831-1.0087452.json
JSON-LD: 831-1.0087452-ld.json
RDF/XML (Pretty): 831-1.0087452-rdf.xml
RDF/JSON: 831-1.0087452-rdf.json
Turtle: 831-1.0087452-turtle.txt
N-Triples: 831-1.0087452-rdf-ntriples.txt
Original Record: 831-1.0087452-source.json
Full Text

Full Text

A PROSODIC ANALYSIS OF N4EKEPMX REDUPLICATION by MANDY NA’ZINEK BEVERLY DALE JIMMIE B.Ed., University of British Columbia, 1979 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES (Linguistics Department)  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA April 1994 © Mandy Na’zinek Beverly Dale Jimmie, 1994  In presenting this thesis in partial fulfilment of the requirements for degree at the University of British Columbia, I agree that the Library freely available for reference and study. I further agree that permission copying of this thesis for scholarly purposes may be granted by the department  or  by  his  or  her  representatives.  It  is  understood  an advanced shall make it for extensive  head of my that copying or  publication of this thesis for financial gain shall not be allowed without my written permission.  (Signature)  Department of  Linguistics  The University of British Columbia Vancouver, Canada Date  DE.6 (2/88)  April  27.  1994  ABSTRACT N+e9kepmx (Thompson-Salish) spoken in the Pacific Northwest, is morphologically complex, and consequently provides a rich data base for analysis within several different theoretical frameworks. Of particular interest to the present research is the further development of the theory of Prosodic Morphology as applied to the different types of N4e9kepmx reduplication. Although specific analyses of selected reduplicative processes from N4e9kepmx and other Salishan languages have been advanced in the literature, there continue to be many issues which require clarification. It is the intent of this analysis to provide empirical data from the five N+e kepmx reduplication types (augmentative, 9 diminutive, out-of-control, characteristic and affective) to support the essential claims of McCarthy and Prince’s theory of prosodic morphology (1.986, 1990), i.e. that templatic morphology does indeed rely on units of the prosodic hierarchy. A further goal is to clarify issues presented in previous analyses, (Thompson and Thompson (1992), Broselow and McCarthy (1983/4), Broselow (1983), and Jimmie and Shaw (1991)). Chapter 1 provides an introduction to the basic N4-e kepmx phonological and 9 morphological structure. Chapter 2 discusses the fundamental units of prosodic organization (mora, syllable, foot and prosodic word) and the constraints on canonical structure and consonant clusters. Chapter 3 provides a descriptive analysis of the five types of N+e9kepnix reduplication processes. Chapter 4 discusses the theoretical framework of McCarthy’s Prosodic Morphology (1986, 1988, 1991) and previous analyses of N4’e kepmx reduplication 9 (Thompson and Thompson (1992), Bell (1982), Broselow (1983) and Broselow and McCarthy (1983/4)). Chapter 5 presents a prosodic templatic analysis of Ne’kepmx reduplication and discusses a number of issues and problems that arise from the data in regard to prosodic morphology. In conclusion, each of the reduplication processes is uniquely defined by: whether the base is prosodically or morphologically defined; how extraprosodicity of the base is characterized; whether the reduplicative template is monomoraic or bimoraic; and whether this template suffixes or prefixes the designated domain. Also, several interrelated issues regarding the N1e kepmx prosodic analysis still require further research: syllable theory 9 (including internal constituency, moraic representation) and the status of schwas and glides.  II  ‘-I.  TABLE OF CONTENTS Abstract  ii  Table of Contents  iv  List of Tables  vii  Abbreviations  viii  Acknowledgement  ix  Chapter 1 Introduction 1.0 Introduction and Background 1.1 Basic Phonological Structure 1.1.1 N+e kepmx Consonantal and Vowel Inventory 9 N{e9kepmx Consonantal Inventory N4-e9kepmx Vowel Inventory 1.1.2 Underspecification of the N+e9kepmx Vowels and Consonants 1.2 Basic Morphological Structure 1.2.1 Root Types: Me9kepnix Words 1.2.2 Affixes Prefixes Grammatical Suffixes Lexical Suffixes Infixes and Morphological Prosodies 1.3 N+e9kepmx Stress System 1.3.1 Thompson and Thompson (1992) Analysis of Stress 1.3.2 Czaykowska-Higgins’s (1993) Analysis of Nxa’amxcin (Columbian) 1.3.3 N+e9kepnix Stress System  1 1 4 4 4 6 7 10 10 12 12 12 13 14 16 16 17 18  Chapter 2 N{e9kepmx Prosodic Representation 2.1 Mora 2.2 Syllable 2.2.1 A Descriptive Analysis of the N+e9kepmx Syllable 2.2.2 Extraprosodicity Status of Glides 2.2.3 Zec’s (1988) Work 2.2.4 Bagemihl (1991) 2.3 Constraints on Canonical Structure and Consonant austers 2.3.1 Canonical Root Structure CVCC Roots -CCC(C)# Roots CCVC Roots 2.3.2 N{e9kepmx Derived Stems #CC- Derived Stems 2.4 Prosodic Categories: Minimal Word and Foot  22 23 25 25 27 28 33 33 34 34 35 43 46 51 51 62  iv  2.4.1 2.4.2 2.4.3  N+e9kepmxcin Minimal Word Lexical Smallest Stem kepmx Vocatives 9 N+e  62 63 64  Chapter 3. N4-e9kepmx Reduplication: Basic Types and Previous Analyses 3.1 Basic Types of Reduplication 3.1.1 Augmentative Reduplication: CVC-/CV3.1.2 Diminutive Reduplication: -CV/-C 3.1.3 Out-of-control Reduplication: -VCI-C 3.1.4 Characteristic Reduplication: -CVC 3.1.5 Affective Reduplication: CV3.2 Multiple Occurrences of N4-e9kepmx Reduplication 3.2.1 Augmentative and Diminutive Reduplication 3.2.2 Augmentative and Out-of-control Reduplication 3.2.3 Augmentative and Augmentative 3.2.4 Diminutive and Out-of-control Reduplication 3.2.5 Affective and Out-of-control Reduplication 3.2.6 Characteristic and Diminutive Reduplication 3.2.7 Three Independent Cooccurrences of Reduplication 3.3 Conclusions  65 66 66 73 79 82 84 86 87 87 88 88 90 91 91 92  Chapter 4. Theoretical Frameworks 4.1 Bell (1982) 4.2 Broselow (1983) 4.3 Broselow and McCarthy (1983-84) 4.4 McCarthy and Prince’s Prosodic Morphology (1986, 1990) 4.4.1 Assumptions and Claims 4.4.2 Prosodic Circumscription (1990) 4.5 Carlson and Bate’s Analysis  93 93 96 99 102 102 104 105  Chapter 5. A Prosodic Template Approach to Nle9kepmx Reduplication 5.1 Issues and Problems : N+e9kepmx Data 5.2 Syllabification and Morafication 5.3 Augmentative Reduplication 5.3.1 Prosodic Base 5.3.2 Prosodic Reduplicative Template 5.3.3 Subsequent Phonological Processes 5.4 Diminutive Reduplication 5.4.1 Prosodic Base 5.4.2 Diminutive Reduplicative Template 5.4.3 Subsequent Phonological Processes 5.5 Out-of-Control Reduplication 5.5.1 Prosodic Base 5.5.2 Reduplicative Template 5.5.3 Subsequent Phonological Processes 5.6 Characteristic Reduplication 5.6.1 Prosodic Base  107 107 108 112 112 115 115 120 121 125 129 131 131 135 136 138 138  V  5.6.2 Reduplicative Template 5.7 Affective Reduplication 5.7.1 Prosodic Base 5.7.2 Reduplicative Template 5.8 Summary and Conclusions 5.8.1 Summary of Descriptive and Theoretical Issues 5.8.2 Identification of Interrelated Issues and Residual Questions  139 140 141 142 144 144 145  Bibliography  148  Appendix I. N{e9kepnix Root Table  152  vi  LIST OF TABLES Table Table Table Table Table  1. 2. 3. 4. 5.  Table 6. Table 7. Table 8. Table 9. Table 10. Table 11. Table 12. Table 13. Table 14.  Phonemic Inventory of N+ekepmx Consonants Underspecification of Vowels Underspecification of N4-e kepmx Consonants 9 Cooccurrence Restrictions in -CC Root Final Clusters Cooccurrence Restrictions in -RC[Oflt] Root-final Clusters Cooccurrence Restrictions in Onset Clusters of Root Forms Cooccurrence Restrictions in #CC[COflt] Cluster of derived stems Cooccurrence Restrictions in #Cl[+COflt]C- Clusters of stems Cooccurrence Restrictions in #C1[+COflt]C2[+COflt] of derived stems Cooccurrence Restrictions in #Cl[OflL]C [+COflt]-] 2 clusters Cooccurrence Restrictions in #C[+coflt]C[+coflt]]$tem clusters Cooccurrence Restrictions in #C[+coflt]C[oflt]]stem clusters Cooccurrence Restrictions in #C[sofl]C[sofl]Jstem Clusters Nle9kepmx Reduplication Types  vii  5 8 9 36 40 46 53 55 57 60 60 61 62 92  ABBREVIATIONS The abbreviations for the cited forms used are listed below: AFF AUG AUG TEMP AUT BENE C CAUS Cl-TAR cont cor CT DEV DIM DIR FMV G HAB TM IMP INDEF INFL INSTR LAB LEXSUF LOC MDL N NOM OC OCP PHAR p1. p055 PRE  Q RECIP RED RELAT s.o. son sonor STAT TRANS  affective augmentative reduplication aumentative template autonomous benefactive consonant causitive characteristic continuant coronal characteristic augmentation developmental diminutive reduplication directive formative glide habitual immediate imperative indefinite inflective instrumental labial lexical suffix morpheme locative middle nasal nominative out-of-control reduplication Obligatory Contour Principle pharyngeal plural possessive prefix uvulars reciprocal reduplication relational someone sonorant sonorant stative transitive  V 3pS 3pO a / /  [ J -  # 6 < *  §  VIII  >  vowel 3rd person-subject 3rd person-object root alpha root form surface form lexical suffix affixation word boundary syllable boundary syllable mora extraprosodic unacceptable form section  ACKNOWLEDGEMENT  I would like to thank the Scw’exmx, ‘People of the Creek, People of the Nicola Valley’, who have supported and encouraged me throughout my studies while attending the University of British Columbia. These qwdmqwamt ‘wonderful’ people include friends, family, Elders, fellow language and cultural keepers, Shackan Band council and staff, Nicola Valley Tribal Council and staff. Also, I would like to thank Dr. Patricia A. Shaw, Dr. Dale M. Kinkade, Dr. Ewa Czaykowska-Higgins, and Dr. Henry Davis for their support, encouragement, knowledge and guidance during my studies. Without these special people, I would never have completed my studies and completed this thesis. Also, a special k”uk”scémx” to Laurence and Terry M. Thompson and Steve Egesdal who first gave me the inspiration and encouragement to further my studies in linguistics and enter graduate school. To my children, Sk’’Isetk’’u and K ’isëist, and my mother who have been patient, supportive 4 and understanding while I completed my studies and thesis, a very, very special thank-you.  Ham ‘e1-  ix  Chapter 1 1.0 Introduction and Background N4-e9kepmx (Thompson-Salish), a northern Interior Salish language, is spoken in southcentral British Columbia along the following rivers: Fraser (Yale to Lillooet), Thompson (Lytton to Aschroft), Nicola (Spence’s Bridge to Nicola Lake) and Coidwater (Merritt to Meadows, approximately 14 miles south of Merritt). The majority of the work on Nle9kepmx has been done by Laurence and M. Terry Thompson who earnestly began to do their field work beginning in 1962 and continued until the early 1990’s. To date, published materials include The Thompson Language (Thompson  and Thompson 1992) and Stylized Characters’ Speech in Thompson Salish Narrative (Egesdal, 1992).  To be published in the near future are “The English-Thompson Finderlist’  t (Thompson and (Thompson and Thompson, 1990ba) and the “Thompson-English Dictionary? Thompson, 1990b).  In addition, numerous articles have been written and published  primarily by Laurence and M. Terry Thompson, Steve Egesdal, and recently, Paul Kroeber, a syntactician.  Prior to the work done in the sixties, the earliest work on N+e kepmx 9  reduplication was done by Herman K. Haeberlin (1918), and early lexical lists were collected by James Teit (1904, 1908). The works of Thompson and Thompson (1990a, 1990b, 1992) have been the primary sources used for this thesis. Although in some cases, additional data from members within the Scw’exmx ‘People of the Creek’ (Nicola Valley-Merritt) community has been elicited to support my analysis. kepmx (Thompson-Salish), like many other Salishan languages spoken in the Pacific 9 Me Northwest, is morphologically complex, and consequently provides a rich data base for morpho-phonological investigation.  Of particular interest to the present research is the 1  further development of the theory of Prosodic Morphology as applied to the different types kepmx reduplication. Although specific analyses of selected reduplicative processes 9 of N+e from Nle9kepmx and other Salishan languages have been advanced in the literature, there continue to be many issues which require clarification. The intent of this analysis is to provide test data from the N+e’kepmx reduplication types against the essential claims of McCarthy and Prince’s (1986, 1990) theory of prosodic morphology, i.e. that templatic morphology is defined in terms of the units of the prosodic hierarchy. A further goal is to clarify unresolved issues presented in previous analyses, that of Thompson and Thompson (1986, 1992), Broselow and McCarthy (1983/4), Broselow (1983), and Jimmie and Shaw (1991). McCarthy and Prince’s (1986, 1990) theoretical framework for prosodic morphology has provided an insightful basis for several analyses of non-concatenative morphology, including reduplication. Within this framework the reduplication process is conceived of as applying to a prosodic or morphological base; it copies this base and maps it onto a morphological frame which is represented by a prosodically-defined template; this template is associated with its base by either prefixing left-to-right or suffixing right-to-left, immediately adjacent to its target. This thesis will present a prosodic morphological analysis of five reduplicative patterns in 9 N+e k epmx, and discuss the particular problems and issues associated with it. Chapter 1 of the thesis will provide an introduction to the basic phonological and morphological structure of N+e kepmx. 9  This will include a representation of the  N4-e’kepmx consonant and vowel inventory (c.f. Thompson and Thompson 1992), followed by a preliminary analysis of the underspecification of the 9 N4-e k epmx vowels and consonants (Archangeli, 1986).  Next the morphological structure of the N1e9kepmx words will be  presented stating the root types, both strong and weak, and the affixes, both prefixes and 2  suffixes. Finally, I will discuss an analysis I adopt for the N+e9kepmx stress system based on Czaykowska-Higgins’ (1993b) analysis of a related Salishan language, Nxa’amxcin (Columbian-Salish). Chapter 2 will discuss the fundamental units of prosodic organization, viz.: mora, syllable, foot and prosodic word, which are relevant to the present research, drawing particularly on the work of Bagemihl (1991) and Zec (1988). Chapter 3 will be a descriptive analysis of the five types of Nle kepmx reduplication 9 processes, specifically: 1) augmentative, 2) diminutive, 3) out-of-control, 4) characteristic and 5) affective. The first three are productive, whereas the last two are not. This chapter will also discuss previous analyses of N4-e kepmx reduplication (Thompson and Thompson 9 (1986), Broselow (1983), Broselow and McCarthy (1983/4)) and Carison and Bates (1990). Chapter 4 will discuss the theoretical framework of McCarthy and Prince’s Prosodic Morphology (1986, 1988, 1991). The theory of prosodic morphology entails a number of claims and assumptions in regard to non-concatenative morphology which should be supported by empirical data from the many languages of the world.  In opposition to  McCarthy and Prince’s (1986, 1990) model, Steriade proposes (1988) a model of modified reduplication where she states that there is a joint occurrence of base copying and one or both of the operations of truncation and segmental insertion. Syllablic adjustment, one of  two mechanisms that accompanies reduplication, employs two constraints: i) prosodic weight of the affix and ii) markedness of its syllable structure. In clear contrast to McCarthy and Prince’s analysis, templates result from the intersection of weight and syllable markedness conditions, rather than existing independently. Chapter 5 will offer a prosodic template analysis of N4-e kepmx reduplication and 9 discuss a number of related issues and problems. 3  Each reduplication process targets a  different base. Moreover, each of the reduplicative processes employs a different prosodic template or characterization of extraprosodicity. Their formal characterizations and their interaction with one another provide an extraordinarily rich empirical testing ground for several tenets of McCarthy and Prince’s theory. Further, there will be a discussion of how each of these reduplication processes interacts with phonological processes, including (i) syllabification and morafication, (ii) stress (re)assignment, (iii) vowel-deletion, (iv) glidevocalization, and (v) vowel coloration and consonantal assimilation. Thus, the basic goal of this thesis is to present a reanalysis of the 9 Ne k epmx reduplicative processes within McCarthy and Prince’s prosodic morphology framework (1986, 1988, 1990).  I conclude that in N+e9kepmx, each reduplication process involves a  prosodically defined  template which is  affixed to  an  adjacent prosodically-  or  morphologically-defined base. Further, the postulated constructs of extraprosodicity and of minimal word are both shown to play a critical role in the proposed account of the N+e k 9 epmx reduplication.  Preparatory to a detailed investigation of the complex  morphophonological interactions outlined above, let us begin with a discussion of the basic phonological structure of 9 N4’e k epmx. 1.1 Basic Phonological Structure 1.1.1 N4-e9kepmx Consonantal and Vowel Inventory N+e k 9 epmx has a significantly complex consonantal inventory. There is a particularily rich post-coronal inventory, with systematic round oppositions as shown in Table 1. As well, there are contrasts in glottalization throughout the stop!affricates and resonant series. 9 N4e k epmx Consonantal Inventory Table 1. shown below is the Ne kepmx consonantal inventory as represented by 9 Thompson and Thompson (1992).  Note however, that Thompson and Thompson treat 4  velars as: simple-(pre)velars and round-(pre)velars and the uvulars as simple-(post)velars and round-(post)velars. In addition, they differentiate coronals into dentals, laterals, post dentals, and alveopalatals. CONSONANTS STOPS  labial p  GLOTTALIZED STOPS  p’  coronal  velar  t (t’)  FRICATIVES  c  c  t  uvular  k  k’ 4,  s  s  RESONANTS  m  n  1  z  y  GLOTALIZED RESONANTS  m’  n’  1’  z’  y’  x  k’  q  k’’  q’  glottal qW  9  qW  h  x” w  ç  Table 1. Phonemic Inventory of Me9kepmx Consonants Retracted coronals, according to Thompson and Thompson (1992), are related to the set of alveolar and palatal segments . Retracted coronals include the phonemic Ic! and 2 and underlying retracted [l(’)]. Thompson and Thompson (1992) claim  !!,  /i(’)/ underlyingly  merges with the plain [l(’)]. Similarly, however the underlying Is! will retract to  [s].  Resonants are treated as sonorants rather than obstruents. Thus, in addition to !m, n, 1, yI and their glottalized counterparts, plain and glottalized /z,  fW/  are also sonorant.  Thompson and Thompson (1992:4) state that resonants are syllabic in certain positions, in particular when interconsonantal between a resonant and an obstruent, when between an obstruent and a pause, or when in initial position. Note that Thompson and Thompson (1992) state that  /r(’)/ and the Iw(’)/ are very similar, and in some cases appear to be in  The gLottalized 1 toanwords’  It! according to Thompson and Thompson (1992)  is  rare and  Limited to apparent  -  2 Thompson and Thompson (1992:7) state that this set of obstruents apparentLy has evoLved from an earLier set. Note I phonemicalLy represent the /1(’)/ and /10/ as the Latter adopting Thompson and Thompson a (1992:3) 3 anaLysis.  5  free variation, for example (1): (1)  pac” paw’  ‘freeze’ ‘freeze’  However, note the following example (2) which has a and a  1w! in the initial consonantal position  If! in final consonantal position of root /wf/’ ‘shake-off-AuG’.  (2)  5 ‘knock seeds or fruits off from plant accidently’ [waI-p-s-t-és] N4-e9kepmx Vowel Inventory There are two sets of Ne kepmx vowels, plain and retracted, represented below in (3) 9 a. and b., although the retracted set is considered “to be less common and to some extent automatic variants of the primary vowels” (Thompson and Thompson, 1992:11). (3) a. Phonemic Vowels:  b. Retracted Vowels  U  e  0  a Thompson and Thompson (1992:40) state that the hi retracts to its counterpart /j and  [i] before  1’! in closed syllables that have underlying stress. They further state that the hi! to [j]  is optional. The /u/ retracts to [o] immediately preceding uvulars (post velars),  /1, and z(’)/. ‘  Similarily IeI retracts to [a]: a) b) c) d) e) f)  between uvulars between rounded obstruents after uvular continuants before /z, z’I except when preceded by a prevelar optionally between labials and uvular obstruents after a labial or postvelar that does not by itself call for retraction, when there is a postvelar or /zi later in the word;  1n Chapter 2, 1 discuss the Morpheme Structure Constraint which states that phonemes with (near) identical 4 place features within a morpheme can not cooccur. The interconsonantal schwa in the root has been lowered by the following pharyngeal. 5 discussed further in Chapter 5.  6  This will be  Thompson and Thompson represent the retracted schwa as both phonemic and phonetic. They state the retracted schwa  []  is sometimes derived from its non-retracted counterpart  /o/, although they qualify their statement that the  [?]  is different from the other retracted  vowels. An example of the retraction of the schwa is shown below (4):  (4) !-pf  ->  [pm-p]  [-?p] ‘speed’ ‘(something) goes fast’  fast-INC  At this time I present no further analysis for the retracted schwa . 6  1.1.2 Underspecification of the N+e9kepmx vowels and consonants Underspecification of N+e kepmx Vowels 9 The following chart, Table 2, presents the underspecification value of features of both sets of N+e’kepmx vowels as presented by Thompson and Thompson (1992). These sets are assumed in the present analysis. Oppositions of particular sets of vowels have been determined by identifying the four basic oppositions by a minimal number of phonologically functional features. Retracted vowels have a phonemic vowels which do not.  [PHARYNGEAL} feature in opposition to  the  The feature [round] is clearly a functional feature, as  evidenced by vowel colouration discussed in section 5.3.5. Therefore, the /u/-[o] pair is set off by being distinctively [round]. The  // is  assumed to be totally unmarked.  Since the [round] vowel and // are as represented above, the have some feature which distinguishes them from  k/,  lu and the /e/ must each  and from each other. Phonological  information as to what features are functional in various rules reveals what the appropriate features are here. These are further discussed in § 5.3.5.  1t does however appear that the feature [pharyngeal] is floating, therefore resuLting in the retracted 6 schwa in the Lexical. form shown in (4).  7  The use of a feature [back] is not fully compelling because /eI crosses into traditional [+ back] territory, although it is unclear as to whether it is phonologically relevant. Note, however, there clearly is a height distinction in the non-[round] vowels which is not apparently functional in the [round] vowels. In order to keep /o/ with no features, both [high] and [low] are needed. For further discussion of  //->[a], see chapter 5.  The following is a preliminary underspecification of N+e9kepmx features analysis of the of the vowels including the retracted ones . Much further consideration will be required to 7 determine a more definitive underspecification analysis to be chosen for further work. N4-e’kepmx Phonemic Vowels: FEATURES:  i  j  e  a  [round] [10]  [high] [PmGEAL]  + +  + +  u  o  +  +  + +  +  +  Table 2. Underspecification of Vowels Table 3. represents a preliminary analysis of underspecified values of both the N4-e9kepmx obstruents and resonants . 8  These are based on discussions with P.A. Shaw (pc.) Glottal stops are [+consonantaL] since they behave with the natural class of obstruents. They are also 8 [+constricted glottis], although not [pharyngeal] since they can be adjacent to pharyngeals. If they were pharyngeal, a violation of the place feature constraint would occur as discussed in Chapter 2. 1, also, present the Eh] as the least marked segment, that is, [+consonantal]. Note however, [h] is epenthetic when on forms which have a vowel in stem—finaL position and a suffix beginning with a vowel, for example [k’nk”enmeh-ü+] ‘critical person, c.f. [n-k”énme-tn] ‘court house. These forms are from Thompson and Thompson (April 1990:132). C+continuant] is considered to be unmarked (Shaw 1991).  8  +  +  +  +  +  +  +  +  +  [sonorant] + [continuant] + [nasal] [lateral] LABIAL + CORONAL DORSAL [round] PHARYNGEAL const.glot.  +  +  k’  n’  m’  m  t  n  +  Resonants:  +  p’  +  -  p  CORONAL [dorsal] [round] PHARYNGEAL [const.glot.]  LABIAL  [conhinuant] [strident] [lateral]  Obstruents:  +  +  1  +  +  c  +  +  +  1’  +  +  +  c’  +  +  +  +  ?  y’  +  +  kw  y  +  k  +  z  +  +  k’  +  +  +  +  +  + +  +  +  + +  +  w’  +  +  w  + +  qW  +  k” q  +  +  + +  +  q’  Table 3. Underspecification of N4’e9kepmx Consonantal Inventory  +  +  +  + + + +  -  qW  +  +  ç  +  +  s  + + +  +  çW  +  +  +  + +  +  +  ç,  +  +  +  -  c  + + + +  çW  +  +  +  x  +  x”  + +  h  1.2 Basic Morphological Structure Like all Salish languages, N1e9kepmx is morphologically complex. Affixal morphemes can be categorized into two basic groups: (i) lexical (roots, lexical suffixes), and (ii) grammatical (prefixes, suffixes and infixes). Forms usually have several affixal types attached to the root. Each of these categories will be discussed in the following sections. 1.2.1 Root Types kepmx has several canonical types of roots with the most common being !CVCI, 9 N4-e athough as claimed by Thompson and Thompson (1990b, 1992), roots range from monoconsonantal to multiconsonantal-multisyllabic forms.  In the Thompson-English  Dictionary (1990b) approximately 2032 roots are listed; these roots are the basis for the analysis discussed in this section. Below are examples which have roots with one consonant, IC! or ICy! 9 as exemplified  in (5): (5)  /X’/  ‘=éni’°  ‘ear’  Roots which have up to five consonants may have three or four syllables as shown in (6)a., although some are considered bisyllabic as in (6)b.,  ((.)  =  syllable boundary)  (6) a.  /+a9.pn.té9/ Ic’akales.p’ü9/  ‘oyster mushroom’ ‘thistle’  b.  !y’tném! [i9t.ném] /wi9nümt/ [s-wi9.-nümt !qwysqne! [qwiys.qne]  ‘in-laws’ ‘beautiful’ ‘Canford’  PLACENAMF.  Twelve root forms of this type occur in the Thompson—EngLish Dictionary (1990b). 9 Following other Salishanists, I adopt the equal signs () for lexicaL suffixes. 10  10  Many of these multisyllabic root appear are lexical forms, although there are some which are not, for example /kéy’i9st/ ‘groan (weakly)’ which unanalyzable (Thompson and Thompson, 1990b.), however, the lexical suffix, !=ey’st! ‘stone’, is very similar to the ending of the lexical form. Of the lexicalized types, 324 were identified in the Thompson-English Dictionary (1990b). The present analysis of the reduplication patterns only considers bi-consonantal and tn-consonantal roots since four-consonantal and five-consonantal roots are less frequent and require further analysis. Perhaps, some of these multiconsonantal roots can be reanalysed in forms of simplex canons since it seems that some of these roots may have morphemes affixed to a CVC root”. N4-e9kepmx roots generally have one or two interconsonantal vowels.  I assume that  Nle9kepmx roots have underlying (albeit maximally unspecified) schwas (Thompson and Thompson, 1992), as opposed to Czaykowska-Higgins’ analysis (1993b) that Nxa’amcin (Columbian-Salish) has vowelless roots and all schwas are epenthetic.  Under the  assumptions adopted here, then, roots in N4-e kepmx are generally considered to have 9 vowels, except for the very rare forms which are monoconsonantal as discussed in (5). The principal rationale for this position is that particular phonological processes require prosodic templates to be copies of prosodic bases, where the copied form has vowels reduced to a schwa or coloured by the effects of adjacent consonants (as discussed in Chapter 5). Some of the N+e kepmx roots are glottalizing roots, where glottals do not surface 9 although these roots will have a glottalizing effect which surfaces only on resonants in  An example of one these potentiaLly reanaLyzeable forms is c’kIkik chicadee which is almost certainLy onomatopoeic. Thompson and Thompson (1990), however, indicate uncertainity with the root by putting a question mark between the root form and the gLoss. The form couLd possibLy be a redupLication form resuLting from a /c’kIkI base.  11  suffixes, e.g. /k’eq”91, [naqw..m1 ‘lick a bowl’. 1.2.2 Affixes Four groups of affixes append to roots: prefixes, suffixes, lexical suffixes and infixes. Often more than one group affixes to a lexical form, creating the highly complex morphological forms common to the Salish language family. Each of the four groups are discussed independently below. Prefixes Thompson and Thompson (1991) state that there are two types of prefixes: nonreduplicating and reduplicating prefixes. Of the first type there are only seven prefixes, which is a small number in comparison with the vast number of N4-e9kepmx suffixes. Stative ‘es, nominalizer s-, and locative n()- are the most frequent prefixess, and they commonly cooccur. These are shown below (7): (7)  ‘(berries, roots) made into pudding with dumplings’ 9es-t-Roo’r-AFFIx(Es) stative-qualitative-pulverize=liquid?  The other prefixes are the reduplicative ones, with an augmentative and affective semantic effect. Examples of these are shown below (8): (8) a. s-müec b. s-m4’-mü4-ec  woman women  nom-(aug)-woman  (9) a. wiyx-ü1-  ‘crying all the time’  cry-habitual  b. w-wiyx  ‘cry’  all-cry  As these constitute part of the subject matter for this thesis, they will be discussed in detail in Chapters 4 and 5. Grammatical Suffixes Suffixes in N4-e’kepmx are more common than prefixes. 12  Again, they can be  reduplicating or non-reduplicating. discussed in more detail below in  Due to the uniqueness of lexical suffixes, they are  § Therefore, in this section only the grammatical  suffixes are discussed. N4e9kepmx forms can have up to seven grammatical suffixes (10) attached to the roots (in addition) to prefixes which may also be present. (10) c’ex’’-min-t-iyxs-e-t-im-es  ‘They congratulate you people.’  happy-RELAT.TRANS-3pS-?-TRANS-3pS-2p0  There are rare instances where suffixes can be ‘cyclical’, that is, a form will have two cases of a particular suffix as shown below (11): (11)  ‘They judge them.’ 12 judge.TRANS-(3p0)-DIR-TRANS-3p0-3pS  As was the case with prefixes, there are allophonic variations to these suffixes which are conditioned by the adjacent environment.  Some of this variation will be discussed in  Chapter 5. Lexical Suffixes Roots can also have lexical suffixes attached to them. Lexical suffixes of this particular  type occur only in three language families of the world: Salishan, Wakashan, and Tarascan. It is noteworthy that they constitute a very productive affixation process in N4e kepmx. M. 9 Dale Kinkade (p.c.) has stated that the maximal number of lexical suffixes in N4e9kepmx could have been over two hundred. Thompson and Thompson have listed at least 133 of these suffixes. The uniqueness of this type of suffix is that they have a lexical meaning attached to them, unlike grammatical suffixes. They should not however be confused with a compounding of stems, since a lexical suffix can not operate independently, as shown  and Thompson (1992:81) state that forms with IIyxs/ can be further expanded. Also the addition of a In—ti creates a new transitive stem. The addition of a second /iyxsl, the first /iyxsl pluralizes the implied 1—0/, 3object.  13  below in (12). (12) xw’é+ es-n-ch=éw’s  ‘road’ ‘the trail/road is fixed’  STAT-LOC-do=road  *ews There are independent, semantically-related lexical forms which enter into true compounding processes.  In Me9kepmx, as cited in Thompson and Thompson (1992),  compounding can occur with two root forms occupying one root position (13) or when the second form is an independent lexical form, as shown below (14): (13) n-cq-xw’é+-ne  ‘I put it right on the road/trail’  LOC-set-road-1-3  (14) +a9=ans 9 s-qáa +a9=ans-s-qáa9  ‘food’ ’ 3 ‘dog’ ‘put horse out to graze’  eat=tooth-NOM-dog  Compounding also occurs with forms that are affixed with a ligature, a compounding connective /-e+/. An example is shown in (15). (15) tm-+-n-I’p=ic’e9  ‘he has no shirt on’  lack-LIG-LOC-under=clothing  Note that the following example, also has the ligature, /-e4-/, but according to Thompson and Thompson the  {{]  has been deleted because of the following [-s], resulting in a allomorphic  form [-ej, as shown below in (16): (16) kwn-e-s-qáa 9  ‘catch a horse’  catch-LIGATURE-NOM-dog Infixes and morphological prosodies Thompson and Thompson (1991) list only six infixes, four of which could be considered allophones to certain reduplicative processes, that is in some cases they occur as an alternative form to the more typical reduplication processes. These infixes are: plural [9j]  The gloss may more likely refer to domestic animal in general. 13  14  and [c’e] and repetitive [a9] and [e(9)j.  Another infix, the inchoative,  [9],  attached after the first consonant of a strong root. In complementary distribution with  is [9],  the suffix, -op, is attached after the C 2 position of a weak CVC root which has the vowel deleted. Examples of both types of inchoative allomorphs are shown below in (17) and (18): (17) Strong root: kw[9]uc ‘become crooked’ crooked[INCHOATIVE]  (18) Weak root: ‘break in two’ q’w’-6p break-INCHOATIVE  Two types of morphological prosody occur in N4-e’kepmxcin in addition to the inchoative above; these are ‘glottalizing’ roots and the specializing extension, [‘].  Thompson and  Thompson (1992) refer to glottalizing roots as those three or four consonantal roots which have a  /9/  in final position, e.g. /zm”/ ‘send-message’. Suffixes with resonants attached to  glottalizing roots will glottalize the resonants following the root. This is shown below in (19) which has two affixal resonants glottalized. (19) zom’-cin’-m’  ‘send a message’  send-message-mouth-MIDDLE  The other morphological prosody is the specializing extension, [‘], which appears to affix to resonants. Thompson and Thompson (1992) state this is unpredictable and not clearly understood.  An example of the effect is shown below (20) on the root, but note the  specializing extension does not occur on the lexical suffix: (20) kn’=éyt kn=éyt  ‘midwife’ ‘helper, assistant’  It can also occur with the diminutive, affective, and repetitive forms as shown below (21): (21) /p’ém/ p’ep’-m’  ‘kindle’ ‘fireplace in a house’  fire-DIM  15  1.3 N+e9kepmx Stress System The N+e9kepmx stress patterns, to date, have not been analyzed in any current theoretical frameworks. This section discusses briefly Thompson and Thompson’s analysis of the N+e9kepmx stress patterns and Czaykowska-Higgins’ analysis of stress-assignment in Nxa’amxcin (Columbian-Salish).  The theoretical approach of the latter provides an  approach for a preliminary analysis of stress assignment in Me9kepmx. 1.3.1 Thompson and Thompson’s Analysis (1992) of Stress Thompson and Thompson (1992:22) state that the position of main stress in a stem is determined by the combinations of possible morpheme types. They state that “main stress falls on the morpheme which by virture of its inherent properties, partly because of its relative position and the inherent properties of accompanying morphemes, is most dominant in that word”. They further state that other morphemes are unstressed or have secondary stress.  Morphemes which attract main stress and are more ‘dominant’ are underlyingly  assigned stress. Other morphemes are underlyingly assigned secondary stress; therefore they never have primary stress. Two N4-ekepmx stress-related rules, (22) a. and b. below, are from Thompson and Thompson (1992). (22) a. main word stress falls on a strong morpheme or, in absence of such a morpheme, on the first suffix that can take stress, or, in the absence of any such suffixes, on the root (prefixes and morphemes without a vowel cannot take stress); b.  unstressed vowels disappear unless they are followed by more than one consonant in a syllable before the main stress.  Further, Thompson and Thompson assign the roots as strong, weak or ambivalent each with predictable characteristics in terms of stress assignment. They define the strong root as taking main stress on the syllable with underlying primary stress. However, if a derived form has two underlyingly stressed syllable stress falls on the last stressed syllable, and the 16  previously underlying stressed syllable will now have secondary stress, which will be deleted at a later stage. However, some weak roots can have underlyingly secondary stress, and they will take main stress if there is no other stressable syllable. In addition to the roots underlyingly having stress, suffixes can also be stressable or nonstressable. Thompson and Thompson (1992) define three types of suffixes: strong, weak or ambivalent. Weak suffixes and non-strong suffixes lose stress in the presence of any other stressable suffix. Greater in number than the weak suffixes, the ambivalent suffixes will take main stress if they are in first position after the weak root. This occurs, in addition, if there is a succession of weak suffixes. With strong suffixes, the main stress remains on the last syllable, as with stems which have a succession of strong suffixes. The previously derived stressed suffixes will now have secondary stress. 1.3.2 Czaykowska-Higgins’ Analysis of Nxa amxcin (Columbian-Salish) 9 Czaykowska-Higgins (1993b) argues that the Nxa9amxcin stress-assignment is accounted for in a principled manner by assuming that two rules of stress-assignment interact with morphological properties of cyclicity and accent. Roots are either strong or weak, and each root type is designated as either “extrametrical assigning” or ‘non-extrametrical assigning”. Further, the suffixes 14 are either recessive [-accent] or dominant [+ accent]. Czaykowska-Higgins elaborates on this by assuming that on cycle II, suffixes adjacent to strong roots are assigned extrametricality by the roots, and therefore can not delete the stress that is on the root; but, on the other hand, on strong roots where there is no assignment of extrametricality on the final suffix, stress deletion of the stem and reapplication of the stress rule will occur. In general, I assume that the same operations  stated earlier (Thompson and Thompson, 1992) prefixes never have stress assigned to them.  17  occur in N4-e9kepmx, which will be further discussed in the following  1.3.3.  Similarly, roots which are weak assign extrametricality to the suffixes adjacent to the root. Weak roots which do not assign extrametricality only get stress if followed by suffixes with no underlying vowels. Suffixes are considered to be accented as either recessive or dominant, where the dominant suffixes will be cyclically assigned stress depending on the weak or strong status of the preceding root and whether it is an extrametricality-assigning root. 1.3.3 N+e kepmx Stress System 9 As stated earlier the N4e9kepmx stress system generally appears to be similar to that of amxcin. Therefore, in accordance with what Thompson and Thompson (1992) call 9 Nxa weak and strong roots, the stress patterns of derived stems of both need to be reviewed. From an initial analysis of the basic stress patterns in N+e9kepmx it appears that there are irregularities in stress assignment. This is consistent with the analyses of Thompson and Thompson (1992), Czaykowska-Higgins (1993b), and other Salishan stress systems. Stress sometimes falls on the root and sometimes on one of several suffixes, and it sometimes falls on internal, non-final suffixes. When it falls on one of the suffixes, stress can fall on either the first or last suffix. Stress assignment also has further irregularities when a lexical suffix and an inflectional suffix are attached to the stem. Again, sometimes it will fall on the former and not on the latter, and vice versa. Czaykowska-Higgins (1993b) states that the strong roots [+extrametrical] and weak roots [-extrametrical] occur more frequently, than their suffixal counterparts. example (23) shows affixes attached to a weak root. (23) Weak roots [-extrametricality]: a. RED-ROOT=LEXSUFF  9 a  =étk”u  ‘haunted lake’ 18  The following  RED-restrict=water  b. RED-ROOT=LEXSUFF=LEXSUFF -x =etk” =üym’xw xa 9 ‘haunted-water country’ RED-restrict=water=country  c. ROOT-INFL-SUFFIX(ES)  9 -s-t-és a  ‘prevent s.o. from doing s.t.’  restrict-CAUS-TRANS-3-3  d. RED-ROOT=LEXSUFF+ INFL 9 -‘ =eyq”’ -wI9x a  ‘develop supernatural power’  RED-restrict=spiritul-ability-DEV  Here the weak root does not have the main stress, which is on the rightmost suffix in all the examples. The above examples in which there is more than one suffix attached to the root exemplify both previous analyses, in that weak roots will yield the stress to suffixes with underlying vowels. A strong root is shown in the following example.  (24) Strong roots [+ extrametricality]: a. (PRE)ROOT=LEXSUFF cék =kst ‘cool hand’ cool=hand  n- cék =tm  ‘(of house) get cool inside’  LOC-cool[INCHOATIVE] = inside b.  (PRE) +ROOT=LEXSUFF=LEXSUFF n -ce’k=énk =xn ‘sole of one’s foot gets cool’ LOC-cool[INCHOATWE] =belly=foot  c.  ROOT +INFL-INFL-INFL cé9k -s-t-es  ‘s.o. causes s.t. to cool’  cooI1NC-CAUS-TRANS-3-3  d. ROOT=LEXSUFF+INFL cék=kst -m  ‘he cooled himself’  cool=hand-MDL  In all the forms in (24), except for (24.b) which has two lexical suffixesstress falls on the strong root. The examples in (23.b) and (24.b) suggest that stress is cyclic since stress is reassigned to the final stressable suffix. The reader is refered to Czaykowska-Higgin (1993b) for other details of this. The following is the N+e kepmx stress rule (25) based on a very preliminary analysis: 9 19  (25) Ne9kepmx Stress Rule: Assign stress to the rightmost stressable element with the following conditions which are dependent on the type of root or suffix which can also be marked as either strong or weak; 1) Strong roots: If the root is strong, stress will fall on the root except when there is more than one lexical suffix or if the lexical suffix is itself strong. 2) Strong suffixes: If the suffix is strong, stress will fall on it except if more than two suffixes are attached, when it will fall on the rightmost suffix. As stated earlier, some of the above examples suggest that stress is cyclic. Other forms which suggest stress is cyclic are the following: (26) a.  ROOT SUFFIXATION STRESS ASSIGNMENT  b.  ROOT SUFFIXATION STRESS ASSIGNMENT  [[9j = etkwu] n/a stress cek [[cék] Stress  -ekst] n/a  In (26.a) stress assignment is not applicable on cycle 1 since the root is weak and as a result the vowel in the root has been deleted. In (26.b) suffixation does not affect stress assignment, since stress remains on the strong root. In summary, in keeping with Czaykowska-Higgin’s analysis, N4-e kepmx roots can be 9 marked for extrametricality and N4-e9kepmx suffixes can be marked recessive or dominant. The basic phonological and morphological Structure of Me kepmx discussed in this 9 chapter is not meant to be an exhaustive analysis; however, it presents the foundation and necessary information required to discuss the subsequent chapters.  In chapter 2, the  fundamental units of prosodic organization will be presented. Discussion will focus on the kepmx mora, syllable, foot and prosodic word, all of which are relevant to the present 9 Me 20  Tz  )JIOM  Chapter 2. N4-e9kepmx Prosodic Representation A definitive analysis of syllable structure in N1-e9kepmx, as is the case with the syllable structure of many other Salishan languages, still requires a considerable amount of work. This chapter, an initial attempt to do such an analysis, is a descriptive and theoretical analysis of the prosodic representation of the N4e9kepmx syllable. In order to determine N4e9kepmx syllable structure, several procedures need to be performed. After a brief discussion of the basic Nle9kepmx syllable structure, I examine first, the canonical form of roots and stems, showing acceptable and unacceptable clusters. Here I adopt Clements and Keyser’s (1985) approach to onset clusters in English, using columnar tables to designate well-formed and ill-formed clusters. In this section, there is also a discussion of N4e9kepmx extrametricality and appendix status (Steriade 1982, Booij and Rubach 1990).  Extrasyllabicity of segments in roots and stems is crucial.  An  investigation of the canonical structure of roots is used to determine the underlying pattern of consonants at both edges. Zec’s (1990) and Bagemihi’s (1991) analyses form the basis for this discussion of syllable theory. Within Zec’s (1990) analysis of syllable inventories, there are four possible language types. Each type is defined by a different relation between moraic and syllabic segments. Syllabic segments are the syllabic peaks in the languages, which occupy either the first mora of a heavy/closed syllable, or the only mora of a light/open syllable. Moraic segments are to the immediate right of the syllabic mora of heavy syllables. Further, Zec claims that syllabic segments are always a subset of the entire set of segments. They can either be a proper subset of or a set coextensive with the set of moraic segments. Since subsequent chapters demand a clearer understanding of what segments are moraic and non-moraic in N4e”kepmx, the distinction will be discussed here. An examination of 22  constraints on acceptable and unacceptable consonantal clusters will be done first.  A  distributional characterization of the status of pre- and post-nuclear segments will further be validated by particular phonological and morphological processes which interact with roots and stems. These will be discussed in Chapter 4 and 5 in more detail. With the overall purpose of providing a conclusive analysis of the 9 N+e k epmx syllable, this chapter is organized in the following manner: in section 2.1 I discuss the mora, in section 2.2 I discuss the syllable, in section 2.3 I discuss the canonical structure of roots and stems, and, lastly, in section 2.4 I discuss the foot and prosodic word. These are the essential components of the framework I use to discuss the main 9 N+e k epmx reduplication processes throughout this work.  2.1. Mora In many current theoretical phonological studies, such as McCarthy and Prince (1986, 1990) and Steriade (1988), the mora is argued to be an element of the prosodic hierarchy. The 9 N4-e k epmx moraic representation motivated here is based on the distribution of the inchoative  [‘p]  (see section and the various reduplication morphological processes  discussed in chapters 4 and 5. The prosodic representation argued for here is shown below (1): (1) a.  6  /Flt CVC Since the inchoative infix,  [9]  for the strong stems is inserted before a stressed vowel of the  strong root, the association of the onset C must be directly to the syllable node as in (1) and not directly to the initial mora, as represented in (2) below (Zec 1988):  23  (2 “ ‘  *  6I”.... JIL  /11  cvc Thus, (3) is a moraic representation of an inchoativized [+ strong] root/stem: (3)a.  b.*  6  //I\  //LI  6 LL  /711  -7iI  é + ‘become cooler’  c’  9  é+  If the N4-e9kepmx prosodic representation included both the initial consonant and the nuclear vowel dominated by a mora, then the inchoative, glottal would be ‘true’ infixation of a segment into the middle of a prosodic constituent as in *(3.b). Broselow and McCarthy (1983) and McCarthy and Prince (1986, 1990) argue against the existence of such infixation processes. Since they convincingly demonstrate that characteristically such cases reduce to prefixation or suffixation to a prosodically well-defined base, I therefore conclude that the initial consonant of N+e kepmx roots is in fact attached to the syllable rather than the mora. 9 This allows the position of the inchoative  [9]  to be characterized as prefixation before the  syllabic (Zec 1988) or nuclear (Shaw 1992) mora. Now, the next issue is whether the segment which immediately follows the left-most mora is either attached to it as in 4(a) or attached to the syllable root as in 4(b), or to another mora as in 4(c). (4) a.  b.  CvC  6  c.  CvC  24  CvC  I take the position that the N4-e9kepmx closed syllable, CVC, is bimoraic as in (4.c)’. The status of segments which are not immediately adjacent to the vocalic nucleus requires discussion. Since it is very common in N1-e’kepmx to have consonant clusters in stem-initial and -final as well as stem-internal positions, two questions arise: what is the prosodic status of these ‘other’ pre- and post-nucleus positions, and how is each attached to the syllable? In the next section, based on preliminary observations of N4’e9kepmx data, evidence will be presented on the prosodic representation of N4-e9kepmx syllables.  2.2 Syllable 2.2.1 A Descriptive Analysis of the Nle kepmx Syllable 9 Preliminary observations show that the N4-e9kepmx syllable minimally can be either open (5) or closed (6)2. (5) open syllables:  (6) closed syllables:  q  //  ‘  é.  n i  //u ‘ear’  s  m u. 4-  é  c  ‘woman’  There is no vowel length distinction, therefore heavy open syllables (CV: or CVV) do not exist. However, there are vowel-glide sequences as shown below (7): (6)  /‘eyk/  ‘kinnickinnick (or bearberry)  /céwt/ s-céwt-mx  ‘ultimate’ ‘youngest child’  NOMINATIVE-ultimate=people  téw-mn  ‘store’  c.f. Czaykowska—Higgins (1993b) for a similar cLaim regarding syLLables in Nxa7amxcin, BagemihL (1991) for Bella Coola and Jimmie and Shaw (1991) for N+e’kepmx. The dot C.) represents the syLlabLe boundary. 2  25  sell-INSTRUMENTAL  twu  ‘sold to someone’ OUT-OF-CONTROL  sell-OUT OF CONTROL  The maximal representation of the N4-e’kepmx syllable is not simply CVC for, frequently there are consonantal clusters in pre- and post-vocalic positions. In fact, as discussed in section 1.2.2, there are some morphologically derived forms which can have up to 6 consonants clustering in final position as (7). (7)  cü.+=qsxtxv  ‘you disconnect his end of it’  disconnect=head-BENE-TRANS-2-3  With respect to syllable nucleus positon, in addition to N4e kepmx vowels represented 9 as syllabic segments, there are syllabic nasals 3 (Thompson and Thompson, 1986:141). Significantly, however, their distribution is limited. That is, syllabic nasals seem to only occur in hi- or poly-syllabic forms; that is, they never constitute the sole syllabic nucleus in a word as shown in (8). (8)  6  /1 p  u  6  IN  /ui n m s  ‘he found/finds it’  Another interesting related fact is that only vowels are stressed; syllabic nasals are never stressed. Both the unstressability of nasals and their distributional limitation are undoubtedly relevant to foot structure, but a detailed investigation of Ne kepmx foot prosody is beyond 9 the scope of the present study. Also, with my native speaker judgement, I consider the nasals in word-initial position to be non-syllabic, similar to Czech (Booij and Rubach, 1990).  Such onset clusters then  constitute a sonority hierarchy violation, as shown in (9.b), but I propose that the initial nasal  1n fact, Thompson and Thompson (1991, 1992) state that in word initial positions liquids are usually 3 syll.abic, as discussed further in this section. However, semi—vowels, they state, wilL retain their semivocalic quality in the same environment; However, if they become syllabic they are preceded by a glottal stop.  26  Examples where a nasal [m] is syllabic in the rightmost  is licenced by extrametricality.  syllable, whereas the [n] which is adjacent to a vowel or more sonorous element is not 4 as shown in (9). syllabic (9) a.  b.  I\1  hr  <m>lam mns ‘his medicine’  /tr  <m> t olt ‘blot clot’  Consonantal clusters at word-initial and -final position of stems generally have nasals, as in the following example (10) where both the LOCATiVE prefix and the INSTRUMENTAL suffix have nasals. (10) iII.L  /I  iU  ii tiI!  II Hflu tn  np  n-pt=Inus-tn  ‘way of thinking’  LOC-?intervene=mind-INST  2.2.2 Extraprosodicity Extraprosodicity always occurs at the edge of a prosodic domain, and it can be either <C(C)>. Booij and Rubach (1990) claim it to be linked to the phonological word and not stray erased. A representation of a syllable with an appended segment is represented by a dotted line from the syllable node to the consonant, and the extrametrical segment is shown with the angle brackets, as below (11): (11)  6  6  I\ LL  I I V  CCC<C>  VC<C>  “me [n] could be considered ambisyllabic, but I do not represent it as such in (8); [m] in the second syllable in (9) is considered the same.  27  <Cc> Three non-moraic consonants in stem-initial/final positions as shown below (12). The examples below show a typical formation of stem-initial forms which have three consonants. (12)  stem initial: a. C1[NowsTA7E] C [Loc] C 2 [Roo 3  s-n-x”üy’t  ‘springtime’  stem medial: b.  C1[N0M or LOC] C2[RooT]  C3[JNCH]  n-c’m-p=ép  ‘(of  basket)  For this analysis, I look at both root-initial clusters and root-final clusters; the latter are more common than the former. I then extend this approach to consonant clusters, both in initial- and final-positions of stems. Each of the allowable or occurring clusters for root and stem forms are shown in the following sections.  Clusters in initial, internal, and final  positions are also examined. Status of glides The status of glides is a contentious issue as stated by (Hyman 1985, Guerrsel 1986, Hunt 1991, Shaw 1993).  Glides, on one hand, have been regarded as the surface  representation of underlying vowels, for example by Shaw (1991, 1993) in her analysis for Nisgha. That is, lu-> [y] or /u/-> [w]. On the other hand, it has been argued that glides are underlyingly consonants (Hunt 1991). Thus, the question regarding Ne kepmx is: are the 9 CVy(’) and CVw(’) roots really /Ci(’)/ or lCu(’)/? At this stage of my analysis, I consider kepmx glides, [y(’)j and [w(’)], to be consonants rather than underlying 9 Ne  /(9)/  or /u(9)/.  Substantive evidence for the underlying representation of glides in N4-e9kepmx is discussed below. Consider the following CVC roots, where C 2 is /y/. Note from data such as (13)a. and b. that although a /y/ can vocalize to [i] in roots where the root vowel  28  // is deleted, such [i]’s  suggests that they are underlyingly non-vocalic. (13)  a.  ..1cy  ‘new’ ‘new boat’  ci-cy=éw ci-cih = éw+ AUG-new=conveyance  fmy n-mi = cm n-mi = cm  =  éyt  ‘cross-over’ ‘interpret’ ‘translator’  LOC-cross-over= mouth =agent  b. fmy’x miy’x-e-cüt miy’x-xi-c  ‘kick’ ‘kick o.s.’ ‘kick s.t. for someone’  kick-BENE-3-3  Note, in contrast, the stress behavior of CVy roots where the nucleus is a full vowel, as in (14.c,d). (14)  c. /Ciy/  sftIy fcíy c’iy=kst ciy-ci=kst-m-tm 5 ciy-ci=kst-mIn-tm  ‘tea’ LOAN ‘open?’ ‘five’ ‘the five of us worked together on it’ ‘  AUG-spread=hand-REL-3-3  d. /Cuy/ 9es-püy=qn  ‘(of container) covered’  STATIVE-cover=head  e. /Cay/ n-qáy-ix  ‘swim’  LOC-dive-AUT  The following example (15) retains the nuclear root vowel /e/ and the 2 C /y/, although one would expect that the /e/-> [] when unstressed, which would then be subject to deletion, triggering vocalization of the /y/. Thompson and Thompson (1991, 1992) posit that the underlying !é/- > [e] is retained here because of its previous stressed status (from an earlier stage of the derivation).  mis form and the one above are cited from the Thompson and Thompson (1990) and are acceptable to Nora 5 Jimmie, my mother.  29  (15)  fmeyqve9  ‘cause rope to be tangled’ LOC-tangle=rope-CAUS-3-3  The form below shows variation in the reduplicated form. Like the examples in (14), the forms in (16)a. retain the [uy], whereas this is reduced to [i] in the second form, (16)b. This latter form is derived by the regular processes of unstressed V-reduction ([uy]  ->  [oy])  and subsequent a-deletion and glide vocalization ([y]->[y]->i) (16)  fmuiy n-muy-müy-1e9=xn-me n-mi-müy-le9 =xn-me 6  ‘immerse one’s legs’  LOC-AUG-dip-in-water-FORM=foot-INDEF  Examples which have glottalized ly’!  as  2 are shown in (17) below. These behave in the C  same way as the unstressed forms in (13): (17)  ..fc’éy’  c’éy’-m  ‘warn about something’  warn-MDL  c’i9-c’éy’-m  ‘warn about things’  AUG-warn-MDL  c’i9-c’ey’-m-ü+  ‘informer, tattletale’  AUG-warn-MDL-HAB  The other glide, /w!, in C 2 position is shown below (18): (18)  %fwek or tw a. b.  n-tuwek=Ik’n-s n-twek=Ik’n-s  ‘walk behind s.o.’  LOC-jump-over=back-3  %Jliw  lI[l}u-tn  ‘Thunder  Spirit’ME, MYTHOLOGY  rumble[DIM]-INST  fpIw  n-piw=le9x  ‘bread’  PROBABLY LOAN, OKANAGAN  LOC-swell-up=bread  %Jlüw  s-lüw=ec’  ‘shredded inner bark of red cedar’  NOM-?warm = bark  ../láw  n-láw=cin-me  ‘gargle throat’  LOC4constrict=mouth-INDEF  These two semantically identical forms are cited from Thompson and Thompson (1990), note however, the 6 second form is not an acceptable form to Nora Jimmie.  30  ..Iq’aw  ‘es-q’áw=qn  ‘have loops [for attaching]’  STAT-attach-by-string=head  fx”üw  wüwmn  ‘traditional spinning toy, top’  ?spin-buzz-INSTR  The data below illustrate that other canonical root forms which have a glide in C 2 position all behave the same way as the /CVG/ roots exemplified above. There are several CVGC roots which have an underlying IyI or 1w! shown below. (19) a. %fméyx  s-méyx  ‘snake’  NOM-snake  s-mi-méyx  ‘snakes’  NOM-AUG-snake  b. %fk”’Iy’t  skwIyt  ‘PLACE NAME’  NOM-(placename) c.  fcéwt  s-céwt=mx  ‘youngest child’  NOM-ultimate=people  d. fc’Iwq’  c’Iwq’-e-s  ‘tear in small pieces’  tear-small-DIR-3-3  However, I do not consider them relevant for this argument; therefore I do not discuss them any further here. All the forms above have examples where the root vowel has main stress. Compare a form below (20) which has a suffix attached to it and the root vowel is deleted, as in the CVG forms.  I9éyk!  (20)  9 é yk ik= éi-p-m  ‘kinnickinnick’ ‘gather kinnickinnick’  kinnickinnick= plant-MDL  With CVC roots, C 1 can also be  /y/. Initial /y/ may be followed by Ii,u,I, but not by Ia!.  It may be hypothesized that this is because height of articulation is significantly different. Examples are shown below (21):  (21)  a.  %fyuq  yoqves yqwyu[y1qw..fls  ‘splash something’ ‘splash s.t.!s.o. a little repeatedly’  AUG-splash(DIMj-INDEF-3-3  b.  9es-t-y1”-p-e  ‘s.t.  has just now disappeared’  STATIVB-QUALITATIVE-disappear-INC-MDL  31  c. [yi  n-y-yi=é+c’i9  ‘quick-witted, intelligent’  LOC-AUG-lucid=insides  d. *fyaC  no examples (Thompson and Thompson, 1990b)  kepmx morphological processes consistently treat the glides as a consonantal 9 Ne segment. Consider, first, the behavior of reduplication as in the example (21)a. above. Evidence of the /y/ behaving as a consonant is shown by the affixation of the two types of reduplication as shown in the example, reduplication, the prefixed (CVC-)  In both instances of  AUGMENTATIVE  and the infixed [C ] 1  DIMINUTIVE,  the /y/ is  treated as a consonant. Also, note that (21)b. is an example which has other types of affixation,  INCHOATIVE  and  MIDDLE.  Here one might imagine that the [y] would vocalize, to  yield *[9esticwpe], but it does not; instead, it retains its consonantal status. A second type of morphological process, inchoative infixation, also treats glides as consonants. Recall from the discussion of (3) in section 2.1 earlier, that the inchoative form of [+ strong] stems is formed by the insertion of a before the first mora.  [9]  after the initial C of the stem, and  As shown below in (22), the initial glides  1w! and ly/ are treated as  (onset) consonants, and not as moraic consonants. (22)  a. /wVCI b. IyVCI  --> -->  [w9VC] w914 [y9VC] y9Ic’’ á 9 y  ‘to fray’ ‘to become loose’ ‘come to be apart’  A further argument for the underlyingly consonantal status of the glides stems from the fact that they occur with phonemically contrastive glottalization.  In this respect, the  glottalized /y’I and /w’! parallel all other segments in the resonant consonant series, e.g. m’, n’, 1’, z’, y’, 1’ and  Note that this parallelism would not exist if [y] and [w] underlyingly  were considered vowels since the non-high vowels counterparts, such as *a or  32  I/ and Ia! have no glottalized  In summary, since the glides !y(’)/ and /w(’)/ function as consonants, I conclude that they are not underlyingly vocalic segments. That is, /CVG/ and /GVC/ roots do in fact occur. I use this as a basis for further discussion in subsequent sections when I discuss canonical root consonantal clusters.  Glides, however, may change their consonantal status; for  example in chapter 5, I provide evidence that glides vocalize when syllables require a nucleus after vowel deletion. 2.2.3 Zec’s Framework (1988) Following Zec’s (1988) syllable typoiogy, I categorize the N4-e9kepmx syllable as a Type One, that is: (i) syllabic segments are a proper subset of the set of moraic segments; and (ii) any consonantal segment can head the second mora of a syllable. N4-e9kepmx syllables can either be open/light (CV or CN, where N=nasal) or closed/heavy (CVC). The subset of moraic segments which can be syllabic includes vowels and nasals. Thus, N4-e’)kepmx nasals are the syllable nucleus, moraic or the onset (as shown earlier in (8)). According to Thompson and Thompson (1992), the property of syllabicity is further extended to the velar [-y] and pharyngeal [cj segments in initial position as shown below (23): (23)  ‘yn-p  ‘he shivers’  shiver-INC  cc-op  ‘It gets ensnared.’  snare-INC  As stated earlier in  § 2.2.1, I consider the initial resonants excluding nasals to be non  syllabic. 2.2.3 Bagemihl’s Analysis (1991) Bagemihi’s (1991) analysis of Bella Coola claims that the theory of prosodic of licensing requires further revisions particularly with the status of the mora. In earlier works prosodic licensing demanded that a segment be syllabified and that it be affiliated to a higher  33  prosodic constituent and to remain in the representation. However in Bella Coola, many have claimed it to be a language with vowelless words where it is necessary to determine the status of the obstruents in relationship to the syllable. It is noted that Bella Coola segments that cannot be syllabified are not necessarily deleted. In fact, Bagemihi concludes that in Bella Coola segments which are linked to the mora cannot be deleted. In other words, obstruents which are linked to the mora cannot be deleted. They also must be in the representation. Thus, the mora which had previously considered to be only part of the syllable, now must have the ability to remain in the representation.  2.3. Constraints on Canonical Structure and Consonant Clusters 2.3.1 Canonical Root Structure The majority of the Me9kepmx roots have the canonical structure CVC (Thompson and Thompson, 1992) which can be subdivided into three types (24): (24)  i) weak roots with an underlying schwa; ii) weak roots with a full unstressed vowel; or iii) strong roots with a full stressable vowel.  There are also tn-consonantal roots of the form CCC e.g. /cp/ ‘break-ice’, CCVC e.g. !c+Iq’/ ‘muddy’ or CVCOC e.g. /k’é9y ‘waste-away’ where the V represents a stressable vowel. It should be noted that in some cases, there are too few lexical forms to uniquely determine the underlying representation of certain tn-consonantal forms. Since the essence of this discussion is the representation of the 9 N+e k epmx syllable, the status of consonants other than those in the core syllable (CVC, CVC and CC roots) will not be discussed further in this section . 7  As stated earlier, the majority of the data provided throughout this paper is primarily from the Thompson— 7 English Dictionary (1990); thus, the root designations are as determined by Thompson and Thompson unless explicitly stated to the contrary. Throughout this anaysis, if I make a different caim on the status or form of the root, I discuss these variations in footnotes or in the body of this paper.  34  I adopt Clements and Keyser’s (1985) methodology for the analysis of constraints on syllable-initial and -final consonant sequences.  The following is an application of their  kepmx #CC- and -CC# consonant clusters in roots. Roots 9 approach to the analysis of N+e which have consonantal sequences greater than #CC- and -CC# are discussed, but they are not exemplified in tables. In  § and tables are used to show root-final clusters. § will be a  discussion of root-initial clusters. #C - initial segments will be represented with the C 2 C 1 1 2 consonants in columns. consonants in rows and the C  Final clusters -C # will be 4 C 3  represented with -C 3 consonants in rows, and the -C 4 consonants in the columns. Allowable clusters are represented by a  “+“,  whereas unallowable clusters are shown by a  Roots  apparently do not occur with clusters at both ends, that is 3 VC Root-final clusters 2 1 . 4 C are more frequent, than root-initial clusters; therefore, I discuss them first. CVCC Roots Although N+e kepmx primarily has CVC or CC roots, there are 148 CVCC roots out 9 of the total of 2032 Ne9kepmx roots examined in this analysis. In this section, I represent the canonical forms of the CVCC roots, and I discuss the consonantal clusters in terms of their sonority relationship with each other. I also propose a constraint which seems to be in effect in regard to the correlation between various consonant clusters. The following is a representation of the 148 roots with consonant obstruent clusters in root-final position. Table 4 presents the particular cooccurrence restrictions found between the stops in these clusters.  35  3 \ 2 C C  p  p’  t  p  -  -  -  p’  -  t  c  k  -  c’  k  q q’  q’  qW  qW  ‘  -  -  k k’ k” kv  q  -  -  + C’  k” k\v  +  -  C  k’  + -  +  -  -  -  +  +-  -  -  -  -  -  -  + -  +-  -  +  -  + -  -  -  -  -  + +  -  -  +  -  qW qW  -  9  -  + + + + + + + + +  Table 4. Cooccurrence Restrictions in CC root-final clusters An issue to be reviewed is how Ne9kepmx clusters in root forms can be represented in terms of the Sonority Hierarchy shown below (25), (Steriade, 1982) in comparison with Hankamer and Aissen (1974), Kiparsky (1979): (25)  Sonority Hierarchy stops/affricates  <<  fricatives  <<  nasals  <<  liquids  <<  glides  <<  vowels  The universal Sonority Hierarchy states that the adjacency relationship between segments in tautosyllabic forms is restricted according to the sonority of each segment. The most sonorous segment will be the nucleus element, and the other segments outside of the nucleus will be progressively less sonorous elements. The sonority value will decrease or will be equal between segments, and the least sonorous segments will be outer segments. Because stop-stop clusters are equal in sonority, the segments in the following -CC# consonantal clusters of CVCC roots do not violate the sonority hierarchy (25). The CVCC 8 in (26) and (27) below exemplify the data presented in Table 4. roots  weak.  forms are represented here being underLying marked for stress if strong and unmarked for stress if  36  (26)  stop+stop#: métk” mék’p k’Iqt  (27)  ‘transport fire’ ‘testicles’ ‘sky’  stop + glottal# : 9 ‘?independent-visible’ cBk” ‘?relentless’ Ieqw9 ‘lick’ cIq’ ‘?encourage-practitioner’ c’uq’’9 ‘point-out’  Now compare the distribution of affricates and stops. Affricates occur in both postnucleus positions, C 2 and C , as shown below in (28) and (29). Based on this behaviour, it 3 is reasonable to posit that aifricates have the same sonority value as stops. Further, this patterning of the aifricates with stops motivates representing both classes as  [-coNT],  as in  Shaw (1991). (28) stop + affric.#: qápc  ‘spring-warmth’  (29) aifric. +stop#: qeck 9écq” iIc’p’ 1 k”  ‘older-brother’ ‘bake’ ‘flea’  Let us turn from a consideration of manner, to a consideration of place. From the evidence in Table 4 above, a striking cooccurrence restriction in these tautomorphemic clusters emerges. Except for a single form, which Thompson and Thompson (1990b) note is from the Lytton dialect: k’éct, ‘tanning stick’, there are no clusters of  [-coNT]  consonants  with similar place features. That is, in dividing the segments of Table 4 according to the major place of articulation categories  -  LAB  (p,p’),  [coR]  (t,  ‘,  c,  c’),  [DORSAL]  (k, k’, k”, kv),  Glottals, although patterning with stops, are grouped differentLy here because of their effects 9 following segments. The status of all /CVC’I roots have glottalizing effect on resonants which follow although questionable, I adopt those forms. These forms are referred to as glottal.izing roots (Thompson Thompson:1990). The surface froms of these CVC roots wiLl not have gLottals, but following resonant(s) in stem will be glottalized. For exampLe: n-c’Iq’-n’-s ‘encourage s.o. to do their best.  37  on it, and the  [PHAR]  qW, (q,  q”),  q’,  it is evident that no two segments from the same place cluster  together. From this, it may be concluded that there are Obligatory Contour Principle (OCP) effects functioning to constrain identity of place in directly adjacent segments. Note that McCarthy (1986) defines the OCP in terms of absolute identity, i.e. “at the melodic level, adjacent identical elements are prohibited”. Here in the Nle9kepmx data that is under consideration, it is not only strictly identical segments which are prohibited, but also partially identical segments, specifically segments which are both  [-coNT]  homorganic (for other  similar cases of identity of place constraints, see McCarthy (1991) and Pierrehumbert (1992). This OCP constraint against identical place specification is an underlying morpheme structure constraint, but it is not applicable at later derivational levels. This is shown by the following data (30), where it is clear that identity of place is tolerated across morphological boundaries. (30)  qwc..c..s..t..es  ‘move s.t. accidently’  move-OC-CAUS-TRANS-3-3  mlám-m  ‘new dancer’  heal-MDL  k”én-na  ‘I take s.t.’.  take-3-3  Two other observations are worth noting from Table 4. First, it is evident that none of the uvulars ever cooccurs with the velars. [nwmoE]  If the velars are exclusively defined by the  node for place, then their failure to cooccur with the  [DORSAL]  segments would  be explained. This suggests, therefore, that the uvular segments in N4-e kepmx might be co 9 articulated as both patterns with  [DORSAL]  and  CORONAL, DORSAL,  [PHARYNGEAL].  and  PHARYNGEAL  The second observation is that  /9/  freely  segments. This would not be expected if /9/  were [PHARYNGEAL] in N+e9kepmx, as it arguably is in certain Semitic languages (McCarthy 1991) and in Nisgha (Shaw 1991). This suggests then that 38  /9/  is  PEAR  in Me9kepmx; this  supports Bessell and Higgin’s (1991a) conclusions about its behaviour in other Interior Salish languages. The following is a preliminary version of the proposed Morpheme Structure Constraint, the Nle9kepmx Place Feature Constraint (31), that is in effect for root final consonant clusters. (31) *  Morpheme Structure Constraint [-cont] [-cont] .place 1 place 1 ROOT [a] [a]  Next, I discuss Nle9kepnix roots in terms of the sonority relationships between the segments in consonant clusters other than  [-coNTINuT][-coNTINuT]  clusters. It will be  shown that most clusters follow the sonority hierarchy, although some do violate it. ) obstruent (C 2 ) sequences which are 3 Table 5 documents the particular resonant (C -  found. All these are cases which show that the more sonorous segments, the resonants, do indeed occur before less sonorous obstruents. It should be noted that  , in striking contrast to the very 3 It! and /9/ occur freely as C  limited distribution of other segments in C 3 position. In terms of radical underspecification theory, /t/ and  /9/  are posited to be the least marked segments of the consonantal inventory  (see, Table 3). Specifically, neither is postulated to have any place specification (CORONAL  being underspecified for It!, as argued in several papers in Paradis and Prunet  (1990); and  /9/  not having any place specification at all). These independent hypotheses in  regard to their lack of phonological place specification correlate directly with their extensive freedom of distribution in C . 3  39  3 \ 2 C C  m m’ n  n’ 1 1’  y y’  z z’ w w’ .y ..y’  c.  çW  p  p’  -  -  -  -  -  -  -  -  -  -  -  -  +  -  -  -  t  +  -  -  + +  -  -  -  -  -  +  -  kE  +  +  -  k”  -  -  -  -  -  -  +  +  -  +  -  +  -  -  -  qW q  q’  qW  + +  -  -  -  k’  -  -  +  k  -  -  +  -  -  -  ÷  -  c’  -  -  + + +  -  -  c  + -  -  -  k’  +  + + + +  -  -  -  +  -+  + -  c, çW  Table 5. Cooccurrence Restrictions in  RC[Oflt]  root-final clusters  Based on the -CC# data above, Table 5, resonants in C 2 and obstruents in C 3 do not violate the sonority hierarchy as shown in (32) and (33).  (32) 1 +sonor 2 C 1+C [Ofl]#: 3 qáyt ‘reach top’ q’Iyt ‘avoid-rain’ 9ey’k Xwüyt  zélt mil’t c’Iwq’ céwt héw’t (33) -C [l] + j#: 2 30 C 1 mk X’ümk’ qwjnt c’m 10 zm’  ‘kinnickinnick’ ‘exit’ ‘dish’ ‘visit’ ‘tear-small’ ‘?ultimate’ ‘rat’  ‘niece’ ‘drinking-tube’ ‘marmot’ ‘eat-soapberry-confection’ ‘send message’  As the other roots which have glottals in final position, there is a glottalizing effect on resonants 10 which are suffixed, e.g. [zm’=cin’-m’] ‘send a message (by word).  40  It should, also, be noted that only some resonants occur in C , i.e. y(’), l(’), and w(’), 2 whereas the following do not occur in C 2 position: /z’/, /-y(’)/, /1’/, /f’! and  /fW/•  Let us now turn to a consideration of the subclasses of root clusters, which entails a reversal of the expected sonority relation. In the tautomorphemic root forms (34) below, fricatives are external to the aifricates (34) aifric. +fric.#: méc’x ücs  ‘blink’ ‘survey (one’s catch)’ ‘abuse’  If fricatives are considered an independent class in the sonority hierarchy, as in the proposal in (25), then this would constitute a violation of the expected sonority ranking. Significantly, however, there are no examples where a fricative in C 2 position precedes an aifricate in C 3 position.  Perhaps this suggests that the classes of stops/affricates and  fricatives should be merged into a single class of obstruents on the sonority scale. A similar conclusion derives from an examination of the sequencing of glottal stop, which was shown above to pattern with stops and fricatives. As illustrated in the data in (35),  /9/  is internal  to fricatives (35.a) and to the fricative resonants (35.b).  (35)  /9/ +  C #: a. 9 +a b.  ‘eat/food’  c’i  ‘PLANT NAME’  z 9 na  ‘goat hair blanket’  Further, the following show both stop-fricative sequences (36), as well as fricative-stop sequences (37). (36) stop+fric.#: ép 9 cItx’’  (37) fric+stop#: n6st c’éxt  ‘choke’ ‘house’  ‘eh?’ ‘brother-in-law’ 41  qwIsp qW  ást X”St  l’at  ‘buffalo’ ‘leg’ ‘rock-cliff’ ‘diverge’ ‘friend’  As these sets of data (35-37) can be accommodated by the more general version of the Sonority Hierarchy which group all obstruents together: (38) Sonority Hierarchy (revised from 25): Obstruents << Nasals << Liquids  <<  Glides  <<  Vowels  The final class of sounds to be considered are the nasals. Roots with a nasal which occur outside of a less sonorous element, a stop, are shown below in (39): (39) stop+nasal#: mütm méq”m  ‘blue grouse’ ‘swamp’  Even though these roots have the same CVCC canonical forms as the roots above, the /m/ in final position after a stop will be assigned syllabic status by the principles of morafication discussed in  § 2.1. That is, these roots surface as bisyllabic, this syllabification being  triggered, as would be expected, by the reversal of sonority in the root-final cluster. In the following root forms, nasals which are more sonorous than fricatives also occur in the most rightmost position of a root. Again, these roots, e.g. [k”am’], will surface as bisyllabic forms as in (40): (40) fric.+nasal:  kvaIn c’é4m’ k”üsn’ k’x”m’ p9zm’ sm’  ‘yellow?’ ‘vegetable-strings’ ‘star’ ‘cluster’ ‘horse-tail’ ‘PLACE NAME, Spuzzum’ ‘balsam root’  It is curious to note that the fricative/nasal forms involve glottalized nasals. On the basis of the available evidence, it is not yet clear whether glottalized resonants should be attributed  42  a less sonorous ranking than plain resonants in N4-e9kepmx (see Zec’s 1988) analysis of Kwakwala). Both nonglottalized (39) and glottalized (40) nasals trigger syllabification, but the distributional anomaly in (40) is nonetheless striking. In summary, if the fricatives, affricates and stops class together, it appears that the N+e’kepmx root-final clusters follow the Sonority Hierarchy. Of particular note are two facts: (i) glottal stop groups with the other stops; and (ii) a nasal external to a less sonorant segment triggers syllabification. Finally, in order to account for the significantly constrained cooccurrence restrictions on consonant clusters in post-nucleus positions, the N4e9kepmx Place Feature Constraint (31) is proposed to be in effect, this prohibiting adjacent consonants which have the same place specification. -CCC(C)# Roots In Me”kepmx, there are only 13 roots which have -CCC# in final position as shown in (41), (43) and (45), and there are only 2 roots which have a -CCCC# cluster (46). The examples are presented and grouped into three categories which are discussed below. The following -CCC# clusters of roots in (41) all begin with a resonant, followed by an obstruent cluster which patterns like the CC# clusters discussed in (41)  /+wéy’st/ /k’len’tx’”/  §  ‘fall’ ‘tule’ II  /q’áy9”!  /pénckl  ‘?cascara’ ‘summer’  The forms in the Thompson and Thompson (Thonipson-EnglishDictionaiy, ‘1990b) do not have any surface forms with 11 the glottal, i.e. q’éy’=e4p, although the root is I.isted in Thompson and Thompson (Tlwrnpson-EnglishDicuonary:1990b) as ‘—final.  43  12 /wew’qi’/  ‘fish’  Assuming that syllables are bimoraic, at this point it is an open question as to how the residual consonants in these clusters should most appropriately be represented. The syllable structure motivated thus far is (42): (42)  6  /L p  e  n  c  k  ‘summer’  This question can be more fully addressed when non-final clusters are investigated. Available evidence, see (45) below, suggests one C can be adjoined as non-moraic to the right-edge of a syllable, and that the final C, being in word-final position, would be licensed by extraprosodicity. Moving on to a consideration of the second category, note that the -CCC# forms in (43) all have a nasal in the final consonantal position. As this nasal will trigger syllabification, the immediately preceding C will function as an onset, and the antepenultimate C will close the preceding syllable.  Thus, all of these consonants fall within the clear parameters of  N+e k 9 epmx syllable structure discussed in sections 2.1 and 2.2, as shown in (44) and (11). (43) /mxéc’xn/ /qwistm!  ‘?purple-huckleberry’ ‘(a type of) saskatoon” 3 ‘?stumble’ ‘barn swallow’  /qwtm/  /qIcec’kn/ (44)  \  6,  I.Li.L  VCCN  The stem form, n-q’m-wew’q4, ‘Bald Mountain, PLACE NAME’ only occurs as a compound. Note this may be 12 a diminutivized form. The Lw’] could be copied from the initial segment. This type of reduplication is discussed in chapter 5. form could be from the root fq”y ‘ripe’, -s ‘T, -tm saskatoon specifically refers to the type for drying.  44  ‘?‘.  Nora Jimmie, my mother, states this type of  The forms in (45) do not appear to have any particular pattern except that a fricative Is! is present in each of the three forms. This follows from the idiosyncratic behaviour of fricatives discussed in the previous section. For the present, these forms will be analyzed like those in (44), with the first C of the sequence being moraic and with the prosodic status of the final 2 consonants indeterminate as in (42):  (45) kéy’i9st sIsq’t to9üstk  ‘groan’ ‘(Place Name) Gody’ ‘catch-fish’  The following -CCCC# forms in (46) can be syllabified as below’ 4 if the nasals are syllabic.  (46) k”iik”.wns  ‘cranberry’ ‘wolverine’  qWilx.qn  (47) -CCCC# 6  tt kW  It  k’’  /  w  n  s  ‘cranberry’  The final form in (46) provides evidence of medial clusters which cannot be accommodated simply into a syllable which is maximally bimoraic with a single C onset. That is, the medial [x] must somehow be prosodically parsed, assuming that N4-e kepmx is a language with 9 exhaustive syllabification, at least up to word margins. For present purposes, it will be assumed that a maximum of one consonant can be adjoined as non-moraic to the right 5 of a bimoraic syllable, i.e.: edge’  using my native speaker judgement in syllabifying the forms, I represent the syllable boundaries with a period (.). Whether this C adjoins to the rightmost mora or directly to the syllable node is immaterial, to the 15 present analyses.  45  A  (48)  /t /1 qW  1  1x  n  q  ‘wolverine’  In summary, the forms do not provide systematic or broad-based evidence for multiconsonantal root-final clusters. They do, however, behave like the -CC# root forms. I therefore conclude that the same analysis can be made for the -CCC# and -CCCC# root forms. CCVC Roots The occurrence of #CC- clusters in root forms as shown above is quite rare; there are 34 out of a possible 2029 roots. These #CC- clusters are shown in Table 6. below: 2 p p’ \C  t  k’  c c’  k  k’ k”  kw q  q’  qW  qW  m n  x” XW  1  y  -y  w  c  Cl  p(’)  +  k(’)  + + + + ++ +  C C’  q(’) qW qW  + S  + --  + + +  x  xw w  + +  z  çW  +  Table 6. Cooccurrence Restrictions in Onset Clusters of Root forms. Of interest is the fact there are 29 occurrences of the glottal stop c’9al ‘tingle’. This suggests that  /9/  /9/  in the C 2 position, e.g.  has some kind of unique status which permits it to occur  between the syllable onset C and the nuclear mora. There are other issues regarding /9/, but  46  these will be discussed shortly. This leaves only 6 forms which have consonants other than the  /9/  in C 2 position. These forms are shown in (49).  As they seem clearly to be of  marginal status, no attempt will be made to modify the general, broadly based syllable canons of the language in order to accommodate these apparently anomalous forms. (49)  plIt  ‘priest’ LOAN ‘rustle’ EXPLETWE and ‘snowb(r)ush’ ‘what-happen?’ ‘climb’ ‘snowb(r)ush’ ‘who’  qwxwp  9kén +k’Iw stá 6 swét’  The first three examples above follow from the sonority hierarchy of (38) whereas the last two examples violate it. Interestingly, /+/ and  Is! are the only two fricatives which are  cluster-initial. (These are the few of the examples which do not have a  /9/  in C 2 position.)  Further, as evident from the leftmost column in Table 6, /4’! and Is! are the only class of C, segments which systematically do not occur with a  /9/  in C 2 position.’ 7 The resonants in  Table 6. are not a complete list, but only the ones listed occur with a  /9/  in C 2 position.  Interestingly, resonants do not occur external to an obstruent. The forms in (49) above are questionable on a variety of grounds. The first one is a loan; the second is an onomatopoeic form (which coincidentally has a second synonymous form, the last example above). It can, perhaps, be analysed as [q4p]  (IqwxINcHoATwE)  The third example has a glottalizing effect on the prefix attached to it as shown below (50): (50)  c’-kén(’)-m(’)  ‘how?, why?, what is the matter?’  PREF?-what-happen-MDL  Given the paucity of such forms, no conclusive arguments can be made regarding how they  Thompson and Thompson (1990b) represent the root form for /swêt/ as /wêtf. The nominative Cs—] prefixing 16 /wêt/ and resulting in Cs—wet]. However, the augmentative reduptication on a /CCVC/ form typicaLLy surfaces as Csu—swêt] (see 3.1). Given the existence of I#lcç)/, 17 interpreted as accidental, gaps.  /#q”9/,  i#q’I, /x,x”,  47  the absences of Ik’(’)/, /#q”9I, I1q”I and /I are  should be parsed. A couple of questions arise as a result of the observation that in all the other initial #CC- forms, the second C is the glottal stop,  /9/.  Could the  /9/  actually be the inchoative  morpheme which is infixed between the first consonant and the vowel of a #CVC root? Or could these IC’?! forms be underlyingly represented as IC’! segments? Consider first the inchoative morpheme  [9],  an infix’ which is inserted between the initial consonant and vowel  of a strong root, that is a root which is underlyingly marked for stress as in (51.a).  [9]  is an  independent allomorph of the inchoative, /-p/, which is suffixed to a weak root, that is, a root which is not marked forstress and whose vowel is generally a schwa, as in (5Lb): (51) a. strong Root: q”?é4 q  /qW/  ‘become dull coloured’  94  I?  qW9ajc  ‘become rancid’  q’sp=ékst  ‘hand, arm gets numb’  b. weak Root: numb-INC=hand  Thompson and Thompson (1992) refer to some forms, [C9VC], which appear to be inchoative, but which cannot be conclusively identified as such because they do not have corresponding simplex /CVC/ forms.  Since there is no evidence that there are  polymorphemic [CVC+ INCHOATIVE] I adopt Thompson and Thompson’s analysis of these as unanalyzable monomorphemic roots. An example they provide to support their argument is (52): (52)  [c9iI’9  ‘it bleeds’  */CIIW/  The second question is whether these forms actually have an independent segment  The glottal infix 1  [1  is discussed in  48  glottal stop in C 2 position, or whether they can be analyzed as having an initial glottalized  IC’!. Evidence in support of the independent status of /9/ in such clusters is shown below; an initial C’ can cooccur with (53) c’9o1 k’w’  [9],  as for example (53): ‘tingle, sting’ ‘loss of power’ ‘sink, submerge’  Of further interest and relevance is the fact that CCVC roots may be divided into two distinct classes with respect to the allomorphic realization of the  AUGMENTATIVE.  One small  subset of these forms, as discussed by Thompson and Thompson (1992), marks the augmentative by a non-productive infix /-‘y-/  --->  9 after the root vowel, i.e. by [9ijl  In contrast to forms in (54)a., those in (54)b. form the augmentative by  [CCV-9i-C].  prefixation of a usual reduplicative (C)CVC- template : 20  (54) a. /cés/ c9é[9i]s  ‘come’ ‘AUG form’ ‘mother-in-law’ ‘mother-in-laws’  &céck/ Icé[9i]ck  0ther augmentative infixes inserted after the stressed vowel are I—ce—I and /—z’e—/. 19 I do not discuss these variants any further.  For this analysis  other canonical stem forms which have the augmentative infix I— 20 y—/ are shown below in the following 7 examp1es ‘?similar ‘cross—sex in—law =éstm 9 ss-9=é[9i]stm ‘AUG. /qéck/  ‘older brother  qe[9i]ck  ‘older brothers  Inésl  ‘go ‘they go ‘we go  ne[9i]s né[?i]t  qW4nu[?iJt  ‘they say’ pt.. persons  ki[’i]c-x  ‘p1. persons arrive at a particular pLace’  x”s-x’’esI[?i]t  ‘they waLk, travel’ ‘s.o. travels a lot’  cü[9i]t  c.f.  x”’zs-x”es’it  49  accompany  ‘happy’  b. /c9éx”/ ce-céx”  ‘AUGMENTATIVE  form’  cece9c9éxw  ‘AUGMENTATIVE  form’  /qw9ey/  ‘?withered’ ‘(of many flowers) withered’  qwa9qw9ey  Referring again back to Table 6, it is clear that the glottals occur internal to all fricatives except for  1 segments which systematically &/ and Is!. Interestingly, these two are the only C  have other consonants in C 2 position. As discussed in the  § 2.3.1, the fact that  /9/  and /Q / can occur next to each other in 21  violation of the place contraint provides evidence that the represented by the  PHARYNGEAL  /9/  in N4e kepmx is not 9  place node. The representations of these two stops are  shown below (55):  (55)  Q  9  LARYNGEAL  PLACE  [coNsT.GLorrls]  PHARYNGEAL  A sampling of the 280/715 other root types, i.e. other than C, CV, CVC, CCVC and CVCC, provides five further #CC- initial forms. These ‘other types’ are documented below  in (56). (56)  /caxJ /q’y’ü9e/ /’ 9 !qwtjxa 22 /çqwqwne/  /céck/  ‘spruce’ ‘pull-down’ ‘louse’ ‘a type of duck’ ‘mother-in-law’  One could perhaps assume many of the stems with initial clusters were originally /CVC-!  refers to the set of uvuLar stops. 22 is possibLe that this and the following form are diminutive.  50  forms from which the 1 V has been deleted. 391 documented cases of CVCVC 22 root forms are recorded by Thompson and Thompson; however, I do not discuss them further here. In summary, I conclude that there are very few #CC- root forms and the vast majority of those have /9/in C 2 position. On the basis of this extensive survey of canonical root form, I conclude that the basic syllable structure of 9 N4’e k epmx root morphemes generally allows a maximum of one C in onset position. One systematic deviation from this is the seeming well-formedness of a  in second position. For the very few forms, (49) and (56) which  /9/  have an initial cluster other than #C9, the C at the left margin will, in the present analysis, be considered extraprosodic. These conclusions are exemplified below (57).  (57) a.  6  b.  /I  6 /.u  6  6  /I /  cj.ka9X (47) ‘spruce’ c.  6  rn (38) ‘purple huckleberry’ 6  /L  us <tk> (45) ‘catch fish’ 9  These syllabifications allow a maximun of one C as extraprosodic in initial position and a maximum of two in final position, with the proviso that at least one of them is coronal. 2.3.2 9 N4e k epmx Derived Stems #CC- derived stems N+e9kepmx derived stems exhibit many more #(C)((C)C)- clustering possibilities. The following analysis investigates these complex sets of clusters, using two major assumptions. The first is, as discussed earlier, that the several sonorancy-based groupings are analyzed  Thi List includes aLL voweL—types. 22  51  independently. Secondly, clusters are also differentiated according to how they are derived, that is in terms of Kiparsky’s (1973) distinction between phonologically-derived and morphologically-derived clusters.  They result from consonants being juxtaposed to one  another through morphological concatenation. Common sources of such clusters include the following prefixes /he9-/, the 2nd person possessive marker; or  Is-I, the nominative marker; In-I  /n-/, the locative prefix or 1st person possessive; and /9es-/ which is the stative prefix.  Cluster derivation is schematized below in (58): (58)  Morphologically derived: [C[CC]rt]  Phonologically derived stems are created by the application of some phonological process to a string. All the phonologically derived forms to be considered here entail cases where an unstressed schwa has been deleted from a weak root, resulting in the neighbouring consonants now surfacing adjacent to one another. This type of situation is schematized below (59), where C represents any Ne kepmx consonant: 9  (59)  Phonologically derived: /CC=ekst/  Under the alternative hypothesis that  []  -->[CC=ekst]  in weak roots is epenthetic, rather than part  of the underlying root, the generalizations which I described here as pertaining to “phonologically derived” stems would in fact be root structure constraints. Interestingly, where unstressed V deletion might be applicable to a monomorphemic stem of the canonical form #C9VC, the results are either a bi- or tn-consonantal cluster with the  [9]  generally moving after the following vowel as shown in (60)a. below, whereas in  (60)b. the  /9/  (60) a.  remains in the original position: .fc’9o1  n-c’ól’=ñs  ‘have one’s eyes sting’  LOC-acrid-eye  There are two others, but I do not discuss them here since they are not as common as the ones listed 23 above.  52  fc’9a1  ‘hand tingles (from cold)’  n-c’á9=kst LOC-tingle=hand  b.  s-k’9ák  Ik”ek  ‘arrival’  NOM-arrive  %/k9éx  s-k9ex=I’t  ‘male youth’  NOM-male youth=child  Considering first phonologically-derived clusters by the rule given in (59), the following Tables 7-13 document according to manner of articulation, the #CC sequences, which are attested in N+e9kepmx. On the basis of these tabulations, the present analysis identifies a number of significant co-occurrence restrictions on place and sonorancy. Note that these tables represent only #CC and not the #CCC-clusters which will be discussed later in  § Table 7 shows which stops and affricates can cluster with each other in phonologically derived contexts. c  c’  k  +  + +  +  2 \ 1 C C  p  p’  p  -  -  p’  -  -  -  -  -  -  -  -  -  -  -  -  -  C C’  k  + + +  q q’ qW  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  +  -  -  -  -  -  -  -  -  + +  -  -  qW  -  -  +  k ” 4 k’s”  t +  -  -  + + -  -  + +  -  + -  -  + + + + -  + -  -  + + + +  k’ -  -  -  -  -  -  k” + -  + + + +  k” -  -  -  + -  +  q  + + + + + +  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  + -  + + +  -  qW  qW  q’ -  -  -  + -  +  -  + + -  +  -  +  -  +  h -  -  -  -  +  -  +  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  +  -  +  + -  +  -  -  “ (+)24  (+) + + + + + + + +  (+) + + +  Table 7. Cooccurrence Restrictions in #CC[COfltI clusters of derived stems As expected, obstruent consonant clusters occur where they do not occur as initial clusters in underlying root forms. In general, the vowels in ICC! roots undergo -de1etion The bracketed forms are inchoativized forms. 24  53  when unstressed, thereby creating #CC- stems. Again, as discussed in  § 2.3.1, there are no instances where obstruent clusters in stems  occur with the same place feature. The constraint previously formulated as (31) can now be replaced by a more generalized revised Place Feature Constraint shown below (60) which outlaws homorganic stop clusters: (60) N+e9kepmx Morpheme Structure Constraint on Place Features:  [  *  [-cont] Place  [-cont] Place  [a]  [al  Given the strength of this constraint against adjacent homorganic clusters, it is significant to note that p’, a  [LABIAL]  consonant, cooccurs with  a rounded consonant, as shown in the  following forms (61). (61)  /pqW/  pqwekst  ‘have rash on one’s hands’  skin eruptions=hand  pqw=use9  ‘(of potato) scabby’  skin eruptions=round obj.  The fact that a  [LABIAL]  consonant !p’/ can cooccur with a rounded consonant may be  considered to be strong evidence that rounding in 9 N4-e k epmx is not represented either by or under the  [LABIAL]  node, since this would violate the constraint in (60). As shown in Table  7, rounded velars as C 1 only cooccur with coronals or a non-rounded uvular as C 2 (viz. k’’)’, k”c, k”’c’, k’’q), and rounded uvulars as C 1 only cooccur with coronals (e.g. qwt qwc). Furthermore, rounded velars and uvulars in C 2 position only cooccur with labials or coronals as C . These cooccurrence restrictions on rounded consonants will follow directly from the 1 postulated constraint in (60) if the true labial consonants !p, p’, m! are represented by a [LABIAL]  place specification in a feature geometric model as in (62)a. below, whereas rounded  obstruents will be represented with an independent distinctive feature [round] under the  54  [DoRsAL]  node, as in (62):  (62)  b. /kv/ Place  a. /p/ Place [LABIAL]  c.  /qW/  ,,,/ .  [DORSAL]  [round]  fce  [ooi.su] [PIR]  [round]  Interestingly, glottalized obstruents freely cooccur as shown below (63): (63)  p’c’I ‘k” c’q’ k’c’ k”’q’c’-  k”c’q’k’  This freedom of cooccurrence follows directly from the fact that [glottalization] is not a place feature and, therefore, not constrained by (60); rather it is a dependent of the LARYNGEAL node  higher in the feature geometry representation.  The following chart, Table 8, represents #C1[+cont]C2[cont]]stem positions. \C 1 C 2 + S  p  + +  p’  + -  t + +  k’  -  +  c +  c’  k  +  + +  -  -  -  -  k’ + -  k” -  +  qW  k”’  q  q’  + +  + +  + +  + +  qW  c  +  + +  -  S C  z z, x  xw x  -  + + + -  -  + + +  -  + + +  -  -  +  + + -  + + -  + + + + -  -  -  + + + -  -  + -  +  -  + -  + -  +  + -  -  -  -  -  + +  +  -  -  -  -  -  -  -  -  -  -  +  +  -  -  -  -  -  -  -  -  -  -  -  + -  +  + + + + +  Table 8. Cooccurrence Restrictions in #Cl[+COflL]C-clusters, of stems: The following examples (64) have rounded velar fricatives adjacent to rounded velar glottalized stops. The Morpheme Structure Constraint appears to need further revisions since the following examples have segments which have the same place of articulation. 55  (64)  /xk’7  xwkvp  ‘dry gulley with gravel, developed from  crevice-INCHOATIVE  earlier erosion  Similarily, rounded velars will also cooccur with labial stops, both plain and glottalized, as:  (65)  /x’p/  xvpóm  ‘pull straight out of ground’  rise straight up-MDL  x’p-Iyx  ‘rise, get up’  rise straight up-AUT  IxpI  x”p’-m  ‘dig up, pull’  rip out-MDL  /xq”7  x”q”éke9-s  ‘prune s.t. (trees) from s.t.’  lessen = hand-3  Moreover, uvular fricatives will cooccur with uvular stops as (64), although they share a common place feature. The status of fricatives preceding stops also raises questions in regard to their independent versus merged position in the sonority hierarchy (or their extrametricality) as discussed earlier. The following are examples which support the fact that identity of place is not a crucial factor between fricatives and stops as it is with stops. Further evidence and discussion will precede a revision of the Morpheme Structure Constraint. (66)  /qI  ‘tight, snug’  q-p wedge-INC  q=etxv  ‘(of animal) short haired’  short fur=fur?  Uvular fricatives also occur with rounded uvular stops as shown below, but only with a reduplicated form: (67)  ‘several snore’ AUG-snore-?-nose  Further Cj[+COflt] segments cooccur with C kepmx #CC- clusters 9 [+CQflt] segments in N4-e 2 in stems, as shown in Table 9. The segments which are identical are given neither a  (-)  designation.  56  (+) nor  + S  +  S  Z  Z’  -  -  -  -  +  -  +  X  Xx”  + +  + + -  ç  -  z x xw  -  -  +  -  -  -  -  + + +  c çW  +  +  + +  -  + -  +  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  + +  +  -  +  +  ç  X  -  +  + + + -  -  -  + -  -  + -  jW ‘  -  + -  -  -  -  +  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  +  +  +  -  -  +  -  -  -  +  -  -  -  -  -  -  -  -  h  -  -  +  -y  -  Table 9. Cooccurrence Restrictions in #Cl[+COflt]C2[+COflt]- of derived stems. It should be noted that !h! and fyi do not cluster with any other fricative when in C 1 position. The Morpheme Structure Constraint, in general, seems to be in effect. Note once again that a difference in [continuant] can overide identity of place, as // and Ic! differ only by the [-continuant] feature of the latter segment. Do fricatives comply with the Morpheme Structure Constraint (60)? The following examples (68) have fricative-fricative onset clusters: (68) /c/  cs-óm  ‘roast-drying of meat over fire’  thy roast-MDL  /sz9/  sz’=éks-tn’  ‘toy’  toy=hand-INST  !zsI  zS=éwt  ‘in the wilderness, etc.’  bend-low=? wçwp !x!  ‘cough excessively’  cough-INC  These examples show that the Morpheme Structure Constraint (60) does not apply to adjacent fricatives in stems. There are also morphologically derived stems where #CCC- clusters will have two  57  fricatives, +- and s- prefixes , preceding an obstruent, as in (69): 25 (69) +-s-cm=éyt-kt  ‘our children’  1pIPOSS-NOM-small=offspring-2  +n-s-cm=éy’t  ‘my children’  1sgPOSS-NOM-small=offspring  An exceptional stem form which has #CCC, that is no vocalic segment that is [+sonorant] or [nasal], is (69)26. (69)  x”st  ‘EXPLETIVE, AN ENDEARMENT TERM USED BY ELDERS TO YOUNGER PEOPLE’  Expletives as refered to by Thompson and Thompson are forms such as (69) which does not necessarily mean the same as in English. As to be expected, derived stems shown below (70) can also have maximally -CCCCC clusters: (70) cu4’=qs-x-t-iyxs  ‘they point a gun at him’  disconnect=point-BENE-TRANS-3-3  cü+=qs-x-te  ‘point the gun at him’  disconnect=point..BENE-IMP  As a speaker, I intuitively would syllabify this form as (71): (71) cu+qs.xtiyxs 27 ?cu.qs.xtiyxs *cu.qsx.tiyxs  ‘they point a gun at him’  A templatic representation of the above forms are shown below: (72) Syllabification representation of -CCCC- clusters: a.6 6  i  cu+qs xtiyxs  lnterestingly, note that this particular form ends in a —GCCC# cluster, s-cm=éyt-kt. 25 This is an expletive form Thompson and Thompson (1990:417) usually said three times by an elderly person 26 to a younger person, especially a child, to show affection. current theoretical frameworks do not prosodically represent this form as I do, the representation of the former form does not intuitively seem appropriate. However, the following form is intuitively incorrect. Nonetheless, this form requires a prosodic representation which is shown in (72)c.  58  b.6  6  cu4-  II  cu+  ‘N  qs  xtiyxs  LJIJI  6 JIJI  c. 6 JIJI  6  ii  qsx  •11  tiy  JJL  x  S  The medial syllable in the representation in (72.b) is exactly what Shaw (1993) motivates as the structure of minor syllables in Semai, Temiar, and Kammu (Mon Khmer) and in Berber. Such syllables function as monomoraic and non-nuclear. The representation in (76.c) follows Bagemihl’s (1991) hypothesis that only “Simple Syllables”, i.e. maximally CRVC, are parsed; other residual segments remain licensed only at the moraic level. Such an appeal to moraic licensing is not otherwise a crucial factor in Ne kepmx. 9  Other than the [q] in this  particular form all the other stranded segments are linked to a fricatives.  It may be  necessary to conclude that voiceless fricatives segments are extraordinary. The following Table 10 is a representation of 1 OI I #C 2 C OI 1 clusters of stems in kepmx. 9 Ne  Recall these clusters are phonologically-derived forms rather than  morphologically-derived forms, and their examples adhere to the sonority hierarchy. A question to consider is whether these sequences follow the Morpheme Structure Constraint  (50) I presented earlier § 2.3.1. and revised in § 2.3.2.  59  /C 1 C 2  +  s  s  p p’  + + + + + + + + + + +  + +  +  C C’  k k’ k”  q q’  -  + + + + + + + + + + +  -  qW  +  qW -  c  x  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  “  x  + + +  -  + +  -  + + + + + +  + +  -  +  +  + +  -  -  x”  -  -  + +  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  +  -  +  +  -  -  -  -  -  -  -  +  -  -  -  -  -  + +  h  -  -  -  -  -  -  Table 10. Cooccurrence Restrictions in  + -  -  -  -  -  -  [+COflt] 2 #Cl[COfltIC  Clusters  As discussed in an earlier § 2.3.1 (37), fricatives and stops with the same place features can co-occur, thus, no additional revisions are necessary to the constraint in (60). Note, however that although they do cooccur they do so less frequently. The following Table 11 provides coocurrence restrictions for #C[sofl]C+coflt]}stems clusters. Contrary to the data provided above, these forms violate the sonority hierarchy, since nasals are more sonorous than fricatives. Also, very few glottalized segments cooccur with other segments in stem-initial positions. +  m II  1 y  z w  + -  -  + -  -  -  çW  + +  s + -  -  -  + + -  -  +  s  c  -  -  -  -  + -  +  -  -  -  x + + + -  +  x” -  +  -  -  -  -  -  -  +  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  + + + -  + -  + -  + -  -  + -  -  +  Table 11. Coocurrence Restrictions in #C[+cont]C+contllstems clusters The above forms show no clear place feature constraint. Segments which have [roundj features also cooccur. No obvious explanation for the violation of the sonority constraint is evident. This 60  could perhaps be due to the special status resonants have, that is, they may be extrametrical since this does not happen with the segments in initial position in root forms. /C 1 C 2  m  n y  z w  k  +  +  + +  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  + -  -  çW  c’  k’ +  p’  + +  1  c  t +  p  -  -  -  -  -  + + -  -  + + -  -  -  +  -  -  -  +  +  -  -  -  -  -  -  -  -  +  +  +  -  -  -  -  k’  k” +  -  k” -  -  -  -  -  -  -  -  -  +  + +  -  + + + -  qW q  q’  +  +  -  + +  -  -  -  -  -  -  -  -  -  -  -  -  -  +  -  -  -  -  -  -  + +  qW -  -  + +  -  -  -  -  -  -  -  -  -  -  -  -  9 -  (+) (+) (+) + + -  + +  Table 12. Coocurrence Restrictions in #C[+cofltCcoflt]]stems clusters Interestingly, some of the resonants do not cooccur with stops when both have the same place feature, but other resonants can cooccur with resonants with similar features. Segments which can cooccur have the following features: example, the [y] can cooccur with [s], which is marked environment involves velars which are marked for [DoRs?.t],  [+coRoNAL], [+RouND].  [+c0R0NAL].  [+R0uND].  For  Another cooccurrence  Segments with the feature  [‘y] and [k, k’], can also cooccur. On the other hand, this cooccurrence does not  occur with uvular resonants and pharyngeals. Segments which do not cooccur are those which are marked for i.e.  [LABIAL],  [pj and [m], or those marked for the feature  [PHARYNGEAL],  uvular resonants and stops.  There is only one glottalized resonant, [y’], which occurs in the C 1 position of N4e9kepmx stems. Examples are shown below (72): (72)  y’h-ém s-ih-éms y’-wi9xy’i-ép  ‘do s.t. well’ ‘good?’ ‘get better, improve’ ‘get sick again, relapse’  The following Table 13. represents the cooccurrences of resonants in C 1 and C 2 position.  61  2 / 1 C C  m  m  (+)-  +  II  1 y z w -y  -  + -  +  -  çW  -  m’  -  + -  + +  -  n + +  n’ -  1 +  -  -  -  -  -  -  -  -  +  + + -  + +  y -  -  -  -  -  -  -  -  -  + + +  1’  + + -  -  + -  -  -  +  y’  -  z +  +  -  -  -  -  z’  -  -  -  -  -  -  -  -  -  +  -  w  w’  -y  -  -  -  -  -  -  -  -  -  + -  -  -  -  -  + + -  + -  + + -  -  -  -  -  +  -  +  -  -  -  -  -  -  -  c  c’  +  -  -  -  -  + + + + -  c”’ -  -  -  -  + +  -  + +  ç.’W  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  Table 13. Coocurrence Restrictions in #C[son]C[sofl]}stems clusters It can be seen that segments which have identical place features do not cooccur, as in (73).  (73)  zy-ém  ‘make channels for water to flow e.g. irrigation’  In conclusion, N4-e9kepmx segments do not cooccur with segments which have identical place features except for adjacent [continuant] segments. Examples below show (74): (74)  nh-éh ch-ém shéw s-zh-éw’s  ‘named’ ‘s.t. is put away’ ‘yawn’ ‘second child’  2.4 Prosodic Categories: Minimal Word and Foot 2.4.1 N+e9kepmx Minimal Word In order to motivate an analysis of the Me’)kepmx minimal word as a bimoraic foot, three arguments are presented. I use some of the same arguments McCarthy & Prince (1990) use for the Arabic minimal word, namely smallest stems, truncated vocatives and borrowed loan words, each of which involves a minimal bimoraic foot, as in (75).  62  (75) Minimal Bimoraic Foot: a. Foot 6  Foot 66  L  2.4.2 Lexical-smallest stem kepmx smallest-stem monosyllabic lexical items that are 9 There are very few N4’e bimoraic. They are shown in (76): (76)  cuk”’ 9 cu he”  ‘finished’ ‘(continually) trying’ ‘high’  Most bimoraic lexical stems have a consonant which affixes to the CVC bimoraic minimal word resulting in CVCC or CCVC canonical structures. One is the nominative prefix, s-, which attaches to the root, as in for example [s-naq”9 ‘a stolen object’. The other is the inchoative which is either an infix roots (see  [9],  with strong roots, or a suffix [-()p] with weak  § 1.3.1 (18) and 2.3.2 (51)). The deletion of the vowel in the weak root results  in the following forms: i) strong roots, [m[9]e’I]/ ‘become rested, and ii) the weak root: [mt p] ‘become cramped from sitting too long’. However, a stem such as  [CjtXv]  ‘house’ is a CVCC form which can still be analyzed  as a bimoraic root, plus an additional segment. Examples from English loan words also have bimoraic forms as in (77): (77) English loan tea iron cat soap  N+e”kepmxized form tiy “ayn pus sup  In summary, the N4’e”kepmx smallest stems must be one of the following: CVC,  CCVC, CVCC, which can all be represented bimoraically. N4-e”kepmx smallest stems can be represented as in (78): 63  (78) Mm word = [pp.jó CVC a. 6  C  b.  CCVC  c.  CVCC  6  VC  CC  VC  1  C  t  VC<C>  2.4.3 N4-e9kepmx Truncated Vocatives N+e9kepmx truncated vocatives, as in many other languages, are minimal words consisting of monosyllabic or disyllabic bimoraic feet. Examples of N+e kepmx names (79) 9 and kin terms (80) which have been truncated to bimoraic forms are shown below: (79) Personal names: C’u’sinek Lapina Suli Cece9 9 Sinci  Vocative forms: C’u9 Lapi Suli Cec Si(y)n  (80) Kin terms: 9 skixza Yeye’  Vocative forms: Kix Yey  ‘mother’ ‘grandmother’  There is another kinship form, 9 ep, which does not have an equivalent English gloss, but it is used to address an older sibling or cousin, male or female.  Following the  N4-e k 9 epmx minimal word constraint, e 9 p is a bimoraic monosyllabic minimal word. Other kinship terms are not truncated to bimoraic minimal words and are therefore not discussed. These truncated vocative forms confirm that the minimum word in 9 N+e k epmx is a bimoraic foot. Interestingly, forms such as [lapi] and [suli] show that this can be satisfied by a sequence of two monomoraic syllables, as well as by a single bimoraic syllable as illustrated in (78). In chapter 3 the descriptive analyses of the five types of 9 Ne k epmx reduplication processes will be discussed and presented.  64  Chapter 3. N+e9kepmx Reduplication: Basic Types and Previous Analyses The N4e9kepmx language has five types of morphological reduplication processes. Three of these reduplication processes, augmentative, diminutive, and out-of-control, are very productive, whereas the other two types, characteristic and affective, are not productive (Thompson and Thompson: 1992). In the first section of this chapter a description of each of these processes will be given. The reduplication processes may be associated with more than one meaning; these will be presented along with the data. Different representations of the various canonical forms and a description of each of the independent morphological processes will be presented. The specffic target and base of each process which will be discussed, although a more detailed explanation of the prosodic and morphological bases will be presented in Chapter 5. Lastly, a discussion of each copied template will be presented, stating whether the template either prefixes or suffixes its targeted base. The second section, 3.2, will consist of a discussion on the interaction of various reduplication processes. One occurrence of each reduplication type is the usual for a stem, but application of two or more different processes can be common; In these cases, the processes not only occur independently of each other, but can be templatically adjacent within a stem. Each N+e9kepmx reduplication type undergoes phonological processes that are ‘across the-board’ in that they also occur in other phonological environments. These processes will be discussed in more detail in chapter 5. This chapter, a descriptive analysis of each of the five types of N+e9kepmx reduplication, refers to the different canonical patterns of the copied base/form and the different canonical targets of the copy. Presented in the first section, 3.1., are basic types 65  in the: augmentative, diminutive and out-of-control, characteristic and affective. In multiple co-occurrences of N4’e9kepmx reduplication will be discussed, and in  §  3.2.  § 3.3 previous  analyses of Me9kepmx reduplication will be presented.  3.1 Basic Types of Reduplication 3.1.1 N4e9kepmxcin Augmentative Reduplication: CVC-/CV Ne9kepmx augmentative  (AuG)  reduplication is used to express the collective,  intensifier and repetitive forms. Thompson and Thompson (1992) state these forms are part of a general notion of “augmented”, interpreted differently according to stem-type, (e.g nominatives, statives and pronominals). Other pluralizing processes in N+e kepmx, these 9 will be discussed briefly in this section, this thesis discusses the most common form which copies the initial CV(C) of the root, and prefixes it to the stem as shown below in (1): (1) {(PREFIx)ICV(C) + (suFFTx(Es))JsThM --> [(PREFIx)CV(C)-%fC7(C)..(surnx(Es))JSThM aug  In general, as represented by the brackets the copied form can either be a CVC- or a CV-  prefix. Shown below are CVC- examples where (2) a. is a collective form, (2) b. is an intensifier form, and (2) c. is a repetition of action. (2) a. CVC- collective forms: k’n’- k’én’i fk’én’i  ‘ears’  AUG-ear  %fai  +c-+ái’  ‘They are cold.’  AUG-cold  b. Intensifier: %fcék  ck-cék  ‘very cool’  Thompson and Thompson (1992) and other SaLishanists note that there is a transitive and intransitive 1 distinction between the forms. Here cold is an intransitive, therefore, subject is plural; whereas a transitive form would pluraLize the object. H. Davis (p.c.) states this is actually an ergative/absoLutive distinction.  66  AUG-cool  */XWéS  ‘something is greasy’ AUG-grease-t-3-3  c. Repetitive: ‘s.o. twists strands together’ AUG-twist-?  %fpu9  p’u9-p’u9-e-cñt  ‘continuously flatulating’  AUG-fiatulate-RESULTIVE-REFL  cu9-cu9=qmn  ‘to hit s.o. in the head’  AUG-punch=head  CVC- forms can also have one of a number of prefixes, including and  STATrVE;  NOMINATIVE, LOCATIVE,  in addition, compound forms can be attached before the stems, resulting in the  following (3): (3) Collective: s-k’m-Ic’mk= éy’t  ‘nieces’  NOM-AUG-niece=agent  n-k’ot-k’tni-m’-tn  ‘fishing places in creek’  LOC-AUG-rod-fish-MDL-INST  9 e s-qX’-qiX’  ‘having many scars’  STAT-AUG-scar  ‘lots of dirt on them’ many-LIG-AUG-dirt  The forms in (4) are also CVC- augmentatives. However, in these forms either the copied template has different consonants than the base, or the copied vowel has not been reduced to a schwa, in contrast to examples in (3) above. (4) a. collective:  ‘it is cloudy’  3 q”i9-q”üy’  AUG-cloud  p’ac-p’ac-t  ‘burned in several places (skin)’  AUG-burn-IM  k’l’-k’l’-mmn’  ‘scissors’  AUG-cut-sheet-INST  Another reduplicative form for this root is c’up-’up-ix grow twisted and cLimbing. 2 This particular form onLy occurs in the reduplicative form, that is it does not occur in a singular form. 3 For example, *q’üy is not a singular lexical form.  67  k’l-k’l-t-és  ‘cut s.t. up into many pieces’  AUG-cut-TRANS-?  The first form can be accounted for by a CVC- copy in which: (i) the [(ij has reduced to a schwa, then (ii) the [y’] has become [j9]. In the second form the vowel retains its original quality since it is followed by a pharyngeal which has a lowering effect. The last two examples associated with the retracted schwa in the root, but the  {+PHA YNGEAL]  feature is  deleted or not copied, while the retracted surfaces as a glottalized 1’. I can nOt account for this at this time, though \fk’ is possibly a glottalizing root, since the [n], the final C of the INSTRUMENT  suffix, has been glottalized. Although this seems to only occur with this form,  other forms have neither the glottalized [1’] nor the retracted  [1] as shown in the last example  above. Other interesting forms below are associated with a repetitious meaning. They have an -e’ suffix in between the copied form and the target form, as shown below. The form also has no vowel reduction of the copied form, though the last example has the last vowel of the root deleted.  (5)  %J1woy’  cwoy[e91cwoyt  ‘doze off and on’  AUG-?-sleep-IM  4 ciy-e9-cIy=kst  ‘sequences of five repeated’  AUG-REPETIVE-open=hand  -ci=ks=qin-m 9 n-ciy-e  ‘five times five’  LOC-AUG-REPETIVE-open?=hand=head-MIDDLE  The other main canonical type of N+e kepmx reduplication is CV-, as shown below: 9 (6) a.  CItXW  ci-cItx”’  ‘houses’  AUG-house  fch  c3-ch-e  ‘s.ts. put away’  AUG-arrange-?MDL  b.  Ic’iy-t  k’i-X’Iy-t  ‘be still, motionless’  AUG-still  other augmented forms of this root have the foliowing prefix, ci—. 4  68  sqThqwy = ép  ‘strawberries’  NOM-AUG-npe=bottom?  This alternation of the copied domain shown in the (6.a) examples seem to be an Obligatory Contour Principle (OCP) effect as stated by McCarthy (1986): “identical adjacent segments are prohibited”. This principle may account for some of the CV- forms since these particular forms in most cases have near identical segments, e.g. cltx” --> cicitx”. If the resulting form were *cit..ciw, [c] and [t], both have coronal features which would violate the OCP assuming the reduplication process occurs, then tier conflation occurs. (McCarthy, 1986)  Violation of the constraint would result in near identical adjacent segments.  However, there no violation with morphologically derived forms which have -C][c- as discussed Chapter 2.  The examples in (6.b) have vowel deletion first and then glide  vocalization of the /y/-> {i} as discussed earlier. The forms below considered bisyllabic root 5 (Thompson and Thompson:1990) have vocalization of the glide as show in (7). (7)  yéxv  ‘teeth’  xiyéxv  AUG-tooth  Whereas, the form below would also be a OCP violation if a CVC were copied. (8)  Jshéw  s-shéw  ‘they yawn’  AUG-yawn  The following forms either have a glottal stop in the base or the stem form and in the copied form (9): (9)  ..fcex  ce-céx”  ‘glad, happy’  AUG-?happy  ci9c9I1v  ‘bleed in several places’  AUG-bleed  fc’z  c’o 9 c’óz  ‘be rather dark, a little dark’ 6  There are many of these bisyl.l.abic roots, therefore there can potentiaLLy be many, many examples of this 5 sort. T&T state that this has an unexpected meaning. 6  69  AUG-dark  ‘get very wet, several get wet’  c’a9-c”áq” AUG-wet  .fc’é  n-c’e9-c’e94-=qmn  ‘have one’s head get cold’  LOC-AUG-chill=’head  These forms unlike the preceding forms do not copy the root, either copy a #CC- of the stem and insert a vowel, or they metathesize. The vowel quality of both the copy and the base must be conditioned by the laryngeal. The examples below (10) and (11) are further examples of a CV- reduplication, these forms are similar to the CVC-wopy, a prefix is attached before the copied form or the root of the singular form, however the C 2 of the root is not copied. (10) a. CV- Collective: fqáy n-qi-qáy-ix-kt  ‘we all swim’  LOC-AUG-move back and forth-AUT-2p ..Jk’ey  fcw  s-ICi-k’ey NOM-AUG-dance  ‘dances’  “es-cu-cuw  ‘houses built here and there’  STAT-AUG-build  ..fcüté4  s-cu-cutél-  ‘sons-in-law’  STAT-AUG-son-in-law fçWy  ‘burned in several places (generic)’ sTAT-AUG-burn  b. CV- Repetitive: %/kéy  n-key-rn  ‘he follows’  n-ki-kéy-m  ‘he keeps following’  LOC-AUG-follow-MDL  c. CV- Intensifier:  NO EXAMPLES AT PRESENT  Again, prefixes attached to the stem are not copied as shown above (8), that is, prefixes do not become reduplicated, but the surface forms only show the first CV- of the root. The above forms delete the copied vowels and the glides /y! and 7w! have vocalized to [i] and [u] respectively. The vowel alternations, of the !e/ ->  70  [} and [i] are phonologically predictable.  The former, schwa, is a vowel reduction of either the stressed and unstressed copied vowel, and the latter is glide vocalization. (Thompson and Thompson: 1992). Also, note however in some cases as in (11), the copied form has a reduced vowel, but the target vowel has not been deleted or reduced. Additional morphemes have, also, been suffixed. This can be accounted for by stress reassigning cyclically, and the /e/ surfaces unstressed and does not reduce. qw4..qwen..cut  (11)  ‘invite people (plural) to accompany’  AUG-invite-DIRECTIVE-REFL  There are several types of idiosyncracies illustrated by the augmentative reduplication. Stem forms have been lexicalized as in (12)a. Another idiosyncracy is shown below in (12) where  AUGMENTATIVE  reduplication appears to occur twice as shown in (12)b. The singular  form ‘k’’IX’p’ does not occur. This may be accounted for by cyclic application or plural reduplication of an inherently reduplicated form . Also in (12)c. the both the target and the 7 base do not have a vowel. (12)  a.  JhIy’ s-hi-hI’ s-hi-hi-hi9  ‘father-in-law’ ‘male-in-laws’  STAT-AUG-AUG-father-in-law  b.  ‘fleas’ AUG-AUG-fleas  C.  .fçm “es-cm-cm-ále’=xn  ‘feet are very dirty’  Other cases where there is only an the reduplicated forms only occur in the augmentative form, and it does not have a non-reduplicated lexical form (13): (13)  kc-kóc  ‘she is a willing worker’  AUG-willing  sx-sx-t  ‘he makes a mistake’  pc. M.D. Kirikade 7  71  AUG-mistake-IM  Other cases include other forms other than prefixation of the reduplicative affix, CVC- and CV-. Forms appear similar to the  8 CHARACTERISTIC  reduplication pattern, for example (14).  The following data are anomalous in two ways. First, if analyzed as 6.qi-prefixation (CVC-) then the C . That is, in (a) [k’] 2 /second 6 slot is not a complete faithful copy of the base C 2 is copied as  [9]  without the oral cavity features; in (b) the complex diphthong [ey] surfaces  in the (stressed) prefix, but is reduced to the simplex nucleus [i] in the (now unstressed) base. Secondly, it is normally the case with prefixation that stress stays on the base, or shifts rightward to a subsequent morpheme. In these data, however, stress shifts leftward onto the reduplicated prefix.  Alternatively, if these data were analyzed as suffixation of a  reduplicative representation of the augmentative, then stress would appropriatelly remain on the (word-initial) base, but it would be the only subset of cases where augmentative was realized by suffixation. (14) áI’-m  9 á-a1’-m’  ‘very steep’  AUG-high-MIDDLE  kéyx  kéy-kix  ‘hands’  AUG-hand  The analysis I adopt for these anomalous form is similar to the ‘regular’ forms. However, stress has shifted from the base to the copied form. Other non-reduplicated augmentative forms include: (15)  i)  ii)  ablaut forms e.g. /sok””/ ‘break, smash, crack (sg.)’, /sIkv/ ‘break, smash, crack (p1)’., and suppletive forms. e.g. mIce9q!  ‘  sit (sg.)’ and /4áq/ ‘sit (p1.)’.  This type of redupLication wiLl be discussed Later in this chapter. 8 T&T comment that this form is unusual. 9  72  Other forms, as mentioned in Chapter 2, are independent morphemes which are suffixed after the stressed vowel. In conclusion, the Nle9kepmx reduplicative augmentative pattern is generally a CVC prefixed before the root.  This pattern semantically referrs to plurals, intensifiers, and  collective forms. There are idiosyncrasies, although some are not necessarily relevant to this discussion. However, some are explainable as discussed in the previous sections.  3.1.2 Diminutive Reduplication: [-CVI/[-C] N+e’kepmx stems undergo a diminutive reduplication process which refers to a smaller item, a small amount or a reduced amount of the base lexical form (Thompson and Thompson, 1992).  This process is also used for references to children’s speech or  belongings. The stem can be one of the following: simple lexical forms, lexically-plural forms, lexically-diminutivized forms. The diminutive affix, -CV or -C, suffixes the consonant and a stressed vowel. The vowel in the diminutive reduplication affix is deleted in the environment of a consonant, that is obstruents and resonants. The brackets in the following examples and subsequent examples in this section designate the reduplicative affix.: (16)  smIvi: DIMINUTIVE:  fqemüt qemü[m’jt’°  ‘hat’ ‘small hat’  STEM:  fk”áf’e  ‘box’ ‘small box’  sTEM:  s-fnüye9  ‘money’  DIMINUTIVE:  s-nü[n’]ye9  ‘a  STEM:  s-fm-yew’  DIMINUTIVE:  s-myé[y]u9  ‘lynx’ ‘baby iynx’  DIMINUTIVE:  small  amount of money’  This form and some of the fo1owing diminitivized forms have a glottalizing effect on most of the 10 resonants. There will be further discussion of this in a subsequent section.  73  Of interest is the fact noted that in many cases the resonant which follows the stressed vowel of the target base is glottalized.  In other cases the resonants which follow the copied  template are glottalized. The above examples have glottalization of the [m, n] but not the [y, cj which Thompson and Thompson (1992) also claim to behave as a syllabic resonant. The following (17), an extension of (16), show the intermediary forms prior to vowel deletion. (17)  sTEM:  ‘hat’  fqemüt qemü[m’u]t ’ 1 qemñ[m’]t  ‘small hat’  %Jk”’áx”e k’”á[k’a]x”e 0: k””á{k’]x”’e  ‘box’ ‘small box’ ‘small box’  DIMINUTIVE: sTEM:  DIMINUTIVE:  VOWEL  ->  sTEM:  s-nüye9  ‘money’  s-nñ[n’u]ye9  small  DIMINUTIVE:  s-nü[n’]ye9  ‘a  sTEM:  s-fmyew’  ‘lynx’  DIMINUTIVE: VOWEL ->  0:  amount of money’  s-m’ye[ye]w’  ‘baby  9 s-m-ye[y]u  lynx’  On the contrary, the copied vowel is not deleted if the the  stem (base form) or  the  template which is reduplicated ends in a glottal stop as shown below (18): (18)  smrvl:  s-fpozu9  DIMINUTIVE:  s-pzu[-zu]9  ‘animal’  RESONANT GL0rFALIzED:  ‘bird (small animal)’  s-pzu[z’u]9  Motivation for these derive from the reduplicative behavior of stems which have a laryngeal in the C 2 position of the stem. And again note, the copied vowel is not deleted in the reduplicated forms in (19):  This form and some of the fol.Lowing diminitivized forms have a glottaLizing effect on most of the 11 resonants. There will be further discussion of this in a subsequent section.  74  (19) a. b. s-pzü 9 --> c. es-c’é -->  The forms with the  9  xé-xes-pzü-z’u-9 9es-c’6-c’e9  ‘a little high’ ‘small bird’ ‘s.t. small laid out’ out’  in the C 2 position are fully regular. Compare, however, the following  where there is no glottal and the augmented stem has not made any compensations at the surface form (20): (20)  2 c’y’é-y”  ‘little basket’  Another phonological process which occurs is with phonemic forms which have a /y(’)/ or /w(’)/ after the copied target. The surface forms result in [i(9)] and [u(9)] respectively as in (21). (21)  9es-c”iy  ‘it is burned’ ‘it is burned a little’  Iwóyt çwÔcwj9t  ‘she sleeps’ ‘she takes a nap’  s-néw’t nu-néw’-t nu-né-n-u9-t  ‘wind’ ‘there is a wind blowing’ ‘there is a little breeze blowing’  The following shows a derivation of the above forms: (22)  ROOT/sTEM:  /1wOyt/  AUG  n/a  DIM: VOWEL DELETION:  cwOwyt  /nu9new’t/ nu9-new’t nu9ne-ne-w’t nu9nenw’t  MORAFICATION:  1q /U 0 çW  R  ftL  çw  9 he nu  nu  9  t  Once the DIM reduplication process has been applied, the copied vowel is deleted. This results in a ‘syllable’ without a nucleus, thereby, necessitating resyllabification of the final  cLaim that the root for this form is Ic’y’éh ‘basket’.  75  syllable. The  !y’!  -> [i9]  or 1w’! -> [u9j which now both have a nucleus.  Other examples of diminutivization occur with English loans which have been N+e9kepmx-ized as in (23): (23) a. b. c.  k’’eü 9épls káh  ‘pig’ ‘apple’ ‘car’  k”eü[u]9 ‘little pig’ ‘)é[9jpls ‘little apple(s)’ ká{ka]h ‘little car’ (530) ká{ka]9  (23.a) stem does not have a consonant in final position, but interestingly the diminutivized form has a glottal stop in final position. Therefore, it provides further evidence that the diminutive template is really -CV. The following forms are not possible *kwesüs or *kwesüsu. The English loan [kah], provides fwther evidence the [h} patterns with the  [9],  a laryngeal.  The diminutive does not necessarily need to fall on the stressed vowel of the root; it can fall on a strong suffix which has a stressed vowel on a stem form as in (24): (24)  [CVC(C) +VCAFFIX(ES)[+strongl]stem --> [CVC(C) + VCnCAFFIX(ES)[+strongl]stem  For example in (25) =áns is a ‘strong’ or an accented lexical suffix:  (25) ax=áns +a9t=á[c}n’s  ‘(a person) eats’ ‘a little person eats’  food = teeth  ese 9 sx” = u[xv1i9t  ‘baby small seals’  STAT-seal=offspring-[DIM]  9is-t-é[tjs  ‘plague, tease, torment s.o’  tease-TRANS-3-3  c[9]é[9]k  ‘get a little cooler’  cool-[INC]-[DIM]  cu-xi[x}-cm-e  ‘Do this for me!’ (DIM)  do-BENEFACTIVE-[DIM]-3obj-IMPERATTVE  As stated by Thompson and Thompson (1992), in some cases there is a contrast between CVC and CC as in the following stem form, es-c’é ‘s.t. laid out’ resulting in the diminutive form, es-c’c’e ‘s.t. small laid out’. At the point in which reduplication is applied the  76  []  is  stressed, therefore it must be present. Here the reduplication process interacts with the vowel quality of the reduplicative affix. Since the copied vowel precedes a glottal stop  [9]  which appears to condition the k/, vowel coloration of the !/ must occur in the environment of the glottal stop. Again, the contrast above suggests that the prosodic affix is a CV. Throughout this analysis I assume the diminutive is suffixed after a stressed syllable, and the diminutive reduplication is a CV which is discussed further in section ?5. There are also examples where the diminutive is applied a second time as shown in (24). In the example below the non-diminutive stem is not an acceptable form, (*). (24) ROOT:  cw’x  ‘snow melts’ *s-cw,ex s-ewé[w’]x ‘creek’ s-cw’é[w’]u9x’a tiny creek’  DIM: DIM:  In an earlier footnote reference was made to resonants which have been glottalized. Here I present further discussion on this, first by presenting some of the pertinent forms’ 3 presented earlier, and I provide additional examples below (25):  (25) Glottalization of resonants: a. weak roots +a’c=áns kn-t-és p’ém-e-s s-pzü’ k”mime  es-sIy,Isoyl 9 q+-q+-iyx qIy-t  s-ké[k]n’ p’é[p’]m’ 9 s-pz’ü[z’u] k”m[m’]i9me” 9 ‘es-s[s]i qo+-q4-I[4i’x 4 9es.wovwyl qaqi’-t cok”-e-t-wá[w’]x”-e9 7 com-cm-[m’ m 9 t e ]i  l3  ‘a little person eats’ ‘partner’ ‘small fire’ ‘small bird’ ‘tiny’ ‘tiny strands are twisted together’ ‘to cheer themsives up a little’ ‘it is a little burned’ ‘it is a little damp’ ‘tug-of-war’ ‘small people, elf ‘very small, tiny’  do not discuss forms which underlyingly have glottal.ized resonants. form is simi Ian ly derived from /“y’/, but meaning ‘a little lopsided’.  77  b. strong roots qemüt c’ék”’-m nqWoxW1etn s-t-pac-pa-pac-t  s-nü[n]y’e? nu[n’]y’-tn qemü[m’]t c’é[c’]k”-m’ nqWo[qW]xWletn 5 s-t-pa’-pa[p]c’-paf’-t’  ‘a small amount of money’ ‘a little purse’ ‘small hat’ ‘he shines a light to attract fish to spear’ ‘sleigh-riding place’ ‘little grey faded ones’  qWjfl=e4  qW4jfl=e[fl]mx  s-méyx  s-mé[m’]i x 7 cI[cltx”-m’ =kst 9 co[c]i  ‘small birch-bark basket’ ‘small snake’ ‘he made a little house for himself ‘five animals or birds’  s-nüye’  cItx”  cIy=kst  (26) Non-glottalization of resonants: ‘pls k’ém-n s-xáy’wi s-pzü’ NO FORM  “é[ ] 9 pls ‘é[k’]m-n wi 9 s-xa[xji 9 s-pzü-zu n-mI[m]+  ‘little apple(s)’ ‘small feather’ ‘dear husband’ ‘small bird’ ‘we (EMPHATIC)’  According to Thompson and Thompson “most often resonants following the stress are accompanied by  //[]//b6  SPECIALIZING,  but this does not always happen and there seems no  basis on which to predict whether it will or not” (Thompson and Thompson, 1992) glottalize the resonant(s). Glottalization of resonants can occur with one or both resonants being glottalized. But, in some cases the glottalizing effect does not occur at all. The glottalized resonants can occur in various syllabic positions, pre- and post-vocalic either immediately adjacent or ?two positions from the nucleus or in the nucleus position. At this stage of this analysis I have not been able to determine where this glottalizing effect occurs. Further examples which could be considered idiosyncracies of the diminutive reduplication are shown below in (27). These, however, appear to be further examples of vowel coloration. (27) ckl[ól’e]=s  ‘catkins’  This form is listed as s-t-pac-pac[pJc’-pa’-t (Thompson and Thompson 1992:90). 15 Thompson and Thompson (1992) use the double sLashes to represent the underLying representation and the 16 single slashes to represent the phonemic representation.  78  c’q-c’q9=ew+[ü+e]=c’e ‘miniature toy boats’ ‘(of child) attractive’ ‘attractive’ cq”-c6cq”’t In conclusion, the diminutive reduplication is a copy of a stressed syllable in the stem with the copied vowel deleted if the syllable is a closed syllable, if the syllable is open the vowel remains. 3.3.1 Out-of-Control Reduplication: -VC/-C The third productive type of Me kepmx reduplication is out-of-control reduplication 9 process. The references are to limited control notions, such as simply out-of-control to spontaneous events and acts achieved with difficulty (Thompson and Thompson, 1992). The out-of-control reduplication affix attaches to a stem which involves the copy of the stressed vowel and the immediately following consonant (Thompson and Thompson, 1992) attaching it after the base, that is, C”’C  --  >C’C-VC, for example as in  (--).  Out-of-control  reduplicative template surfaces in the following forms: [-VC], [-R/-q, [-‘QC] and [-V]. These are discussed below in the examples (28), (29) and (30). (28) [-VC] forms: xüss a.  ‘it foams on its own’  foam[OC]  c’’-Ic’  ‘s.t. got pitch on it’  pitch[OC]  ‘cooled off’ cool-OC  tüp-p-s-ne  ‘I smash it up accidently’  break[OC]-CAUS-3-1  +ep-p=qIn  ‘sunset’  extinguisth-OC=head  9ecqW=nsse9 bake=berry[OC]  ‘berries dried and useless because baked by the SUn’  79  b.  c1tltXw  ‘a house finally becomes available’  house[OC]  9 e s ‘üq’’-oq”e  ‘managed to have a drink’  drink[OC]  ‘it finally got completed’ complete[OC]  c.  céw’-u’  ‘discussed’  talk[OC]  17 n-c=éw’-u-s  ‘s.t. laid on trail by s.o.’  LOC-Iay-Iongobject[OC]-3-3  Most of the above examples (28.a) reduce the copied vowel to a [a], but in the (28.b) examples the vowel has not been reduced to a schwa, but either it retains its original vowel quality or is conditioned by its environment. Examples above have been rounded by the following rounded consonants. The last forms in (28.c), however, has vocalization of thw [w’] -->  ] as discussed in the earlier sections (see section 3.2). The reduction of the copied 9 [u  form is the more common form. The examples in (28) are either of stems which are derived tautomorphemic roots as a result of the out-of-control reduplication. Other out-of-control examples are morphological derivations. Therefore, substantiating the fact that it is a copy of a stem. The second set of examples are shown in (29) copied vowels are deleted. In most cases they are followed by a resonant except for the last three forms (cited in Thompson and Thompson: 1992). These forms do not have a resonant as the copied consonant, yet the vowel of the copied form is deleted. In all those cases the segments are [coronal], that is, [c], [c’] and (29) [-R]  [+}. n4-ém’-m’  ‘s.t. goes into s.t. unnoticed’  carry[OC]  The root here is Ic/ Lay—long—object. 7 ‘ InterestingLy, this form deletes the schwa rather than h— epenthesis as one might expect. Thus at the surface, the root appears as a single consonant. The onLy other example which has a vowel after the root consonant is Lc—eyexke—tn], that is, Clay long object—?scale— intrument] ‘scales of snake’  80  ‘how one appears (the way one can can not help)’ LOC-look[OC]  es-cy-’y-e 9  ‘stuck in upright position accidently’  STAT-stuck[OC]-MDL  yóS”-I”=an’i-n-s  ‘they manage to keep (an event) secret from s.o.’  hide[OC] =ear-?-?  s 9 9üy-y=u  ‘happen to get together’  get-together[OC -?topsurface  8 cm-e-m=üse’  ‘of small berry (no control on size)’  small REPETITIVE-[OC] = round object  [-C]  ‘it gets full suddenly and unexpectedly’ fuIl[OC]-MDL  ‘you manage to move it’ move[OC]-CAUS-TRAN-3-2  ‘es-c-+-e  ‘(of liquid) spurt suddenly’  STAT-spirt[OCI-MDL  Compare the following example (30) which has the lateral affricate  [4W]  which has not had  the deletion of the vowel. (30)  9u=[]xv  ‘manage to enter’  enter=house[OC]  Out-of-control rduplication process also surfaces as shown below in (31) where the vowel is deleted, or the vowel is not reduced to an unstressed schwa as in the above examples. (31) [-‘C] ch-éh  ‘s.t. is put away’  put-away[OC]  ‘you managed to put it away’ put-away[OCJ-CAUS-TRAN-3-2  c’ix-x  ‘grabbed, grasped’  grab[OC]  Note, in the last example, above, does not delete the root vowel.  Thompson and  Thompson represent the root of this form as /cy’, therefore both of the vowels of the root would otherwise be deleted. The stress of both these weak stems also must be on the suffix  and other forms with the same root are unusual, where the out—of—control reduplication does not occur immediately after the root or the stressed sylLable, but after the repetitive. Thompson and Thompson (1992:117) have occurrences of the repetitive /[e’?]I in between the augmentative reduplication.  81  which adjoins the root as discussed in chapter two. At present I have no analysis as to why one would be a fully stressed vowel, and the other a stressed schwa.  Thompson and  Thompson account for this by stating that since it is followed by a suffix that can take stress; that is, stress has shifted cyclically to the inflectional suffix. The fourth surface variation of the out-of-control reduplication is a singular -V. This occurs with forms which have a glide in final position of the out-of-control base or stem as in (32). It should be noted that both examples below are acceptable. (32) -V  ciiw-u cüw-w  ‘done, made, fixed’  do-OC  In conclusion, the out-of-control reduplication is a copy of the stressed -C’c’, and it surfaces as one of the following [DC], [R], [C], [V] and [-‘c/C], and the template is suffixed to a stressed vowel.  3.1.4 Characteristic Reduplication: -CVC Characteristic reduplication process as stated by Thompson and Thompson (1992) extends the meaning of the base, has a general characterizing quality and is not productive. A copy of the stem’s first two consonants (and usually associated vowel) is inserted directly after the stem-inital syllable, resulting in an infix if other morphemes are involved -C(V)C. Examples taken from Thompson and Thompson (1992) are shown below in (33). (33)  k’aq’-Icq’-t  ‘thorny’  perforate[CHAR]-IM  cáq-cq-t  ‘spruce grouse’  tame[CHAR]-IM  cêp-cp-t  ‘s.t. that stains s.t. red’  smearblood[CHAR]-IM  9üp-9pn=ekst  ‘ten people’  both?[CHAR]=hand  82  As shown in earlier discussions of other templates of the copied forms, -CVC copied forms generally have a full vowel reduced to a schwa.  So is the case with the above  examples in (33). Whereas, the following forms (34) do not have a -CVC copy, but the copied vowel has been deleted surfacing as either -VC or a -CC copy which are not identical to the stem form. (34) zéw’[zu9]-t  ‘tiresome’  annoy[CHAR]-IM  n-wén-wn  ‘early’  LOC-early[CHAR]  Above in (34) the first example and below (35), also the first example, are examples which show the copied vowels are conditioned by the adjacent consonant to the right. Note one of the following form has an exact copy of the root. In reviewing this further, the root is /miy’! ‘distribute-plural’, therefore the copied form has been reduced and the [y’] has vocalized as discussed in earlier sections.  In some cases the features of the adjacent  consonants are [laryngeal] conditioning the immediate segment  In another case, the  following morpheme, suffix in this case, begins with a vowel therefore the final consonant of the copied template is not deleted or reduced to a schwa, [a], or an [i]. However in one case the following lexical suffix is affixed to the root-CHAR, but an epenthetic [h] has been inserted. (35) 9 mé-mi t  ‘its very contagious’  distributePL[CHAR]-IM  ci-cy=él’qs  ‘have new clothes  new[CHAR] = clothes  ci-cy-éle9=xin  ‘have new shoes’  new[CHARI-? = foot  ci-cih=é4-x’  ‘have a new house’  new[CHAR]=house  Note however, the following example in (36) which has not had a vowel reduction of the  83  copied form as earlier examples discussed. However, the two segments of the root and the copied template have similar  [PLACE]  (36) cñ+-cu+=ke’  features.  ‘seven people’  point[CHAR]=digit  In conclusion, the characteristic reduplication template, -CVC, -CRC -CC or a suffixes the first bimoraic syllable of a stem.  3.1.5 Affective Reduplication: CVThompson and Thompson describe the affective reduplication as a process which “derives stylistic variants connoting special attitudes ranging from familiarity, perhaps with overtones of nostalgia, to extreme specialization’. Further Thompson and Thompson (1992) state the CV- prefix is often reduced to a [a], but in some forms variation is heard between /e/ and  //  and a few cases show consistently /e/.  accompanied by the specializing extension,  Very often, but unpredictably, affectives are  [‘1.  The form is unproductive, but provides many  derivatives some of which are shown below (37): (37) t-tz-e 9  ‘arrow head’  AFF-protrude slightly-FORMATIVE  c-cp=Ikn’  ‘sun goes down over mountain into ocean (sunset)’  AFF-fit in=back  p-py’éx  ‘very tired’  AFF-tired  x-xw’é  ‘trail’  AFF-road  n-c-céw’s  ‘exhausted, tired out’  LOC-AFF-fatigued  n-ta-tm=ltn  ‘there are no berries in the basket’  LOC-AFF-lack=harvest  Again, as previously discussed similar processes occur here as with other reduplication processes. One of these is the çw..1wóyt  ‘sleep erratically’.  [+PHARYNGEAL]  feature of the vowel which is not copied as  Also, the vowel of the base is deleted if a strong suffix is 84  attached, and the prefix before the target is not copied. However, the following examples in (38) are conditioned by the consonants in the immediate environment of the vowels in question, that is, the rounding of the vowel in the environment of [+roundj consonants and the Lowering of the vowel in the environment of the laryngeal  [9].  9 (38) nkwe*kücel  ‘(s)he goes downstream’  LOC-AFF-descend water  ‘carry s.t. around eating it’ AFF-grasp-LIGATURE-food  9e-9uiy-m’  ‘laugh’  AFF-laugh-MIDDLE  ° 2 ce-cItx”  ‘a bunch of houses together’  AFF-house  In some cases, as Thompson and Thompson state, the affective form is the more common lexical form (39), that is the non-reduplicated forms are not lexical forms, or it is used less often. Also, as with the other reduplication processes the non-affective form appears with other affixes as shown below in the derived forms below (39). Thompson and Thompson (1990) refer to some of the stems as affective stems. These stems, I presuppose, do not appear in the non-affective reduplication pattern. (39) t-té-  ‘no (NEGATIVE)’  AFF-no  we-wIyx  ‘he cries’  AFF-cry  ccuikwe  ‘by oneself  AFF-finish-FMV  (40) s-p-plánt  ‘skunk’  NOM-AFF-skunk  0ne wouLd expect the copied vowel to assimilate the [+roundj feature and become Lu], but it does not. 19 should be noted that the affective form has a full voweL EeJ as opposed to a schwa. If it were a schwa the form would be ambiguous with the augmentative form. There are other cases where the copied vowel does not reduce to a schwa or is it coLored by its environment.  85  9 s-c-cu  ‘pattern’  NOM-AFF-do  9es--6p  ‘it is pleated’  STAT-AFF-pile flat  n-k-kic-n’-cm-s  ‘he haunts me’  LOC-AFF-?-1-3  And, again the vowel is deleted in forms which have a resonant as in (41). Note however that in (41) the vowel is retained in the copied form. (41) ccwoyt  ‘sleep erratically’  AFF-sleep  Some forms are ambiguous as shown below (42): (42) n-c-citx”-tn’  ‘empty house’  LOC-AFF-house-INSTRUMENT  n-c-citx”’-tn’  ‘winter village, group of several houses’  LOC-AFF-house-INSTRUMENT  In summary, affective reduplication prefixes a lexical root with C, Ce or C. As the other reduplication processes also have phonological processes. In most cases, the copied vowel is reduced to a schwa, whereas some others are conditioned by their environment. There are, however, cases where the copied template is a Ce- which I can not explain at this time.  3.2 Multiple occurences of N4t kepmx reduplication 9 N4-e9kepmx reduplication processes, as to be expected, should be able to occur within the same lexical forms. The 9 N+e k epmx reduplication is generally not limited in the number of instances the different types of reduplication can occur. However, they can not occur randomly. The amount of occurences is semantically limited, and the ordering relationships between some of them are evident. The following will present the various acceptable forms of multiple occurences of N4-e9kepmx reduplication. First, I present the data and then provide an explanation where necessary. 86  3.2.1 Augmentative and diminutive reduplication The application of both the augmentative and diminutive form is quite common. These reduplication types can either target the same part of the stem or different parts of the stem. Since the diminutive does not necessarily target the initial part of base form but the stressed -CV syllable. Examples of these are shown below in (43): (43) STEM: AuG: DIM: AUG/DIM:  cItx” ci-citx’’ Ci[C]tXv cici[cjtXV  ‘house’ ‘houses’ ‘small house’ ‘small houses’  AUG: AUG/DIM:  cak-cék cak-ce[c}k  ‘pleasantly cool (weather)’ ‘a little cooler’  ROOT:  k”’am  STEM:  k’me9txn  ‘small foot’, =xn ‘foot’ ‘have small feet’  Cewt s-céwt=mx scé-cut=m’x t = m’x= qI[q]n’ = kst 9 s-cu s-cu9-cu9t=m’x=qi[q]n’=kst  ‘ultimate’ ‘youngest child’, =mx ‘people’ ‘last child in family’ ‘little finger’, = ekst ‘hand’ ‘little fingers (on both hands)’  AUG/DIM: ROOT: STEM:  DIM/GLOT: DIM/GLOT: AUG/DIM: GLOT.  Since both reduplicative process function frequently, the co-occurences of them is not unusual. The examples above also show the phonological processes which occur with these reduplication processes, that is vowel reduction, resonant glottalization, and ordering relationship.  3.2.2 Augmentative and out-of-Control reduplication The augmentative and out-of-control reduplication processes can also cooccur. Both these interact with each other and have independent phonological and morphological processes. Since these reduplication types can target different parts of the stem no further discussion is required at this time.  87  (44)  ‘hang over’ ‘(of plants) droop, wither’, =qin  ROOT:  oc: AuG/oc:  +p’-p’-qin  ROOT: AUG/oc:  cw n-cu-cuw-w = Ip  ‘head’  ‘(of plants) droop, wither’ ‘make,do’ ‘decendants’  3.2.3 Augmentative and affective Both the augmentative and affective reduplication processes prefix the ?inital CV of a stem. The example below shows that both the augmentative and affective reduplication as identical copies of the base [tz]. (43)  AuG/AFF: AFF:  AuG/AFF:  t3-t-t4z-e 9 9e-üy-m’ e-9uym’ 9 ‘e-  ‘arrow heads’, /tQz/ ‘protrude-out’ ‘laugh’  ‘laugh’,plural  Are the results the same because the [t] and the [zj are near identical segments, and that they both have the same place feature? 3.2.4 Diminutive and out-of-control reduplication In N4e9kepmx, the diminutive can also be used with the out-of-control reduplication. Note the following example (44) has a ‘lexical’ suffix: (44) RooT: OUT-OF-CONTROL: DIMINuTIvE:  ‘enter’ ‘go in without been noticed’ ‘a small container’  [4-em’ n-4-ém’-m’ n-4-é-+m’-m’-tn LOC-ROOT-DIM(2)-INST  (45)  ROOT:  x”’us  ‘foam’  OUT-OF-CONTROL:  US-S  ‘foaming’  DIMINUTIVE:  xvu[x\]ss  ‘a little foam’  Note the following example has out-of-control reduplication and then diminutive is applied.  Stress has  cyclically been reassigned,  therefore the  diminutive has been applied to a later  derivation.  (46)  oc/DIM:  x’y’=aq-[aq}s-e-tI[t]yxs ‘they carried a little baby’ carry=nose[OC]-DIR[DIM (on a stretcher or coffin) 88  Note the example in (46), in that it has two reduplication processes applied to this stem form. The out-of-control reduplication applies to the stressed vowel of the root. Then, the diminutive reduplication applies to the lexical suffix. The out-of-control reduplication and its interaction with diminutive reduplication must have an ordering relationship. From the above examples it is evident that out-of-control reduplication must precede diminutive reduplication. An ordering relationship is necessary since both out-of-control and diminutive target a stressed base. If stress was not already assigned neither the first or the second reduplication process would have an identifiable target, thereby not being able to be applied. Other evidence supporting the fact that out-of-control reduplication precedes the diminutive reduplication is shown in (47). (47)  OUT-OF-CONTROL:  “1S-S  DIMINUTIVE:  xwu[xw]s..s  ‘foaming’ ‘starting to foam a little’  The above form has the base of the root undergoing the out-of-control reduplication applied first, the the diminutive reduplication applied next. If the diminutive were to apply before the out-of-control the following derivation would result (48): (48)  DIMINuTIvE:  XWUXWS  OUT-OF-CONTROL:  *xWl[xWl  [u’]s  ‘a little foam’ ‘starting to foam a little’  Also, note the following form which has a lexical suffix (48): (49)  OUT-OF-CONTROL:  xwy=aq[aq]s  DIMINUTIVE:  X”’  aq[aqjs-e-tl[-t]iyxs  ‘person/corpse lifted in a stretcher/coffin’ ‘person (e.g. child) lifted in a stretcher’  Again, to account for the form (49) we must assume that stress was assigned at the point at which out-of-control took place. The first suffix, -aqs, which is a strong suffix has had the stress cyclically assigned after it has been attached. But when the second inflectional suffix is attached stress has again been cyclically assigned. Therefore, out-of-control is ordered 89  before stress is assigned to -tiyxs thereby creating an environment for the diminutive. To summarize this section the following derivations (50) show the variations of reduplication and how they interact with one particular root.  (50) a. ROOT: b.  INSTRUMENTALIZER:  ouT-oF-coNTRoL: d. DIsTRiBuTIvE: e. OUT-OF-CONTROL: f. DIMINuTIvE: g. DIsT/DIM/oc: C.  ém’  ‘enter’  n4ém’-tn *n4émem.tn  ‘a container’ unacceptable ‘containers’ ‘to carry s.t. unnoticed’ ‘manage to carry things’ ‘small things are carried unnoticed’  n-4-m’-4-ém’-tn n-4-ém’-m’ n-4-e[-i-]m’-m’ n-m-+ém’-(?e)m’  In (50.b), an acceptable form, a locative and instrumental affix are attached, whereas the same form which has had the out-of-control reduplication applied to it is unacceptable, perhaps due to the restrictions of the morphological process. In (50.d) this form is simply a plural form of (50.b), and (50.e) is an out-of-control form.  However, (50.0 is an  application of diminutive after the out-of-control reduplication as discussed earlier.  The  (50.g) example has three applications of reduplication showing that the augmentative is independent of the the diminutive and out-of-control. between the latter  two  But again, we see the interaction  showing the ordering relationship and cyclicity of the reduplication  processes.  3.2.5 Affective and Out-of-Control Still another multiple occurrence of reduplication processes occur with affective and outof-control. The following example (51) shows this: (51) AJ-oc AFF-OC  These  two  ce-cIt-gtx’’ c-cun-ém’-m  ‘directing s.o., they telling s.o. what to do’  are  reduplication forms occur isolated from each other; therefore, there is no 90  significant phonological process or interaction between the two.  3.2.6 Characteristic and Diminutive: Characteristic and diminutive reduplication also cooccur within the same stem as shown below in (52). ‘a little, somewhat, kind of sour’  (52) c’ü[c’]l’-c’l’-t  Again, the diminutive reduplication process must occur after characteristic reduplication process. Otherwise, the characteristic copied form would not be a copy of the root/stem form.  3.2.7. Three independent cooccurring applications of reduplication The example (53) below shows a rare example of three different reduplication processes occurring within the same root and stem.  (53) c’n-cé[c’]n’-c’n’  ‘small grasshoppers’  AUG-grasshopper-[DIM]-CHAR  Again, significant interaction occurs between the three since they occur within targets and result in reduplication in three different locations. It should be noted that characteristic reduplication must occur befor diminutive. And augmentative must occur before diminutive. Therefore, the ordering relationship among the three must be either as shown in (54) or  (55). --->  (54)  AUGMENTATIVE  (55)  CHARACTERISTIC --->  CHARACTERISTIC --->  DIMINUTIVE  or AUGMENTATIVE  --->  DIMINUTIVE  Whereas, (56) below shows the ordering relationship between out-of-control and diminutive.  (56)  OUT-OF-CONTROL --->  DIMINUTIVE  91  The relationships between the other are not known at this time, it will not be discussed any further since the relationship between the ordering relationships and the reduplication processes are not crucial to this analysis.  3.3 Conclusions In conclusion, the reduplication processes can be represented below in the Table 14 as shown below:  REDUPLICATION TYPE  TYPE OF AFFIX  TARGET  augmentative  prefix  CVC morphological stem  CVC-/CV-/C  diminutive  suffix  CV  CV-/C-  out-of-control  suffix  CV bimoraic syllable  -VC/-C  characteristic  suffix  CVC or CVC, first of root  -CVC/-VC/-CC  affective  prefix  ?stressed CV, lexical root  CV-/C-!C-I(Ce)  COPIED FORM (TEMPLATE)  (BASE)  of stem  Table 14. Me’kepmx Reduplication Types  92  OThER  glottalization of post templatic resonants, ordering relationship  glottalization of post templatic resonants  Chapter 4. Overview of Theoretical Approaches and Issues Building on the descriptive analyses of N+e9kepmx reduplication reviewed in the preceding chapter, the focus of this chapter is to provide a critical overview of previous theoretical accounts of reduplicative behaviour in Salish, and to outline the theoretical framework of Prosodic Morphology which will be adopted in the present analysis. Specifically, §4.1 summarizes Bell’s (1983) insightful discussion of internal -C diminutive reduplication in Shuswap (e.g. [sqee] ‘dog’, [sqeqeJ ‘little dog’ within Marantz’s (1982) framework.  § 4.2 discusses Broselow’s (1983) treatment of the diminutive and distributive  reduplication in Lushootseed as compared with Thompson and Shuswap, and §4.3 presents Broselow and McCarthy’s (1983-4) subsequent discussion of this same body of data, as well as of the -VC reduplication patterns in Lushootseed. This general overview will illuminate the particular nature of the difficulties which Salish reduplication created for the early theoretical frameworks of autosegmental templatic morphology. The subsequent developments within the Prosodic Morphology approach of McCarthy and Prince (1986, 1990) are then outlined in §4.4, and Carlson and Bate’s (1990) application of this framework to out-of-control reduplication in Spokane is discussed in §4.5.  As  McCarthy and Prince’s (1986, 1990) theoretical framework provides the basis for the analysis to be presented in Chapter 5, the discussion here intends to provide the essential background of theoretical assumptions and analytical mechanisms which crucially underlie the present approach, and to document the specific types of Salish reduplication data which as yet have not received a fully satisfactory and systematic theoretical treatment. 4.1 Bell (1982) Bell observed that the Shuswap diminutive reduplicative pattern, internal C reduplication, had implications for Marantz’s (1982) 93  theory of non-concatenative  morphology. Two areas of identifiable issues Bell stated are (i) that the domain of a copied melody borrowed by a reduplicating affix may be an entire word and (ii) problems with the procedure as proposed by Marantz (1982). Therefore, an alternative procedure is proposed as discussed in this section. Evidence presented below in (1) reveals that the Shuswap diminutive places a copy of the C preceding a stressed vowel such as: (1)  Shuswap diminutive : 1 sqéfe a. sqéqe pésXk°e b. péps&s..k°e cq’ékp c. cq’éq’Xp kpqin’ d. kpqIqn’kn  ‘dog’ ‘little dog’ ‘lake’ ‘little lake, pond’ ‘fir tree’ ‘little fir tree’ ‘her head aches’ ‘my head aches’  Bell’s first observation in regard to the diminutive reduplication is that it applies across morpheme boundaries, including inflectional endings may be crossed.  She refers to  Marantz’s (1982) theory of reduplication, ‘that internal C reduplication (can) ignore boundaries implies that a reduplicating affix may borrow an entire word’s phonemic melody, not just the stem’s melody”. Another conclusion which reinforces Marantz’s theory is that the locus of reduplication is fixed only with respect to stress which has been assigned before reduplication. Since Bell accepts the claim that Salish languages generally can be divided into three stress classes: strong, variable and weak, the position of the Shuswap reduplicationg affix can be determined at the level of the word (Bell, 1982:334). Thus, she claims that reduplication may copy the entire phonemic melody of the relevant domain  These are Shuswap diminutive reduplication forms presented by BeLL (1982). 1 have a glottalized [q] in the root, that is the second consonant.  94  The example in (1.c) does not  which will be copied, and association will proceed from right to left. Bell concludes that entire melody of the relevant domain is copied in reduplication. It also allows parallel treatment in respect to the position of the reduplicating affix relative to the relevant portion of the skeleton. Thus, Bell proposes a variation in the procedure, although there is an undesireable consequence. Both analyses are shown below in (2) and (3): (2)  Marantz’s analysis: kpqI  I  cvccv  (3)  n’kn  I ccc  +c+  kpqi  kpqI  cvccv  +c+  n’kn  I  I  ccc  a. Bell’s analysis (336:(8)) kpqIn’kn(-->)  kpqi  n’kn  cvccvccc  IIH\ cvccv +c+  ccc  HI  Bell states when the reduplicating affix is inserted, association lines between the melody and portion of the skeleton following the infix are broken as in (3). Then the entire melody is broken and association proceeds from right to left, as in (4), the correct output results. (4)  kpqI  IH i\ cvccv  n’kn  +c+ ccc k  p q I h’k n  Thus, Bell’s analysis copies only the relevant domain is copied rather than the entire phonemic melody. The undesireable consequence, however, is since infixation causes a disruption of association lines; however, unlike Marantz’s analysis the “internal reduplication” is a subspecies of infixation. 95  Since’s Bell’s (1982) analysis of Shuswap is applicable to N4-e9kepmx, this early analysis on Salish language revealed an insightful discussion of internal  -c  diminutive  reduplication which identified some problematic areas. 4.2 Broselow’s Analysis (1983) In her previous analysis of Thompson diminutive reduplication,Broselow (1983) states that the diminutive reduplication is a prefix C(V) which copies the consonant preceding the stressed vowel, e.g. [q’ümqn] ‘head’ [q’oq’umqn’j (diminutive). She makes several related claims and assumptions:  1) the diminutive reduplication is internal i.e. infixing in the  Interior Salishan languages (namely in Nlt9kepmx and Shuswap); 2) that the diminutive is entirely predictable, in that the reduplicated form is a copy of the stressed vowel of the base and the consonant preceding it; and 3) that the diminutive affix can also occur between two morphemes, i.e. is not necessarily morheme-internal, e.g. [s-t’malt] ‘cow’ [s-t’malt-Iy’t]. Broselow concludes her discussion by stating that the diminutive infix is subcategorized to occur before a stressed syllable rather than before a morphological stem. In other words, the author assumes that the diminutive infix is attached to a phonological constituent, rather than to a morphological constituent. The fundamental question she addresses here is “at what level and in what ways infixes are related to the other morphemes of a word” (1983:345). Derivations to illustrate how her analysis actually applies to a given form (including distributive 2 and diminutive) is shown below : 3  The term is the same as augmentative which I have chosen to use throughout this thesis. 2 For further expanation of the representation in (5) and (6) refer to Broseow (1983:340). 3  96  (5)  word stem  dist. Si!  ii \  cvc  stem Si  ii  cv  sil’  III  cvc  6 V (6)  dist. t’  u m  I I I  cvc  word stem stem t’  u m a m  I I I  aIt  I N\  cvcv+cvccc  66 She also deals with doubly reduplicated forms which I refer to as cooccurrence application of reduplication where both the diminutive and distributive are affixed on the same root, e.g. [s-cuw] ‘something done, work’, [s-cw’cócw’j (distributive-diminutive). She states that the two reduplicated forms are associated at different levels of representation; i.e. first, the distributive is prefixed to the stem, whereas the diminutive is is attached to a syllable; that is both subsequent affixes are associated at different levels as shown in (6). The basic tenets of Broselow’s actual analysis are that the base form of the diminutive affix is an infix (compared to Lushootseed which she concludes is a prefix) attaching to some phonological representation as shown in (6) With this analysis areas which were retained in Broselow and McCarthy’s (1983-84) analysis state the following exist (i) internal reduplication exist and (ii) copy and association of melody as either prefixal or suffixal reduplication,  On the otherhand, a more set of  restrictive operations exist including: (i) two types of internal reduplication, “true’ and  97  “apparent” infixing reduplication, infixation is either the entire melody or to a prosodic constituent. This dichotomy resolved what Bell (1982) considered to be a “problem”. Broselow concludes the “diminutive as an infix is not included in the morphological structure at all”. The present analysis agrees with Broselow that the base of diminutive reduplication is prosodically defined, but disagrees that this should disqualify the reduplication affix from having morphological status within the word. Thus, in the analysis proposed here, the diminutive suffix is indeed treated as having full morphemic status. Depending on where stress falls, the diminutive may (e.g. (a) [si[si]lj ‘little piece of fabric’ or may not (e.g. (b) [s-t’malt] ‘cow’ [s-t’malt-Iy’t] result in the morphological root surfacing as a discontinuous constituent (e.g. [s-t’malt-i[tijy’t]. examples below (H (7) a.  b.  =  Compare Broselow’s (1993:339)  Haeberlin):  Thompson s + t’omállt’ (H) s + t’omámallt’  cow diminitive  s + t’omallt’ét s + t’omallt’étet  calf diminitive  qoesp (H) qoiqsp  buffalo diminitive  qospé o 9 t t 9 qospépe  young buffalo diminitive  Although Broselow and McCarthy treat diminutive as prefixation, I consider the domain of the diminutive affixation site a stressed syllable, and the suffixation is immediately after the first  i  of the stressed syllable. This analysis pre-empts having to manipulate stress  shift leftward as shown below (8): (8)  ROOT: DIMINUTWE: VOWEL-DELETION:  ‘buffalo’  qWIsp  qwj[qijsp [qwI[q]sp]  ‘little buffalo’  98  ROOT:  qwIsp  LEXICAL SUFFIX.!  q”isp=IC’e9  ‘buffalo hide’  [q’’isp=ipc’e9]  ‘small buffalo hide’  STRESS REASSIGN: DIMINuTWE:  4.3. Broselow and McCarthy (1983-4) Broselow and McCarthy’s (1983-4) paper addresses the general issue of infixing or internal reduplication. They argue for a particular conception of infixing reduplication which is generally compatible  with the theory of prefixation and suffixation with a minimum of  constraints. They do not succeed, however, in reducing all cases of classical “infixation” to either prefixation or suffixation. They must conclude therefore that there is a typological distinction of two particular types of internal reduplication. “True” infixing reduplication involves infixation within a template, copying of the entire stem melody, and association of the infix with that melody from either the left or the right. The other type, “apparent” infixing reduplication, involves prefixation or suffixation of the reduplicative morpheme to a prosodic constituent  -  a syllable or a metrical foot  -  rather than to a morpheme; copying  and association then applies as in the case of normal prefixal and suffixal reduplications. This analysis provides a more restrictive set of operations. With true infixation, the entire stem melody is copied and the direction of association is specified, whereas when the reduplicative affix is prefixed or suffixed to a prosodic constituent, the melody of the constituent is copied and association proceeds in the unmarked direction of association as with other reduplication types.  Among the several languages which they draw on for  empirical Support for their conclusions, Broselow and McCarthy refer to two Salishan languages, Thompson [here referred to as N4e kepmx] and Shuswap [here referred to as 9 ] for this analysis. They conclude that these languages have apparent infixing 4 Szx”epmx  mis is the N4e 4 kepmx form for Shuswap. 7  99  reduplication. The basis of their analysis for these languages is the diminutive reduplication. As is claimed in the traditional Salish literature (e.g. Kuipers (1974), Thompson (1990), Broselow and McCarthy (1993-4) note that the diminutive in Szx’’epmx is stress sensitive: a copy of the consonant preceding the stressed vowel in the base form appears after that vowel in the diminutive. Recall (4.2) that Broselow’s (1983) earlier analysis claims that the reduplication of Szx”epmx and N4-e9kepmx both prefix CV- to the stressed syllable, shift stress forward onto this CV, and delete the original vowel of the base (except before a resonant in Szx’’epmx and before a laryngeal in N4-e9kepmx).  Broselow and  McCarthy (1983-84:66-67) however note that Broselow’s (1983) analysis is not fully compatible with the general in that forms which have an initial consonant cluster cause a problem for the analysis e.g. cqép ‘fir tree’ cqéq+p ‘little fir’. The solution that Broselow and McCarthy propose entails a reanalysis of syllable structure. That is, following Newman’s (1947) analysis the Coast Salish language Bella Coola Broselow and McCarthy assume “that Shuswap also has vocalic consonants” (1983-84:67).  Consequently, they claim that the  prefixation of Szx”epnix forms can be shown as (9) which has a [c] in initial position. (9)  6  6  c  qé+p  UNDERLYING REPRESENTATION  Diminutive reduplication then prefixes to the second, i.e. the stressed syllable: (10)  6  6  PREFIXATION  OF  CV  TO  STRESSED  SYLLABLE  C+ c  CV+  C  VCC q é4p  To draw on the “syllable” status of initial consonants in obstruent clusters would undoubtedly be profitably recast in the light of Bagemihl’s (1991) more recent arguments that such obstruents are demonstrably not syllabic, but would be analysed a independentllllly moraic. 100  However, Broselow and McCarthy’s essential claim here may be interpreted as a constraint against complex onsets: if the stressed syllable has a simple onset, then the CV- prefix is positioned directly before it, for example: (11)  C[CV-[qe+p]6  A more contentious aspect of Broselow’s (1993) diminutive analysis, adopted in Broselow and McCarthy (1983-84), is her claim that it is consistently prefixal.  While  Broselow’s treatment of the diminutive as a CV- prefix in Lushootseed seems empirically well motivated, Broselow and McCarthy (1983-84:66) claim more generally that: In all these related [Salish/MNJ] languages, then the diminutive is a CV prefix; the languages differ only in the constituent to which the CV is prefixed (The stem for Lushootseed [sic] and the stressed syllable for Thompson and Shuswap). This contrasts with the widespread treatment of this reduplication pattern by traditional Salish scholars as insertion of a -C(V) after the stressed vowel. The major empirical consequence of Broselow (1983) and Broselow and McCarthy’s (1983-84) claim that diminutive reduplication in Thompson (N4’e9kepmx) is prefixing is that they would have to account for the fact that stress “shifts leftward from the base onto this “prefix”, and that the root vowel then deletes (unless followed by a (12) (PREFIX)-CV-fC’’C(C)--> CV-C’7(C) Base Dim. Redup. citx” xé s-pzü9 fwóyt  cicItxv ‘house’ ‘high’ xe-xe’ ‘animal’ sp-zu-zü9 ‘she sleeps’  [9].  Stress Shift  c1citxv xé-xe s-pzü-zu9 Iwo coy’t  Base-V deletion cI-ctx” -  5 ¶moct  Broseaw does not present any data which has a Cy] —> [19], however I represent it as such in this form. 5  101  A major difference from the previous analysis is that the stress on the base form is deleted or shifted and the original stem/root vowel subsequently deleted except before  /9/  thereby creating the proper result. While deletion of unstressed (or detressed) root vowels is not uncommon in Interior Salish, what is particularily problematic is the requisite treatment of this stress “shifting’ leftward from the root to a prefix, as this type of phenomenon is otherwise unattested in these languages. A further consequence of treating diminutive reduplication as suffixation rather than prefixation relates back to the issue discussed above of how to treat initial CC- clusters in order to “prefix” the diminutive between them. Note that under the suffixation hypothesis, it is no longer critical to split the prosodic constituency of an initial CC- sequence. Rather, the problem is now redefined as how to identify the post-stressed-vowel insertion site in terms of prosodic constituency. This issue will be addressed in chapter 5. 4.4 Prosodic Morphology The central assumptions and claims from both McCarthy and Prince’s theoretical prosodic morphology framework of (1986) and subsequent theory of prosodic circumscription (1990)6 which will be drawn on in the present analysis are discussed below. 4.4.1 Assumptions and Claims McCarthy and Prince’s article (1986) provides a theoretical basis for nonconcatenative morphology. In word derivation, a strategy requiring a target to accommodate a base must be defined in categories and rules of prosody provided by syllabification and stress/accent. Templatic morphemes must be defined prosodically rather than segmentally. Prosody places  Donca Steriades (1988) model discussed 6 elsewhere.’ will not be discussed here.  in  RedupLication  102  and  syllable transfer  in  Sanskrit  and  strong conditions of adequacy on templatic theory; in particular, templates can only (be) count(ed) “up to two” and only moras, syllables and feet can be counted. The templates must be structurally adjacent to an element, this constituting a strict locality condition. Further, a condition on templatic interpretation states all elements in a template must be categoricallyly satisfied, that is the Satisfaction Condition. Prosodic categories within the McCarthy and Prince framework (1986) include the following categories: prosodic word, foot, syllable, monomoraic syllable, bimoraic syllable, and core syllable (defined essentially by stipulation as a CV syllable). Mapping parameters which allow for the prosodic reparsing of a copy of the base accommodate the templatic domain. As in Marantz (1982), the three following assumptions are made.  First, the entire  segment melody of the reduplicative domain is copied onto a new plane. Secondly, the mapping of the segmental material into the template is directional and continuous: left-to-right for prefixes, right-to-left for suffixes, and free choice for root-and-pattern systems. Lastly, the domains of affixation are fixed prosodically and morphologically. In order to characterize phonological structure and morphological consequences McCarthy and Prince (1986) appeal to two basic principles:  the Prosodic Morphology  Hypothesis and Template Satisfaction Condition. These are stated below: (13)  Prosodic Morphology Hypothesis. Templates are defined in terms of the authentic units of prosody  McCarthy and Prince, following Selkirk (1980), Hayes (1989), Hyman (1985), etc., assume that the units of prosody defined by the Prosodic Hierarchy are: mora (i’), syllable (6), foot (F), prosodic word (W), and so on.  Following Shaw (1992,1993), 1 assume that this  hierarchy must be extended to include a distinction between nuclear and non-nuclear moras.  103  The second condition constrains mapping into the template: (14)  Template Satisfaction Condition. Satisfaction of templatic constraints is obligatory and is determined by the principles of prosody, both universal and language-specific.  4.1.2 Prosodic Circumscription (1990) In order to account for misalignment between the edges of a templatic domain and of the base, McCarthy and Prince (1990) propose that morphological domains may be circumscribed according to well-defined prosodic constituency. (14)  Prosodic Circumscription of Domains. The domain to which morphological operations apply may be circumscribed by prosodic criteria as well as by the more familiar morphological ones. In particular, the minimal word within a domain may be selected as the locus of morphological transformation in lieu of the whole domain.  Several phenomena within the theory of prosodic circumscription of bases require attention. To discuss these, McCarthy and Prince (1990) introduce the following formulism to specify mechanisms of morphological functions which operate on a base. (15)  B o *  O B:o B/o C E o:<C,E>  = = =  = = = = =  base function an operator which designates the relationship between the two factors. operation factoring on the base residue constituent edge  =  One phenomenon is extrametricality which defines the location of an edge for purposes of a given rule or sets of rules is characterized by the constraint shown below: (16)  Factoring Imposed by Phonological Constraint = B4*B/ B  104  Below as in (17) provides an example which shows how the base is divided. (17)  Definition of Operation Applying under Extrametricality 0/4) (B) = B:Ii * O(B/)  (18)  Definition Operation Applying under Positive Prosodic Circumscription 04 (B) = O(B:) *  In the analyses which follow extrametrical constituents will designated as angle brachets. The prosodic criterion always selects the minimal base. McCarthy and Prince propose that the prosodic circumscription of the base to which a rule applies is limited by the following constraint: (19)  Minimum Word Constraint. Positive prosodic circumscription of a base may only appeal to the category Minimal Word. That is, in: 0:o <C,E>, C = minimum word  McCarthy and Prince further claim that the minimal unmarked foot is two moras, and that two moras is the lower limit on word size. In support of their hypotheses, McCarthy and Prince discuss truncation of words where the template for truncation is defined by the minimal word as its base. This approach also allows that a language may characterize a templatic affix as a minimal word. 4.5 Carlson and Bates’ Analysis Carison and Bates (1990) have have applied McCarthy and Prince’s framework of nonlinear phonology and prosodic morphology to the analysis of out-of-control reduplication in Spokane Salish, paying particular attention to “the stress system...(and)...surface variants of the OC reduplication that occur with root shapes other than canonical CVC” in the Spokane language (1990:73). They state that the out-of-control reduplication as copying a prosodic constituent, the mora of the root and since the melody of the mora (VC) is copied, right to left association of the suffix can never incorrectly link the final consonant of CVCC  105  root. Their representation is shown in the following example (12): (20) ac’c’ ‘observe’ (9ac’ [sj ‘watch’) After copying and association (20) would have the following representation as in (21). (21)  ac’ CVC-VC-C ac’c’, ‘ac’  ->  stress and Unstressed Vowel Deletion  ->  Further discussion in Carlson and Bates state weak CVCC roots regularly form OC as C C C VC C. In summary, they account for this form as single or double mora copying. Issues in question, they conclude with, are the status of stress in morphologically based languages and the syllable weight distinctions. The Carlson and Bates’ analysis does not clearly state what constitutes as a mora. If they conclude that there is single mora copying should that not only include the vowel? Or, is that specifically referring only to the second mora, and if it does what of the instances they refer to as weak CCVC root derive [cnnip’p’] meaning ‘it suddenly got banded’. Again, what is the status of the mora? Is it, as in (22)? (22)  a. 1st 2nd  VC b. cnnip’p’  The forms which are further analyzed in Carison and Bates in regard to other canonical forms are not discussed here, since comparable forms are not clearly evident in N+e9kepmx. N+e’kepmx does not have similar forms as those present in (22); therefore, I do not discuss it any further.  106  Chapter 5. A Prosodic Template Approach to N+e9kepmx Reduplication N+e9kepmx as shown in Chapter 3 has five independent reduplication forms designating specific morphological processes. In this chapter each of these processes will be presented within the prosodic morphological framework of McCarthy and Prince (1986, 1990). Specific issues related to the application of this theoretical framework to N4’e’kepmx data will be discussed in  § 5.1. As it is crucial to the prosodic definition of templatic domains, the  N+e9kepmx syllabification and morification presented in chapter 2 will be reviewed again in  § 5.2. Specific analyses of each of the five reduplication processes will be presented in  § 5.3 to § 5.7. Discussed will be the appropriate representations within the framework of prosodic morphology of the base and template, the role of prosodic circumscription to characterize extraprosodicity, and the subsequent phonological processes relevant to each reduplication process.  5.1 Issues and Problems: 9 N4’e k epmx Data The reader is referred back to chapter 3 where the five 9 N4-e k epmx reduplicative morphological processes were descriptively identified. Now the representation of each of these patterns needs to be determined as to whether the reduplicative template is in fact prefixed or suffixed, whether the base is defined prosodically or morphologically, what the appropriate form of the prosodic template is, and whether extraprosodicity is in effect. With respect to the potential role of prosodic circumscription, it is important to ascertain what defines the “minimal word” in 9 Ne k epnix, and whether this plays a critical role in the definition of the base and/or affixes in the reduplicative morphology. A particularily challenging issue is the definition of the prosodic category of some of the reduplication templates for N+e9kepmx. For example, the template for the reduplicated  107  diminutive (e.g. s-mü[m]+ec) and out-of-control (e.g. q’c-c-s-t-és) forms have a single consonant as the copied form. However, the prosodic criteria defined in McCarthy and Prince’s theoretical framework do not characterize a single consonant as a prosodic level: therefore, how is the single consonant defined as a premoraic (onset) and as a single mora in second position? Similarily, the out-of-control reduplication pattern -VC is not definable templatically in terms of authentic prosodic unit, since nothing within the current theories adopted by McCarthy and Prince (1986, 1990) recognizesany internal constituency within the syllable above the moraic level. Thus, the units of prosody as presented and claimed by McCarthy and Prince, do not suffice or completely address the 9 N4-e k epmx templates. Another problem as discussed in chapter 4, is the fact that the location of the stress in the base is the locus of infixation in the diminutive and out-of-control reduplicative patterns, e.g. ‘es-9a-’I{ ] ‘have several stripes, 9 a sharp end,  OUT-OF-CONTROL’.  DIMINUTIVE’  and 9es-t-k’át[t]-e ‘pierce something with  That is, the place of infixation is not the end or the beginning  of an edge-defined syllable or foot as to be expected, but more restrictively, the stressed syllable. Detailed treatment of this issue is given in § 5.4 and  § 5.5  below.  5.2 Syllabification and Morafication 5.2.1. Mora As presented earlier, McCarthy and Prince (1986, 1990) claim the mora is the lowest element of the prosodic hierarchy. Recall that the motivation for the N+e9kepmx moraic representation is based on the inchoative infix  [9]  which was discussed earlier in  §  and the various reduplication morphological processes discussed in chapters 3. inchoative infix,  [9],, The  is crucial for the argumentation that the onset C must directly link to  the syllable node as in (1) and not directly to the initial mora, as represented in (2) below  108  and earlier in (1)  § 2.1.  c\  /t  cvc (2)  6  Al  cvc Thus, (3) is a moraic representation of an inchoativized {+strong] root/stem: (3)  a.  6 /N //I’ll,  b.  6  //\\ é 4-  //U  ] 9 m[  m[9] é 4-  ‘become less painful, lesser pain’  Recall if both the initial consonant and the nuclear vowel were dominated by a mora as in (3.b), then the inchoative glottal would be ‘true’ infixation into a prosodic constituent. Broselow and McCarthy (1983) and McCarthy and Prince (1986, 1990) argued against the existence of such infixation processes, reducing the majority of apparent cases of infixation to cases of prefixation or suffixation to a prosodically well-defined base.  If the initial  consonant of N4-e9kepmx roots is in fact attached to the syllable rather than the mora, then the position of the inchoative can be defined as prefixation to the first mora. The other issue is in regard to the segment immediately following the left-most mora. Is it attached to the preceding mora as (4)a., or attached to the syllable as in (4)b., or dominated by another mora as in (4)c. (4)a.  6  CvC  b.  6  /\  CvC  c.  6  /N  CvC  Earlier, I posited that the Nle9kepmx closed syllable, CVC, is bimoraic, as the prosodic  109  form represented in (4.c) . 1  5.2.2 Syllable Preliminary observations show that the N4e9kepmx syllable can be either open (5) or closed (6)2.  There is no vowel length distinction nor tautosyllabic vowel sequences;  therefore no heavy syllable of the form CV: or CVV exists.  (5) a. open syllables: 6 6  / /  )‘  é.  n i  b. closed syllables: 6  ‘ear’  s  6  ,‘/f H m u m’.  +  é  c  ‘little woman’  Vowel-glide sequences, nonetheless, occur as shown below (6): (6) a. I’eyk/ b. /céwt/ s-céwt-mx  ‘kinnickinnick (or bearberry) ‘ultimate’ ‘youngest child’  NOMINATIVE-ultimate=people  c.  téw-mn  ‘store’  sell-INSTRUMENTAL  cf. d. tawu  ‘sold to someone’  OUT-OF-CONTROL  sell-OUT OF CONTROL  The above examples, except for (6.d) show the glides [y/wj as coda, and therefore they close the syllable. At the surface as shown in (7) the maximal representation of the 9 N+e k epmx syllable in word-final position has consonantal clusters in post-vocalic positions. Earlier in  § 1.2.2,  it was shown that morphologically derived forms can in final position have up to consonants clusters as in (7). (7)  cül- = qsxtxv  ‘you disconnect his end of it’  1 c .f. Czaykowska—Higgins, 1993a for a simiLar cLaim regarding syLlables in Nxa’amxcin.  2 T he dot (.) represents the syllable boundary.  110  six  (6)  disconnect=head-BENE-TRANS-2-3  The syllable nucleus position can be filled by either a vowel or a syllabic nasal. Thompson and Thompson (1990) state liquids in word-initital position are usually syllabic, semi-vowels in the same environment retain their semivocalic quality, and if syllabic must be preceded by a glottal stop. With their limited distribution, syllabic nasals only occur in multisyllabic forms, never constituting the sole syllabic nucleus in a stem. Shown below in (8) is an example which has a syllabic nasal. (8)  6  6  u  n rn s  A /f  ‘he found/finds it’  Only vowels are stressed; syllabic nasals 3 are never stressed. Although I earlier considered proposing that the initial nasal is licenced by extrametricality, I will represent the nasals as being moraically licensed consistent with Thompson and Thompson’s (1992) analysis. Consonantal clusters with and the  INSTRUMENTAL  LOCATIVE  prefix, n-,  suffix, -tn, at stem-initial and -final position respectively are very  productive, and thus, occur frequently. As shown in (10), both affixes are syllabic nasals. (10)  6 L.  6  L  I  n  6 /r\  I,  /L  fL IL  6 L  il/Il p  tiflus  tn  n-pt=Inus-tn  ‘way of thinking’  LOC-?intervene=mind-INST  Extrasyllabic segments are represented with the angled brackets. (11)  V  C<C>  VC<C><C>  me nonstressability and distributional limitation of nasals are undoubtedly relevant to foot structure, 3 but this will require further study  111  In conclusion, the N+e9kepmx representation is maximally a bimoraic syllable [p4iJó. As will be motivated in the remainder of this chapter, this corresponds directly to the maximal template.  5.3 Augmentative Reduplication The Ne9kepmxcin augmentative reduplication process presented and discussed earlier in chapter 3 will be further examined and represented prosodically here. In the subsequent sections, I present several hypotheses for discussion.  These are in regard to the  augmentative reduplication prosodic base, reduplicative template, extrametricality of the prosodic base and template. I also discuss ensuing phonological processes that come into effect as a result of the augmentative reduplication and morphophonological outcomes.  5.3.1 Prosodic Base Based on the data shown in chapter 3 (e.g. [s-ma+-mü4-ec] ‘women’) I postulate the prosodic base of the N+e9kepmx augmentative reduplication is the morphological root. Upon initial analysis, the augmentative operation appears to be an ‘affix’ surfacing either at the beginning of the stem or after the first consonant of the stem. Therefore, the first question is when the initial C should be outside of the reduplicative domain. The first argument here is in regard to an initial [s].  Is the initial [s] a part of the  morphological root, or is it a prefix? Interestingly, the initial [sJ of a stem can be either a prefix or the initial consonant of a root these will be shown below. As shown by Thompson and Thompson (1992), the nominative prefix is an [s-].  Now  on the other hand, the [SI can also participate in the reduplication process. The two variable forms representing each type is shown below as (12):  112  (12) a.  JseplIl  b.  %fSéfl  sop-seplIl san-sen  s-fmü4’ec  s-ma+-mfrec  Consequently, a morphological distinction on the status of the initial [s] must be made between the two reduplicated forms. Again, the examples above and their reduplicated forms are: (13) sap-%fseplIl  (14) san-Jén  AUG-bread  (15) s-ma+-fmülec  AUG-pine cosmetic  NOM-AUG-woman  Thus, the reduplicated prosodic base targets the morphological root, excluding any prefixes. Another argument to maintain the fact that the base is a morphological root is discussed here where I argue that the base is not the syllable.  Using the same data as evidence  particularily, the third form above (15), where the base is the root. I maintain that if the prosodic base were a syllable, the form would not be as shown below in (16) but rather as shown in (17): (16)  Em  (17)  base  augtemp  {L trfl  l”é  base 6•  6  u  *  /cN  L  Sm  base  I  augtemp _6  ii  é c  smu  base• 6  6  m u 4- ec  Additional data show that the augmentative reduplication of CCVC and CRVC roots provide further evidence that the prosodic base is indeed the morphological root. In fact, it is clear that the base of each form includes the initial premoraic consonants of the base or the root.  113  (18) CCVC bases: +k’iw’-ix  1k’-+k’Iw’-ix  ‘they climb’  climb-AUTONOMOUS  climb[AUG]-AUTONOMOUS  qWxp  qwxqwxop  snowbrush  AUG-snowbrush  ‘snowbrushes’  tüt  tu9-t9i49]t  ‘little boys’  little-boy  AUG-little boy[DIM]  Thus, the following (19) is a prosodic representation of the augmentative reduplication of CCVC roots. (19) base 6  /iLJ qWx  aug. template 6  base  qWx  qWxp  [1 I  With CCVC bases as shown below in (20), reduplication copies the initial two consonants of the morphological root, irrespective of prefixes to the left. (20)  n-put  n-p1-plit  LOCALIZER-priest  LOC-AUG-priest (English loan)  However, note the following in (21) where a compound form has a lexical root  Cv?/  meaning ‘much, many’ and the ligature I/  concatenating it with the second lexical root /k’k’/ ‘dirt’. Note that reduplication applies to the rightmost lexical root. (21)  (22)  ‘lots of dirt on them’ many-LIGATURE-AUG-dirt  aug. template 16 I I\ I IiFLP  base 6  I”—’  Lk’k’I Thus, I claim from the above examples the base of the augmentative reduplication is the morphological root, not a prosodic base.  114  5.3.2 Reduplicative Template The augmentative reduplication forms on CCVC bases below establish that the template is a bimoraic syllable [p4.u prefixed to the base. Further, unlike the base, the onset of the reduplicative prefix is a single consonant. (23) CCVC bases: +k’Iw’-ix  4k’4k’Iw’-ix  climb-AUTONOMOUS  climb[AUG]-AUTONOMOUS  qWxp  qwxqwxwp  snowbrush  AUG-snowbrush  /t9iit/ t9üt  4 tu9-t9i[9]t  little-boy  AUG-little boy[DIM]  it 9 x  x9-x9i[9]t  ‘little boys’ ‘scout/raider’  AUG-scout[DIM]  One example below (24) shows that CRVC bases behave the same as CCVC bases. That is, the base of these canonical types includes both the initial CR premoraic (or prenucleus) consonants. (24)  n-put  n-pl-p1it  LOC-priest  LOC-AUG-priest (English loan)  The analysis below shows how these behave in similar fashion: (25) base  aug. template  6  6  put c9ex”’  pal 9 ca  //r?  /±  base 6  p1 it c’ éx  5.3.5 Subsequent Phonological Processes Several phonological processes occur in conjunction with the reduplication operations.  Little boys a  []  appears to be an anomalous in that the reduplicative form has a  as expected.  115  [U]  instead of  Mechanisms in effect are deletion, Obigatory Contour Principle, vowel colouration, glide vocalization and stress assignment. First of all, some variation of the [CVC] template is evident with forms that have deletion of segments. The data below (26) appear to require a different template [CV-] for the augmentative. Instead, I present them as evidence that the Obligatory Contour Principle (OCP) is in effect, as shown by the starred outputs. (26) Obligatory Contour Principle effects: a. cItx” ‘house’ sg. ci-citx”  ‘houses’  AUG-house  *CjtCft)W  b.  9e-9uy-m’  ‘he laughs’  9e-9e’uy-m’  ‘they laugh’  ?AUG-AUG-laugh-?  e * u 9 e y..m 9 c.  s-nen’eke’  ‘orphan’  s-ne-nen’eke9  ‘orphans’  NOM-AUG-orphan  *snenneneke9 Thompson and Thompson (1992) in their analysis do not suggest how one can predict when the final C in the template will be filled or not, although they claim it will not be when the copied template CVC has identical consonants in the C 1 and C 2 positions as in (26b,c). This alternation in the copied domain is, I propose, a direct effect of the Obligatory Contour Principle: as stated by McCarthy (1986), “identical adjacent segments are prohibited”. Although this principle accounts for a CV- prefix in forms which would otherwise result in the concatenation of identical consonants, the form in (26.a) has merely near identical segments as in (27): (27)  a.  citx”  -->  cicItx” *citcitxW  116  b.t  c [-cor.rr]  [-com]  [coRoNi]  [cooi]  [+ STRIDENT] If the resulting form were  *Cjt.Cftxw  both {c] and  [t] would have [coRoN] features.  I  conclude then that these adjacent segments identical in manner and superordinate place would violate the OCP; therefore, the first deletes. A second apparent irregularity occurs with vowels, where alternations such as /e/  ->  []  are phonologically predictable. These are cases of reduction of the unstressed copied vowel (Thompson and Thompson, 1992) as shown in 28: (28)  fc’é+ c’+-c’é+  ‘chill’ ‘cold’  AUG- chill  Note in (29), the base has an unstressed, but unreduced vowel.  Here, the cyclical  reassignment of stress to the following stressable suffix leaves the [e] unstressed; but because it was stressed on a previous cycle the [e] does not reduce to [o] (Thompson and Thompson, 1990): (29)  qw94qwe4ncut  ‘invite people (plural) to accompany’  AUG-invite-DIRECTIVE-REFLEXIVE  Another variation occurs when the reduced  []  deletes.  In (30) a copied vowel has been  deleted, and the [y] has vocalized. (30)  Vowel deletion of the template: k’iy—t ‘still’ k’i-Ic’iy-t ‘be still, motionless’  The derivation for (30) is shown below: (31)  RooT: VOWEL REDUCTION: VOWEL DELETION: GLIDE-VOCALIZATION:  /I(’iy-1’Iy-t/ I(’y-k’iy-t )‘ y-k’iy-t )(‘ i-I(’iy-t [‘ i—Ic’Iy—t]  117  Other alternations include roots which are not ICVC/ but, according to Thompson and Thompson (1992), they propose /CVC’C/ roots exist. If they indeed are 2 3 V 1 C ’ (7C C roots, the vowel after the C 1 has been deleted as shown below in (32). The first vowel in the root is unstressed subsequently deleted after the application of the reduplication process. However, if the copied vowel in the template were to be deleted, the remaining cluster [#s pzp} would violate constraints. (32)  V deletion of 1 1 CV C ’QC roots: 9 s-pz-pzu ‘birds’ NOM-AUG-bird  Still other phonological processes occur, but in specific interconsonantal environments. That is, vowel colouration occurs in the environment before uvulars and velars. As shown in (33)a. and b. the reduplicant vowel, schwa, surfaces as pharyngeal.  [+PFwmGEu]  [a] preceding a  The reduplicant vowel in (33)c. also has been colored by the preceding  segments. That is, in this case the [round] feature has colored the schwa in the copied 5 form. (33) Phonological processes a. p’af-p’ác-t  ‘burned in several places’  AUG-burn-MDL  b.  ‘he broke my door down’ AUG-break=opening=house-DIREcTIVE-TRANSITIVE-3-ISG  c.  ép  ‘strawberries’  NOM-AUG-ripe=bottom  Note however, unlike (33.c), the form in (34) does not have vowel colouration of the copied vowel. One must therefore ask: does vowel colouration of [+round] only occur with [-cont] obstruents, or under conditions of left-to-right directionality?  Complicating these issues  further, observe that the stressed vowel in the base of the root /t’! ‘straight’ has been  root.  Thompson and Thompson (1990a) note that the SPECIALIZATION EFFECT has occurred on the resonant Cy] of the 5  118  affected by the [round] feature spreading right-to-left from a continuant.  When this  adjacency is interrupted by the diminutive infix [t], the vowel colouration no longer is applicable.  (34) a.  tvt6t  ‘straight, direct, accurate’  AUG-straight-IMMEDIATE  cf. b.  to”-t[t]-t  ‘correct, right, the right way’  AUG- straight[DIM]  Other phonological process occur with the reassignment of stress. The example in (35)a. has stress on the suffix; the  []  is thus in a non-stressed position but is not in a context for V  colouration, nor is it subject to deletion, presumably because it is preceding a resonant coda consonant. Compare this with (35)b where the unstressed  []  does delete if the following  resonant is functioning as an onset rather than a coda. (35) Stress effects: a. s-k’m-X’mk=éy’t  ‘nieces’  NOM-AUG-Jhiece=agent  b.  ‘m-’m=éke-s  ‘chop off/down branches’  AUG-chop=branch-30-3S  In contrast, the reduplicant vowel in (36) is also unstressed, but nonetheless exhibits V colouration. (36) Laryngeal effects: %fpu9 -p’u9-e-cüt 9 p’u  ‘continuously flatulating’  AUG-flatulate-RESULTIVE-REFLEXIVE  The vowel retention of (36) above is a result of the immediate adjacency of a is perhaps attributable to the rounding to [u] as an effect of the  [9];  its quality  [p].  In (37) two phonological processes have occurred: (37) Morafication: fcw’ cu9-cu9=qIn  ‘to hit s.o. in the head’  AUG-punch=head  First of all, the root vowel loses stress and then deletes after the suffixation of the lexical 119  suffix /-qin/; and then the [w’] vocalizes to [u9]. That is, the monomoraic [w’] splits into a bimoraic [u ] form. Glides can not take the nucleus position, thus the occurrence of the 9 phonological split. Both of the examples below, (38) and (39), have specializing effects occurring with the reduplicated forms. In (38) the shifted from the  [PHARYNGEAL]  feature of the C 2 consonant of the root has  Ill to the root vowel. Also, both the base root and the copied template  (38)a. now are glottalized, [1’]. At this time, I am unable to account for these effects. The movement of the [pharyngeal] feature in both examples in (38) is not copied onto the template. Consequently, the status or condition of this feature needs further discussion. Thompson and Thompson (1992) address the former by stating the roots have a pharyngeal feature, and they refer to the glottalizing effect as a specializing effect [‘]. (38) Special(izing) effects: k’l’-k’l’-mIn’  ‘scissors’  AUG-cut-sheet-INST  k’l-k’l-t-és  ‘cut s.t. up into many pieces’  AUG-cut-TRANS-?  In conclusion, I have argued that the appropriate analysis of the augmentative reduplication is (39): (39)  Base: morphological root Template: prefix to the left edge of the root a bimoraic [ i’p] syllable. 6 Extrametricality: non applicable  Derivational prefixes occur outside of the base and the template. Several independent phonological processes affect both the template and the base, that is vowel reduction, deletion, vowel colouration, morafication, stress reassignment and special(izing) effects.  5.4 Diminutive Reduplication The diminutive reduplication surfaces as a [-C] or [-CV] copy infixed into the 120  N4’e’kepmxcin form. Given that infixing can be more insightfully analyzed as suffixing to a prosodically defined base, the present analyses focusses particularily on the prosodic representation of diminutive reduplication process. First of all, I propose the prosodic base is a stressed [p] 6 where the onset can maximally have one consonant. Secondly, I propose the prosodic template is a monomoraic syllable, 6u, suffixed to the prosodic base, where the second mora of a stressed syllabic base is extrametrical. In the concluding section, I discuss the subsequent phonological processes relevant to the diminutive reduplication process. 5.4.1 Prosodic base The diminutive reduplicated forms as shown earlier in chapter 3 appear in several forms. Consequently, in identifying the prosodic base as shown below (39), several factors must be discussed. First of all, the base may be tauto- or hetero-morphological, and may involve either the root or a derived stem. In reviewing the diminutive reduplication process, it is evident that the base for the diminutive reduplication is not morphologically defined, but is instead prosodic, i.e. the stressed syllable of the stem. This is in fact the traditional claim of many other Salishanists; what needs formal definition in the present context is the form of the syllable. In current terms, the base is prosodically represented as a stressed monomoraic syllable, [p]6 as shown in (39). (39) Diminutive reduplication: a. qemüt qemü[m]t Jhat  b. c.  ‘small hat’  hatf DIM)  4ax=áns  +a9=á[]n’s  Jèat=teeth  eat=teeth[DIM]  is-t-és  9is-t-e[t]s  tease-TRANS-3-3  tease-TRANS-3-3  ‘a little person’ ‘plague, tease, torment s.o.’  121  cu-xI-cm-e  d.  cu-xI[x]-cm-e  ‘Do this form me!’  Each of the examples in (39) have targeted a different location of the stem.  For  example in (39a.) the target is the root, whereas in (39b.) the target is the lexical suffix. Now the example in (39c.) the target is an inflectional suffix (pronominal marker), and in (39d.) the target is another derivational suffix (benefactive). Thus, I conclude that the prosodic base is the stressed syllable. Also, I argue that the onset of the stressed syllable is a single consonant as shown below in (40). Recall from the discussion in chapter 2 that the Ne kepmx syllable is maximally 9 [i]6 or minimally []6.  The examples below have pre-nuclear consonantal clusters;  however, the templatic representation of the reduplicant must be restricted to only have one consonant in the onset as only the single C immediately preceding the nucleus is copied in the reduplication process. Consequently this is stated as a condition on the onset below. (40) Diminutive reduplication base: h] , onset = single C 6 a. c[9Jek ‘become cool’ c[9]é[9]k ‘get a little cooler’ b. s-%fpzü9 s-pzü[z’u]9  ‘(large) bird, animal’ ‘small bird’  Therefore, the initial C of (40)a. is not part of the prosodic base. Nor is the initial C [p1 of the root in (40.b). The nominative [s] in (40.b) is independently excluded from the base domain because of its prefixal status.These are prosodically represented as in (41). (41)  a. base  base 6  HR I  II c s-p  9ék  c s-p  DRT  zñ  ek _zu 9  Although I claim the stressed 6 of the stem is the base, a major issue is how to effect the infixing of the diminutive mora into the stressed bimoraic syllable, [p4.u]6. Since forms 12 Zt  can have a stressed syllable with a second , an explanation is required for the status of the second p. I propose the second p of a stressed syllable is subject to prosodic circumscription since the template is affixed immediately after the monomoraic stressed nucleus. (42)  domain of affixation site: a. prodically defined domain: stressed syllable b. suffixation: after first jj. of stressed syllable  5.4.2 Diminutive reduplication template In this section evidence is considered in regard to whether the proposed monomoraic template [j.i.]6 is suffixed or prefixed to the base. Several points will be discussed to maintain the suffixation hypothesis presented here. Current prosodic theory (McCarthy and Prince; 1986, 1990) claims reduplication templates must either be a prefix or suffix. For this analysis, I adopt the several claims and conditions as presented by McCarthy and Prince concerning nonconcatenative morphology. The first argument presented here deals with whether the template is a prefix or a suffix. In conjunction with the prefixation of the template additional phonological processes would need to be applied.  On the other hand, if suffixation occurs, the final output is  derived by fewer phonological processes; thus a simpler analysis can be presented. The prefixation analysis entails particular complexities with respect to stress behaviour. The major issue here is to determine how stress is copied, and once it is copied, how is it deleted in the base? Although I consider stress to be a separate process, it is significant that there have been no other cases where stress is copied. Consequently, I do not propose the copying of the stress from the prosodic base, although it is the target of the reduplication process. Therefore, stress cannot be deleted from the base or copied template. I conclude then that the reduplication template is a suffix. However, to substanstiate  12.3  the claim, I present the following derivations to evaluate whether the template is better treated as a prefix (43 ff.) or a suffix (48). Broselow (1993) argues that the template is a prefix; therefore I present the possible prefixation analysis first.  (DRT  =  Diminutive  Reduplication Template) (43)  Prefixation: base a.  DRT  [6  6  /  /pfl  [p 1 .t x”  éI  base  DRT  6  6  6  I’  I-’  b.  s c.  mñ + e c  s  base  6 J1L  Il Ii \  n é 4- m x  mu m ü DRT  66 qW4-j  d Ijt  11  q”’+i  n e  ‘little house’  x”  6  \N  4- e c  ‘small woman’  -  6  6  L  né  4- m x  Two derivations which illustrate alternative ways of treating the base V-deletion and stress reassignment phenomena: (44)  a.  Prefixation: ROOT:  fcitx’’  fmü+ec  DIMINUTIVE:  cic1tx”  BASE VOWEL DELETION:  Ci-CtX’  STRESS REASSIGNMENT:  cjctxw  mu-mül-ec mu-m4-ec mü-m+ec mfl-m’+ec  RESONANT GLOTrALIZATION:  s-mü4-ec [s-müm’4-ec]  NOMINATIVE:  b.  ouTpuT:  ci-ctx”]  ROOT:  fcItx”  DIMINUTIVE:  CiCItXw  BASE STRESS DELETED: STRESS REASSIGNMENT:  Ci-Citx” c1-cjtx”’  BASE VOWEL DELETION:  CI-CtX”  RESONANT GLOrrALIZATION: NOMINATIVE: ouTpuT:  n/a [ci-ctx”]  mi+ec mu-mül-ec mu-mu+ec mñ-mu4-ec mu-m4-ec mü-m’+ec  n/a ‘little house’ 12  s-müm’+ec [S-müm’+ec]  ‘little wnian’  A crucial question is why stress re-assignment applies and the base stress is deleted. Perhaps it would be simpler to assume copy of the stress along with the melodic material, then the deletion of the copied stress. Therefore, another prosodic representation and derivation would be as in (45): (45)  Prefixation: base  DRT  LEThTXW  -  L!1  Thus, two alternative revised derivations, assuming stress is copied, are presented below (46): (46)  a.  b.  ROOT:  ..fcítx  DIMINUTIVE:  cI-cItx’’  BASE VOWEL DELETION:  Ci-CtX”  ROOT:  fCitX”  DIMINUTIVE:  CI-C1tX”  BASE STRESS DELETED:  cicitxv  BASE VOWEL DELETION:  clctxw  Although the above analysis can be made to generate the right output, I do not adopt it, particularily since there is no principled way to account for the shift or copy of the stress. In contrast, these stress facts are not at all problematic under the suffixation hypothesis, as illustrated below in (47): (47)  Suffixation: a. base  base DRT  [H  F/Th’  c I t x”  b.  C  base 6  m  qW4-j n  é  -  c  i  base DRT 6 6  /1_il  S  I  [/ F 3 Nt t  X”  ‘little house’  6  /FJ\  + ec + mx  /I  m ñ m u q+ n é n e S  12  4- ec + mx  ‘small woman’ ‘small birch bark basket’  The derivation below shows subsequent deletion of the unstressed V in the reduplicant. (48)  RooT:  .fcitx”  DIMINUTIVE:  cI[ci]tx”  VOWEL DELETION:  Cl[c]tX”  As shown, in this analysis no ad hoc mechanism is necessary to shift stress. Further, the V-deletion in the suffix follows directly from its unstressed post-tonic status. Another issue is that the prosodic category which defines the reduplication template surfaces as either a monomoraic (or core) syllable [ji]6 or a single consonant [C]. Recall according to McCarthy and Prince’s (1986) framework, it is necessary to define the template as one of the prosodic categories. The issue therefore is the prosodic status of the single C reduplication cases. The suffixation analysis above offers a unified treatment of both the -CV and -C cases by positing that the template consistently copies and maps the [CV] sequence, and that a subsequent general rule of unstressed vowel deletion then applies to delete the vowel under certain circumstances, specifically unless it is followed by a laryngeal  An alternative approach would be to consider that the -C cases represent the general diminutive reduplication process; the template might then be defined as a single mora. But if the template is a mora, the diminutive reduplication process could not account for examples such as in (49).  The process would never allow for copying the vowel and  therefore result in unacceptable forms such as: [xe9] ‘high’ than [xe-xe] ‘a little high’ and [9es-c’e9] ‘s.t. laid out 9  ->  --> [*XéX9]  or  [*XX9]  rather  [*9esc’éc’9] or [‘es-c’-c’é’] rather  than [9es-c’e-c’e9] ‘s.t. small laid out’. The following derivation results in [*spzuz9] or [*s.pz..zu91 (49)a. rather than [s-pzü-zu-9] (49)b.  12  o  (49)  Suffixation: a. base  I  s pz u b.  9  ‘animal’  *[o1  >s p  zu  base  DRT  [Tj z  9  DRT *  ?  Ill9  s-pt ü  ‘animal’  s-p  -->  Note that because the vowel quality before the  \“1 9 zu  z [9]  is systematically identical to that of the  stressed vowel of the base, an epenthesis analysis would not account for this. Therefore, I conclude that the diminutive reduplication template is a monomoraic syllable, []6. Subsequent phonological processes occur in conjunction with the diminutive reduplication process, and these are discussed in the following section.  5.4.3  Subsequent phonological processes  In the previous section, I stated the deletion of the copied vowel of the template occurred in the reduplication process. Motivation for this derives from the reduplicative behavior of stems which have a laryngeal in the C 2 position of the stem. The copied vowel followed by a laryngeal is not deleted in the reduplicated forms.  Note the following  examples in (50):  (50)  xé s-pzü9 es-c’é  ‘high’ ‘animal’ ‘s.t. laid out’  -->  --> -->  xé[xe]9 s-pzü[zu]9 9es-c’é[c’e]9  ‘a little high(er)’ ‘small bird’ ‘s.t. small laid out’  The retention of the copied vowel with the [“] in the adjacent position is fully regular. Another phonological process is with phonemic forms which have a the second mora of the base. These segments vocalize to diminutive reduplication forms, as in (51).  127  [j9]  /y(’)/ or /w(’)I as  and [u9] respectively in the  (51) c’’óy’-t nu-néw’-t  ‘she sleeps’-->  c’’o[c’9i9t  ‘she takes a nap’  ‘there is a --> wind blowing’  nu9-né[n]u9-t  ‘there is a little breeze blowing’  After the deletion of the copied unstressed vowel, the /y’/ and 1w’! become [i9] and [u9] respectively. Since the final sequence -CG’t is not syllabifiable, the glide vocalizes, creating the necessary nucleus and retaining the glottalization independently (as was seen earlier in the  AUGMENTATIVE  reduplication of the root fcw’ in (37)). The two derivations in (52)  represent the vocalization of the glides discussed above: (52)  RooT/STEM: AuGMENTATWE: VOWEL DELETIoN: GLIDE VOCALIZATION: DIMINUTIVE: VOWEL DELETION:  /r’oy’t/ n/a n/a n/a co[1wO]yt [ç.W]yt 0 çW  GLIDE VOCALIZATION: {IWOIWI9t]  /new’t/ new’-néw’t nw’-néw’t nu9-new’t nu9-né[nejw’t nu9-ne[njw’t nenu9t] 9 {nu  Once the diminutive reduplication process has been applied, the copied vowel is deleted since it does not have a laryngeal following the stressed vowel of the target. This results in a sequence of segments which cannot be syllabically parsed; this necessitates vocalization of the glide. The /y’/  ->  [j9]  and /w’/  ->  [u9] both now serve as the nucleus of the final  syllable. In contrast, note the behaviour of the copied /y’/ in a stem which is vowel-final, as shown in  (53): (53) a. c’y’é b.  c’y’éy’  ROOT/STEM:  /C’y’e/  DIMINUTIVE:  c’y’é[y’e]  VOWEL-DELETION:  c’y’e[y’j  GLIDE VOCALIZ’N:  n/a  ‘small basket’  [c’y’éy’]  Here there is no stem-final glottal causing the copied vowel to remain, nor is there  1Z3  motivation for glide vocalization. Therefore, the above analysis demonstrates that diminutive reduplication template cannot be simply a consonant. To summarize, I have argued that the diminutive reduplication receives a unified and principled account under an analysis which claims the following:  (56) Diminutive reduplication process: base: stressed syllable of stem, onset = single C prosodic circumscription: a non-nucleur p at the right edge of the base is extraprosodic template: suffix a monomoraic syllable, 6i  5.5. Out-of-control reduplication A prosodic analysis of the out-of-control reduplication process detailing the prosodic base and template will be discussed here. It appears the prosodic morphological framework as presented by McCarthy and Prince (1986, 1990) does not adequately capture aspects of the N+e9kepmx out-of-control reduplication, particularily the template. This issue as well as other issues will be discussed in subsequent sections. I hypothesize that the base for out-of-control reduplication is prosodically defined as a stressed bimoraic syllable [ji.p.,]6 with the onset extraprosodic, and the template is a bimoraic syllable [qi]6 which is suffixed to the base. Each of these claims will be discussed in detail in the next several sections, as well as the phonological processes associated with out-of control reduplication, or those which work in conjunction with the reduplication process in question.  5.5.1 Prosodic base As stated above, I postulate that the prosodic base of the out-of-control reduplication is a stressed bimoraic syllable. Three aspects of the prosodic base of the out-of-control lad  reduplication are directly relevant here as earlier presented by Jimmie and Shaw (1992). The first area to be discussed is in regard to N4-e9kepmx roots which have initial C clusters. As discussed earlier, few CCVC exist; different canonical subclasses are shown in  (56). Occurrences of glottal stops  [9]  in second position (56)e. are the most frequent class.  What forms do occur with out-of-control reduplication indicate that the existence of onset clusters does not affect the affixation site, which is consistently after the first segment which follows the stressed vowel.  (56)  Out-of-control reduplication on CC-initial roots:  a.  f+k’Iw ..fstax f+céck  k’Iw-u-wiyx n/a 4-céc-c-k  ‘climb’ ‘snowb(r)ush’ ‘mother-in-law’  b.  fq’p  n/a  ‘rustle (onom); expletive;snowb(r)ush’  c.  fplIt  n-plIt-t  ‘become a priest’  d.  mlám  mlám-m  ‘heal, affect by supernatural spell, new dancer’  e.  C9VC ../k’9ek  )c”ék-k  ‘s.o. finally managed to arrive’  ENG.L0AN  The second type of data of interest are roots with a final cluster. As Thompson and Thompson (1992) state: ‘if a stem involves more than a single consonant after the stressed vowel, the affix is automatically infixal”. Data exemplifying this are given in (58): (58) Out-of-control reduplication on -CC final roots: C1tXW cit-it-x’’ ‘house’, ‘house finally becomes available’ 9écq’’  9éc-c-q”  ‘bake’, ‘s.t. was accidently baked’  ücs  üc-c-s-c-  ‘abuse’, ‘to be intentionally abused’  This suggests that the base is truly defined as a bimoraic syllable [p]6, where a maximum of one post-vocalic consonant can attach to the end of the syllable. The third type of data provides compelling evidence that the base must be syllabically 130  defined. Again, these facts are observed by Thompson and Thompson (1992) who state: “With longer bases involving suffixal stress, the out-of-control reduplication operates on the stressed syllable of the underlying form, regardless of the position and valence of the root involved”. In the forms in (58), a lexical suffix has attracted stress away from the root, and it is this stressed syllable which functions as the base for out-of-control reduplication. Once again, the syllable must be defined in terms of a maximum of one post-vocalic C.  (58) Out-of-control reduplication on stressed syllable of derived stem: a. ../1-a9 4a9x=án-n-s kn ‘I manage to eat (with difficulty).’ The above data suggests indeed that the base is defined not morphologically, but prosodically as the stressed bimoraic syllable. However, the issue of extraprosodicity needs to be discussed in that neither the onset segment nor any additional segment(s) at the right edge of the stressed syllable are ever copied. As shown below (ORT  =  out-of-control reduplication template) the onset <C> is  not copied; therefore I conclude that it must be circumscribed as extraprosodic, i.e. outside of the prosodic domain of the base.  McCarthy and Prince’s (1990) theory of Prosodic  Circumscription allows for a “constituent” at the edge of a domain to be designated as extraprosodic. In a case such as /cItxv/, this is possible as there is only a single segment functioning as the syllable onset. This single segment is interpretable as a constituent, and can therefore be designated extraprosodic at the left edge. Where this approach becomes problematic, however, is in cases like those documented in (56) above where there is a complex onset, e.g. (56.c) [plit] [n-plIt-t]. (60) Suffixation: base  base ORT  \\1  \  <c>  i t  W  <c>  i t  it  13,  x”’  ‘little house’  (61)  Suffixation: base  <p1>  [r j i  t  base ORT 6 6  <p1>  frj[ 1  t  i  t  ‘become a priest’  Unless the “onset” were given formal recognition as a prosodic constituent (which no standard version of moraic theory allows for), it is not possible to characterize the sequence <p1> as a single extraprosodic constituent. Since the data in (56) clearly establish that the appropriate descriptive generalization is that the “onset”, whether it be a single C or a cluster, is not copied in the out-of-control reduplication, these data therefore point to a serious inadequacy in the theory of prosodic circumscription in not being able to formally characterize this behaviour. Returning to examples like (60)  [cItx”]  ‘house’ which have a complex coda in the root,  the out-of-control reduplication 1 [cft[it]x” shows that the base must also be prosodically ] circumscribed at the right edge. That is, to account for the apparent “infixation” of the outof-control reduplication in between the two final consonants of the CVCC root, an additional segment beyond a single (moraic) coda must be prosodically circumscribed as outside of the domain of the out-of-control reduplication. To summerize the above discussion, it has been argued that in order to identify the base for out-of-control reduplication, both onset segment(s) at the left edge and any residual coda segments at the right edge must be considered extraprosodic, i.e. [<CC>VC<C>]. Although McCarthy and Prince’s theory of prosodic circumscription allows for a single constituent at an edge to be circumscribed, it does not accommodate the fact that an onset sequence (as in (61)) must be eliminated from the domain.  132  Quite independently of the theory of prosodic circumscription, an interesting generalization emerges from a consideration of the prosodic structure in (60) and (61). Under the assumptions argued for in chapter 2 that (i) syllables are maximally bimoraic, and (ii) onset segments link directly to the syllable node, then the following prosodic generalization is evident: the content of the base which functions to determine both what is copied and where the affix is positioned is precisely definable as all and only the moraic content. 5.5.2  Out-of-control Reduplicative template  Turning now to a consideration of the appropriate form of the template, the reduplicative suffix, as shown earlier in chapter 3, is either a single [-C] or a [-VC]. Although there are surface templatic variations, I propose that the template is Examples supporting the fact that the template is bimoraic syllable  [jiji]  . 6 [.qi]  suffixed to a  stressed bimoraic syllable where onset is extraprosodic are discussed below. First of all, I present data in (62) that appears to have a single [R] as the prosodic constituent. (62)  ném’-m’  ‘s.t. goes into s.t. unnoticed’  carry[OC}  LOC-Iook[OC]  es-cay-y-e 9  ‘how one appears (the way one cannot help looking)’ ‘stuck in upright position accidently’  STAT-stuck[OC]-MDL  “uiy-y=u9s  ‘happen to get together’  get-together[OC]-?topsurface  In chapter 3, I discussed these anomalous forms as well as the following in (63): (63) a. q”ác’-c’-e  ‘it gets full suddenly and unexpectedly’  fulI[OC]-MDL  b.  qwaccs4exw  ‘you manage to move  move[OC]-CAUS-TRAN-3-2  13  it’  c. 9es-c6+-I-e  ‘(of liquid) spurt suddenly’  STAT-spirt[OC]-MDL  In (63.b) the base is not stressed, therefore with the affixation of the derivational suffixes it is evident that stress must be cyclic. If it were not the case, the stress would be on the base. Czaykowska-Higgins (199b) presents evidence that there is a class of cylic suffixes including transitivizing ones. Thus, the example above (63.b) must be a cylic transitivizing suffix which triggers cyclic application of stress. If the template was indeed a single p (or a sole C ), the prosodic representation would 2 be as follows in (64) where an explanation would be required for the extraprosodicity not only of the onset but also of the first (64)  [ii].  Out-of-control reduplication with a single 2 C : * * base base ORT L  •1 <qw>  a Lz  <qw>  I  a  I  z  I  z  ‘perspire, sweat’  But as shown in a previous examples (61), the out-of-control reduplication template also surfaces as a bimoraic syllabic template. If the template is hypothesized to be a bimoraic syllable [Fq’] 6 which is suffixed to the prosodically circumscribed base, then there must be explanations for the deletion of the first mora (or copied vowel) and the syllabification of forms which have no nucleus. These will be discussed below with the supporting data.  5.5.3 Subsequent phonological processes: V-Deletion and Glide Vocalization The deletion of the unstressed suffix vowel of the template creates a pair of identical resonants in sequence. These are not geminates; in fact, the second resonant is realized as a syllable nucleus. As shown below:  13  ‘  (65)  Out-of-control reduplication: Glide vocalization of a single C : 2 base base ORT \N  -  1111  <c> (66)  I  LI  I  é w’  <c>  LI  H9  Hi  L é w’  u  ROOT:  céw’ ‘discuss’  our-oF-coNTRoL:  céw’-w’  COPIED VOWEL DELETION:  CéW’-W’  GLIDE VOCALIZATION:  C W’  ‘discussed’  9 u  These occur in cases where the prosodic base has the glottalized  [y’]  and [w’] in the second  mora position. Significantly, the bimoraic 6 template remains as a bimoraic syllable  [jiji],  that is either a [j9] or a {u9]. Cases with a non-glottalized [yJ or [w] in the second mora position behave differently. After the schwa deletes, the [y] and [w] vocalize (as above) to [i} and [u] respectively. The second p.. however remains empty, as shown below in (67). Given (i) that there is no other melodic material to fill the mora; and (ii) that the language does not have either long vowels or geminate consonants, there is no principled way of filling this second mora after glide vocalization. (67)  Out-of-control reduplication: glide vocalization of a single C : 2 base base ORT ILFL  c (68)  ui w  c ii w cw cêw-w  COPIED VOWEL DELETION:  CêW-W  GLIDE VOCALIZATION:  Cw-U  VOWEL COLORATION:  C UW-U  conclude  that the  II u 0  ‘done, made, fixed’  ‘do, make, fix’  RooT: OUT-OF-CONTROL:  Therefore, I (69)  H  out-of-control reduplication is as  follows:  Out-of-control reduplication: Base: stressed bimoraic syllable where onset and any non-moraic coda segments are extraprosodic Template: suffix bimoraic  syllable [p..p..} 6 135  5.6 Characteristic Reduplication On the surface, the characteristic reduplication (informally -CVC) is very similar to the augmentative (informally CVC-) reduplication process. However, the former suffixes to the base (cf. § 5.3), whereas the latter prefixes to the base. Also, the former is not productive, whereas the augmentative is very prolific. Let us consider the comparison more deeply, in the light of a formal characterization of the properties of the base, template, and subsequent phonological processes. 5.6.1 Prosodic base The prosodic base of the characteristic reduplication process is very similar to the augmentative except for a very major difference. reduplication is prosodically defined as the first  6 [FLp]  The base of the characteristic of the stem. As argued earlier, the  augmentative base is morphologically defined as the stem. Like the other reduplication processes, the characteristic reduplication also has surface variations. In most cases the prosodic base is a bimoraic syllable [iji]6, as shown in (70). (CT (70) Characteristic reduplication: a. cáq[cq]t  =  characteristic template)  ‘spruce grouse’  tame[CHAR]-IM  cep[cp]t  ‘s.t. that stains s.t. red’  smear blood-[CHARI-IM  b.  Base 6  Base 6  I I  IPI’ I  caq cép  CI’ 6  I I  caq cép  cq cp  t t  On the contrary, the following examples in (71) have what surfaces as a monomoraic syllable as the prosodic base.  13:  (71) Characteristic reduplication, -CC: ci[cy]=él’qs a.  ‘have new clothes’  new[CHAR=c1othing  ci-cy-éle=xn  ‘have new shoes’  new[CHARl4foot=foot  b.  base  base  CR 1111  II  H  cy coy  ii  cy ciø  cy  cOy  ele=xin éle=xn  I would like to argue, however, that the above examples do have a bimoraic root /fcoy/ which has undergone deletion of the root vowel [o]. Then the [y] has filled the nucleus, but in order to fill it the [y] vocalizes.  Thus, following Thompson and Thompson’s (1992)  analysis of roots, I propose that in cases where the realized base surfaces as an unstressed monomoraic syllable, the roots are in actuality !CoG! roots. The effects of these types of roots on the template will be discussed in the next two sections.  5.6.2 Reduplicative template I propose that the characteristic template is a bimoraic syllable [ii,u]6 which is suffixed to the first bimoraic syllable.  In the instances where the template is reduced to a  monomoraic template, as above in (71) and shown below in (72), subsequent phonological processes have come in effect. As in earlier examples, the empty mora has been fified by the vocalization of the glide after the unstressed copied vowel deletes and the [constricted glottis] feature fills the second mora, surfacing as a  [9].  (72) Characteristic reduplication: glide vocalization a. zéw’[zu9]-t ‘tiresome’ tire[CHARI-IM  b. base  base  CR t  II z é w’  Ii z é w’  zu  ‘  t  ‘tiresome’  Other phonological effects are shown below in (73). Here both the root vowel and the copied vowel have been deleted. The deletion of the root vowel (schwa) has caused the vocalization of the /y/ -> [i]. Similarly, deletion of the unstressed vowel in the reduplicative suffix has triggered glide vocalization there too. An epenthetic [hj fills the onset for the following syllable, since a vowel-initial lexical suffix has been attached to the previously vowel-final stem. (CT  =  characteristic template)  (73) Vowel deletion and h-epenthesis: a. ciø[ciø]=hé+x” ‘have a new house’ new[CHAR] = house  base  CT  li1i.i, ci0  I1P ci0  1I  J  h é + x”  ‘have a new house’  In summary, I conclude that the characteristic reduplication is as in (74): (74) Characteristic reduplication: Base: first bimoraic syllable of stem Template: suffix a bimoraic syllabe [pp.] 6  5.7 Affective Reduplication On the surface, the affective reduplication appears also to have a counterpart, the diminutive reduplication process, in that they are very similar. Both templates are the same except the affective prefixes to the base, whereas the diminutive suffixes to the base as shown earlier in this chapter. However, like the characteristic reduplication, the affective is not productive whereas the diminutive is very productive. The prosodic representation of the base and the template will be discussed and presented in the following section, along  13S  with relevant subsequent phonological processes. 5.7.1 Prosodic base I propose that the prosodic base of the affective reduplication is the lexical root. Although very similar to the diminutive, this is a very significant difference. Recall the prosodic base of the diminutive reduplication is a stressed monomoraic syllable. However, with the affective it is the first monomoraic syllable of the root.  (75)  9 t-tz-e  ‘arrow head’  AFF-protrude slightly-FORMATIVE  c-cp=ikn’  ‘sun goes down over mountain into ocean (sunset)’  AFF-fit in=back  p-py’éx  ‘very tired’  AFF-tired  In cases like the last form above, the base appears to have a complex onset, i.e. two consonants. Thompson and Thompson (1990) however represent the underling roots of the three forms above as (76): (76)  ..ftz  ‘protrude’  fcp cp-p  ‘fit in’ ‘get fitted into position,  ..fpy’ex  ‘tired’  INCHOATIVE’  Thus, I conclude that the prosodic base of the affective reduplication can be generalized simply as the lexical root. (77) Affective reduplication prosodic base: lexical root a. n-c-céw’s ‘exhausted, tired Out’ LOC-AFF-fatigued  n-t-tm=ltn  ‘there are no berries in the basket’  LOC-AFF-lack=harvest  13  b.  n c.  template  base  6  6  JI  fiji  ca  cew’s  template  base  6  6  6 Ii  n  I  ta  fiji  I  I ta  mltn  5.7.2 Reduplicative template The template, a monomoraic syllable [ji]6, is prefixed.  In virtually every case the  template has not been reduced to a single consonant as with other reduplication processes, i.e. the vowel in this prefix is never subject to unstressed V-deletion (unless the prefix is an initial resonant (see (83)). Although Thompson and Thompson (1992:115) claim that the ‘vowel of the prefix is most often reduced to Ia/in regular fashion, but in some forms variation is heard between /e/ and Ia! and a few cases show consistently [e]’. All my data have the reduced vowel [a] except for two forms which have an (78)  9 ta-tz-e  [e].  Forms derived from weak roots as shown below:  ‘arrow head’  AFF-protrude slightly-FORMAT WE  ca-cp=íkn’ ‘sun goes down over mountain into ocean (sunset)’ AFF-fit in=back  pa-py’éx  ‘very tired’  AFF-tired  xa-xw’éI  ‘trail’  AFF-road  n-ta-tam=1tn  ‘there are no berries in the basket’  LOC-AFF-lack=harvest  s-pa-plant  ‘skunk’  NOM-AFF-skunk  140  9 s-c-cu  ‘pattern’  NOM-AFF-do  9es--p  ‘it is pleated’  STAT-AF’F-pile flat  However, other forms which have the copied templatic vowel as a schwa are strong roots as shown below (79): (79)  c-ci’ik”-e  ‘by oneself’  AFF-flnish-FMV  n-c-céw’s  ‘exhausted, tired out’  LOC-AFF-fatigued  n-k-kIc-n’-cm-s  ‘he haunts me’  LOC-AFF-?-1-3  Again, as previously discussed similar processes occur here as with other reduplication processes. One of these is that the  [+Pi.iRmGEj  feature of a vowel is not copied. Also,  the unstressed vowel of the base is deleted if a strong suffix is attached. Further, there, are vowel coloration effects.  The following examples in (80) are  conditioned by the consonants in the immediate environment of the vowels in question. That is, (80)a. shows the rounding of the vowel in the environment of {+round] consonants, and (80)b. shows the vowel [e] in the environment of the laryngeal  [9].  However, (80)c. in  contrast with (80)a. and (80)b., the [e] is not conditioned by its environment. (80) Vowel coloration effects: a. kwukwe+cIn  ‘carry s.t. around eating it’  AF’F-grasp-LIGATURE-food  b.  9e-9üy-m’  ‘laugh’  AFF-laugh-MIDDLE C.  we-wIyx  ‘cry’  AFF-cry  In some cases, as Thompson and Thompson state, the affective form is the more common lexical form as in (81). That is, either the non-reduplicated forms are not independent lexical forms, or the non-reduplicated form is less often used. Alternatively, as with the  141  other reduplication processes, the non-affective form appears only with other affixes. Thompson and Thompson (1990) refer to some of the stems as affective stems. These stems, I presuppose, do not appear in the non-affective reduplication form. (81)  t-té  ‘no (NEGATIVE)’  AFF-no  The vowel is deleted in forms which have an initial resonant, as in (82). Note however that elsewhere the vowel is retained in the copied form before an obstruent. (82)  1wwóyt  ‘sleep erratically’  AFF-sleep  Still some forms as shown below (83) are semantically ambiguous: (83)  ‘empty house’ LOC-AFF-house-INSTRUMENT  n-c-cItx”-tn’  ‘winter village, group of several houses’  LOC-AFF-house-INSTRUMENT  In summary, I conclude that the template is a monomoraic syllable. (84) Affective template: temp  base 6 1 RI-  ii  we  wiyx  AFF  -cry  ‘cry’  These generalizations are stated in (85):  (85)  Affective reduplication: Base: first monomoraic syllable of root Template: prefix prespecified monomoraic syllable, [RI 6  5.8 Summary and Conclusions 5.8.1 Summary of descriptive and theoretical issues In conclusion, each of the reduplication processes is uniquely defined by: (i) whether the base is prosodically or morphologically defined; (ii) how extraprosodicity of the base  14Z.  is characterized; (iii) whether the reduplicative template is monomoraic or bimoraic; and (iv) whether this template suffixes or prefixes the designated domain. Several inadequacies with the theory of prosodic circumscription exist. As discussed earlier in  § 5.5.1, the designation of a prosodic constituent with the out-of-control  reduplication is problematic. The initial cluster of CCVC roots requires circumscription. Since an onset cluster cannot be designates as a single extraprosodic constituent, this cannot be formally characterized. A striking dichotomy is evident with the N4-e9kepmx reduplication once the prosodic representations of each of the prosodic bases and reduplicative templates are analysed. The generalization which emerges that the augmentative,  affective and characteristic  reduplication templates are all prefixed to the morphological base.  In contrast, the  diminutive and out-of-control are both suffixed to a prosodically defined stressed base. That is, all the prefixing reduplication processes attach to a morphologically defined base (the root), whereas all the suffixing reduplication processes attach to a prosodically characterized base. 5.8.2 Identification of Interrelated Issues and Residual Questions Several interrelated issues regarding the N4-e9kepmx prosodic analysis require further research. The primary issue is the representation of complex onsets (pre-nuclear) and codas (post-nuclear) segments. Particularily problematic are those that are in medial positions; although they do not occur frequently, they nonetheless occur. At present I am unable to take a clear position on this matter, since the status of the Cs at both edges cannot be explicitly motivated.  For example, if the prosodic representation of initial and final  clusters is simplified to a maximum of one onset and one coda by extraprosodicity, as in (86)a. and (86)b. below, then one might expect forms of type (86)c. 14  (86) a.  6  b.  ...  <C> CVC  c.  CVC <C>  <C> CV C <C>  The third example above, however, is really not an acceptable form since N4ekepmx roots like fCCVCC do not exist. Thompson and Thompson (1990) designate almost 400 N+e9kepmx roots as CVCVC, many of which have a  []  in either first or second V position. If the status of the  []  is  epenthetic, many of these designated 391 roots would require reanalysis, resulting in CCVC or CVCC root forms.  If this were the case, clearer evidence for the actual  prosodic representation may surface. On the other hand, since many N4e’kepmx stems can be shown to have glide vocalization as a result of schwa deletion, the existence of glides [y] and [w] must exist in the N4’e’kepmx root.  Particularily since the N4-e9kepmx  augmentative template can not be both CVC- and CV-; it must be either the CVC- or CV-. Earlier in this chapter I presented an analysis where I claim that the augmentative template is CVC-. If the augmentative template had epenthetic schwas, a problem becomes evident. That is, how would the Prosodic Theory represent the prefix for a CC- template with subsequent epenthetic schwa? This argument, however, does not supercede the fact that epenthetic schwas do exist. They must occur in some forms, primarily when there is an unsyllabifiable consonant cluster. Thus, I claim that there must be two different types of schwas in 9 N+e k epmx: (i) underlying schwas in roots and (ii) predictably derived schwas. Although this is a fundamentally important issue, further exploration of the status of the schwa is beyond the scope of this thesis and requires a detailed investigation. As stated in chapter 2, CCVC and CVCC roots were less than one-tenth of all roots designated by Thompson and Thompson (1990). With the CVCC roots which occur more  144  frequently (almost four times as many as the other type of root), apparent sonority violations occurred with forms which primarily were fricatives, therefore one must posit as with other languages, that there are consonant clusters which are acceptable. For the present, I adopt the following assumptions although they may be subject to further investigation. With the exception of [CR- and [sC- initial clusters in roots, I adopt a simple {CVC]6.  As there is no clear evidence for coda clusters functioning  phonologically, I take the cautious position. Another assumption is that underlying schwas do in fact exist. As well, I have assumed the existence of underlying glides, based on the reduplicative effects of forms which have a glide in C 2 position. kepmx reduplicative process has verified and clarified many 9 This analysis of N+e reduplicative issues, but much further work is necessary in several other phonological areas, i.e. syllable theory (including internal constituency, moraic representation), lexical phonology, stress, the sonority status of fricatives.  14  Bibliography Archangeli, Diana. 1988. “Aspects of Underspecification Theory.” Phonology 5: 183-207. Bagemihi, Bruce. 1991. “Syllable Structure in Bella Coola.” Linguistic Inquily. 22.4:589644. Bell, Sarah J. 1983. “Internal C Reduplication in Shuswap”. Linguistic Inquiiy 14.2:332-338. Borowsky, Toni. 1989. Structure Preservation and the Syllable Coda in English. Natural Language and Linguistic Theoty, Vol. 7, 1989. Bessell, Nicola J. 1992. “Toward a Phonetic Phonological Typology of Post-Velar Articulation.” Ph.D. Thesis, University of British Columbia. Bessell, Nicola J. and Czaykowska-Higgins, Ewa. 1991a. ‘The Phonetics and Phonology of Postvelar Sounds in Moses-Columbian Salish (Nxa’amxcin)”, ms. presented at the CLA, Kingston, Ontario. Bessell, Nicola J. and Czaykowska-Higgins, Ewa. 1991b. “Interior Salish Evidence for Placeless Laryngeals. NELS 22. GLSA: University of Massachusetts, Amherst. Pp. 35-50. Broselow, Ellen. 1983.”Salish Double Reduplications: Subjacency in Morphology”, Natural Language and Linguistic Theory, 1:317-346. Broselow, Ellen and John McCarthy. 1983-4. “A Theory of Internal Reduplication”, Linguistic Review, 3:25-88. Carison, Barry. 1989. “Reduplication and Stress in Spokane” International Journal of American Linguistics 55:204-214. Carison, Barry F. and Dawn Bates. 1990. “Modeling out-of-control reduplication in Spokane”, XXV International Conference on Salish and Neighbouring Languages, Pp. 73-79. Carison, Barry F. and Laurence C. Thompson. 1982. “Out of Control in two (maybe more) Salish Languages”, Anthropological Linguistics 24: 5 1-65. Clements, G.N. and S.J. Keyser. 1983.  c-v Phonology.  Cambridge, Mass.: MIT Press.  Czaykowska-Higgins, Ewa. 1993a. The Phonology and Semantics of CVC Reduplication in Moses-Columbian Salish. University of Montana Occasional Papers in Linguistics, No. 10. Czaykowska-Higgins, Ewa. 1993b. ‘Cyclicity and Stress Moses-Columbia (Nxa’amxcin), Natural Language and Linguistic Theory 11:197-278.  14b  Salish  Doak, Ivy G. “Another Look at Coeur d’Alene Harmony,” International Journal ofAmerican Linguistics, 20. Egesdal, Steven M. 1992. Stylized Characters’ Speech in Thompson Salish Narrative. University of Montana Occasional Papers in Linguistics No. 9.  Guerssel, Mohammed. 1986. “Glides in Berber and Syllabicity.” Linguistic Inquiiy. 17.1; 1-12. Haeberlin, Herman K. 1918. “Types of Reduplication in the Salish Dialects.” International Journal of American Linguistics. Vol. 1; 154-174. Halle, Morris and J.-R. Vernaud. 1987. “Stress and the Cycle,” Linguistic Inquiiy. 18.1; 45-84. Hankamer, J. and J. Aissen. 1974. “The Sonority Hierarchy”. In A. Bruck, R.A. Fox & M.W. Lagaly (eds) Papers from the Parasession in Natural Phonology of the CLS. Pp. 131-145. Hayes, Bruce. 1986. “Inalterability in CV Phonology”. Language 62. 321-352. Hayes, Bruce. 1989. “Compensatory Lengthening in Moraic Phonology,” Linguistic Inquily. 20:253-306. Hunt, Kathy. 1991. “The Status of Glides in Gitksan”, Paper presented at WECOL 1990. ms., University of British Columbia. Hyman, Larry M. 1985. A Theoty of Phonological Weight. Sciences 19. Dordrecht: Foris Pubi.  Publications in Language  Ito, Junko. 1986. “Syllable Theory in prosodic phonology”. Ph.D. Dissertation, University of Massachusetts, Amherst. Jimmie, M. N. and P.A. Shaw. 1992. “Out-of-Control Reduplication in N1-e9kepmxcin”. Paper presented at the 31st Conference on American Indian Languages, Society for the Study of the Indigenous Languages of the Americas (SSILA) in conjunction with the American Anthropological Association (AAA), San Francisco. Katada, F. 1990. “On the Representation of Moras: Evidence from a Language Game”, Linguistic Inquiiy 21:641-646. Kirkham, Sandra. 1991. “A Preliminary Analysis of Reduplication in Lushootseed: A Prosodic Approach,” XXVI International Conference on Salish and Neighbouring Languages, 241-257. Kiparsky, Paul. 1973. “Abstactness, Opacity, and Global Rules.” In 0. Fujimura (ed.), Three Dimensions of Linguistic Theoty. Tokyo: TEC. Pp. 5 7-86. 141  Kiparsky, Paul. 1979. “Metrical Structure Assignment is Cyclic.” Linguistic Inquiry 10.3;421-442. Kiparsky, Paul. 1985. “Some Consequences of Lexical Phonology.” Phonology Yearbook 2:83-138. Kuipers, A. 1974. The Shuswap Language. Mouton, The Hague. Marantz, 1982. “Re Reduplication.” Linguistic Inquiry 13.3:435-482. Mattina, A. 1989. “(V)C 2 Reduplication in Colville-Okanagan, with Historical Notes”. unpublished paper, presented at the International Conference on Salish and Neighbouring Languages. McCarthy, John. 1986. “OCP Effects: Gemination and Anti-Gemination”, Linguistic Inquiry. 17:207-263. McCarthy, John. 1991. “On Gutterals.” Manuscript, University of Massachusetts, Amherst. McCarthy, J. and Alan Prince. 1986. “Prosodic Morphology”, ms., University of Massachusetts, Amherst, and Brandeis University. McCarthy, J. and Alan Prince. 1988. “Quantitative Transfer in Reduplication and Templatic Morphology.” In Linguistic Society of Korea (ed.) Linguistics in the Morning Calm 2. Seoul: Hanskin. Pp.3-35. McCarthy, J. and Alan Prince. 1990. “Foot and Word in Prosodic Morphology: The Arabic Broken Plural”, Natural Language and Linguistic Theory. Vol. 8:209-283. Newman, 5. 1947. “Bella Coola 1: Phonology.” International Journal ofAmerican Linguistics 13-3:129-134. Paradis, C. and J-F. Prunet. 1991. The Special Status of Coronals: Internal and External Evidence. Phonetics and Phonology. Vol. 2. New York: Academic Press. Rubach, J. and Booij, Geert, E. 1990. “Edge of Constituents Effects in Polish.” Natural Language and Linguistic Theory. Vol. 8.3; 427-463. Selkirk, Elisabeth. 1980. “The Role of Prosodic Categories in English Word Stress”. Linguistic Inquiry 11:563-605. Shaw, Patricia A. 1991. “Syllable Structure in Northwest Coast Languages: Melodic Skipping in Nisgha”. Paper presented at the Conference on Linguistic Research at U.B.C. Shaw, Patricia A. 1992. “Templatic Evidence for the Syllable Nucleus.” In Proceedings of Nels 23. Amherst: GSLA  Shaw, Patricia A. 1993. “Minimal Prosodic Constituency.” Paper presented at WECOL, University of Seattle, Washington. Shaw, Patricia A. and Walsh, Linda. 1991. “Syllable Structure in Nishga.” ms. Steriade, Donca. 1982. “Greek Prosodies and the Nature of Syllabaification. Ph.D. Dissertation, MIT. Steriade, Donca. 1988. “Reduplication and syllable transfer in Sanskrit and elsewhere.” Phonology, 5:73-155. Teit, James A. 1898. Traditions of the Thompson River Indians. American Folklore Society, Memoir 6. Boston and New York, Houghton, Mifflin. Teit, James A. 1904. Unpublished lexical lists, texts, miscellaneous material. Boas Collection. American Philosophical Society. Philadelphia. Thompson, L and M. Terry Thompson. 1990a. Unpublished ms. University of Hawaii.  “The English-Thompson Finderlist.”  Thompson, L. and M. Terry Thompson. 1990b. ‘Thompson-English Dictionary.” Unpublished ms. University of Hawaii. Thompson, L. and M. Terry Thompson. 1991. Appendices to the Thompson Grammar and the Thompson Dictionary, forthcoming publication. University of Montana. Occasional Papers. Thompson, L. and M. Terry Thompson. 1992 (orig. 1986). The Thompson Grammar. University of Montana Occaisonal Papers in Linguistics No. 8.  van Eijk, J. 1985. The Lillooet Language: Phonology, Morphology, Syntax. Doctoral thesis, Universiteit van Amsterdam. Zec, Draga. 1988. Sonority Constraints on Prosodic Structure. Ph.D. Dissertation, Stanford.  14?  Appendix I: Me9kepmx root table (Thompson and Thompson, 1990b.) The following underlying root representations proposed by Thompson and Thompson are organized (i) in terms of canonical forms and (ii) into classes phonologically defined by the onset consonant(s). symbol  C,CV Co  CVC  ez  CCVC  CVCC  CVCVC  OTHER  ‘ké(n)  ‘áyn ‘écq’’  ‘éqe’  9j9 or  écon ’on 4 ‘ék” ‘élès  ‘él’  ‘etuq’’  ‘émayx  ‘éq’ ‘éh  9 é p ‘esx’’ ‘éyk  éwi 9 itom 9 Ic’om 9  ‘ék’ 9 é m  ‘ey’ ac’x 9  imec 7  ‘éyicq”  ixic 9  9 è lile’ 9 ??? ‘Mile ‘sxeh  9áJ áz  11  ‘ipn  IyOw 9  ‘ép’  ‘oys  9uyüxW  ‘ée  9jW  ew’ 9exw  ücs 9  9 ü poy ‘ispIn ‘upon  9uy9  ‘éy  ‘üq”e’  ‘éy’ ‘1k’ Ik’ ‘iq’  ‘amálà ‘élmon ‘épols ésne x 9 étwon ‘éyc’qe’  91k’lOxW 9Iswe+ ‘imoyx” im’on 9 jnowo 7 ‘istk 7itxVe  “C’  7 6 m 6s 9  h  ---  h?m  héh  héw’t  hécow  hékw  hék”i’  hem hem’ hen’ hey hIc’ hip hIy’ hCl hCp  henoy’ hèléw’ hIhè’  150  howéyi  symbol  C,CV  CVC  c  co cü(n)  ceq cex” coy cow cow’ coc coh cox”' cok” cal’  ca4corn con cop caq coq”' coq”  CVC  CCVC  CVCC  CVCVC  OTHER  cáz  c’és éx” 9 c  céwt céw’s 9 cok cIkn cItx”‘’  céteh céw’ek cokeJ comèh com’ox conuip con’és copo cowes cow’éx cken c’i4-on cImè+ ckté+  cénmn cépnI cétwi co+’il coq”ollm cowéx”t 49 ciqe ca?z  a1 9 c’ c’ a 1 m aq 7 c’ o1 1 c”  c’é+m’ c’éxt ) 9 c’ex( 9 c’om 9 c’i 9 c’iq’ c’iwq’ c’Cp’n’ c’üq”‘” 9 c’ul’  c’áleh c’áq’èn’ caqwa1  c’á m 9 os c’ák”è c’ályè c’ámà9q”‘ c’éyè c’èk’épè+ c’è+óy’s c’èpi9es 9 c’okolosp’u c’ok”la’ c’o+ax”' c’od-huy’ c’oolépè c’kIkIk  céc  cék cék”' cern cén’ cép céq céqW  céw céw’ céz  cik cik’”  cIq cIq”'  cI  cas cot  dy  co  cI’  co”' coy  cól  cok coc’  ciik”'  ciw  cC+ cCq cCt cItx”' ccc”'  cuw  c’  c’ac” c’ek 9 c’o  c’a-y  c’áp c’áq c’áq’ c’ék’  c’ak c’ok’ c’ak”' c’ol’  cékW  c’o4c’ok’  c’ép’ c’éq  c’am  ceqW  c’orn’ c’on c’oq’ c’oq”'  c’éw’ céx”' c’éy’ c’cl’  c’ok””o’l coloxW c’ol c’olIq’ c’olü”' c’ol’ox c’o+op c’o+I’  coqV  c’Ik’  c’amoq’  c’os  c’Ip’  c’om’éx  c’aw  c’Iq’  c’ome  c’aw’  c’Iw’  c’oqi’  c’áJ  c”?ál  c’oteh  c’é4’ c’érn’ c’én  151  c’èli’is c’ènéc’  symbol  C,CV Ca  cont’  ç  CVC  CVC  caxV  c’I  c’awén  c’ac  c’Ix”  c’aw’a+  c’ac”  c’iy  c’ax”e  c’ay  c’ic”  c’aaz’  c’az  c’k’  c’ayap’  c’aI  c’óq”  c’ayax  c’aV”  c’áJ’  c’ayaV”  c’  c’ük”  c’ayüV”  9 c’i  Cu1CW  c’ay’éh  c’üJ c’ám c’üm’ cuq\r c’us  c’imaq’ c’itam’ c’iwaq’ ” 4 c’ik”utk  CCVC  CVCC  CVCVC  OTHER  cam cal’ ç9m  c?fl çó  k  ka (UNR  ken kew kac kay ka+ kam’ kan kan’ kas kac kaxW kaz’  káh káz Ice’ kém’ ken kép kéw kéw’ kéx” key kéz kIJ kIc kik kI+ kin kix klc  éx 9 k  k’e’ k’e+ k’ew’ k’ac k’ac’ k’a4-  k’éIc’ k’ém k’ép k’és k’éw’ k’éx  k”w’  kéyk kéyx x 9 ki kIyxW  k’éct k’éy’ k’iwIc’  152  kálac aw’ 9 ké kéze’ képay’ képCw kètax” kM’ax” kataz kaxam kayè’ kay’ax”' kikax kikh kInat kayè’ kaz’ah  k’é’ay k’éce 9 k’écèh k’émè+ k’émCs k’ém’èc  kàwpüy ke’+és kèknIy kèle kéntih kéy’i’st kalákèk kélzam kaná’ye k9ltan kikxkay kali  k’a’é+ay’t k’èk’nim k’éyni k’atni k’+aqe’  symbol  k’ coil’  k”  C,CV Co ---  ---  CVC  CVC  k’aJ k’ok’ k’am k’am’ k’on k’an’ k’oq k’oq’ k’os  k’éX”  k’éy’ k’il k”il. k’ik’ k’ip k’ls k’It k’Ix  k’ot  k’Ix”  k’ax k’ax” k’oy k’ay’ k’ k’o+  k’Ic  k”oc k”ol k”'oJ k”al’ ’o+ M k k”om k”on k”on’ k”os ’ot M k k”a? k”ay’  k”ál’ kvac k”'át k”'á k”ác” k”'ék’ k”'él k”'én k”éw k”'éw’ k”éy k”éy’ k”ók’  k”oz k”iw  CVCC  CCVC  k”awk” k”ax”m’  k”U’t k”üsn’ k”usp  CVCVC  OTHER  k’étye’ k’olIc’ k’o4-a% k’opax” k’atox” k’oyox” k’oy’ep k’Isém k’itox”  k’owéye’  k”ocok k”ok’ok’ k”ayat k”'ètIh k”upoy’ k”upoy’ k’”üceh k”ümeh k”üye k”ÜZe  k”áü  k”am’ k”'o’unoy k’oha’q kVa1hélox k”'oltepIs k”ol’IqIq k”'a+tAzne k”'am’y’éx” k”oncá?tn kVIykes k”ük”Svfls  k”Is  k”Ul  k”it kwh kwiy kwu kw k”üm k”flm’ kwup  kwey  kwek  kwiyxw  k”ac  k9y’t  k””ok’ kwam  k”éy kw6k kwu+  kwow kwaxw kwoy  k”'Iq’ k”Iw kwjy  k”ák”ès kwaxweh k”atec kaxtrum kwene9 k”esjxw kVacok kwoloc k”ak’ak  153  kwenmeh rek’um’ye kwewe k”oc’y’és kwokjye kwomoyxw kwotnoy è 9 k”oy’1 kvulowxw  symbol  k”’ cont’  C,CV Co  CVC  k”oz  k’”oz’ k’”il kWUC  bc’ bo.y lop loq loq’” lo low lox loW lof lof”  4-  CVC  4-ac  4-ac  4-e” 4-ew’ +oc  4-ak  4-op’  +áy’ +éc +ék +ék’ +ék”' +ék’  4-oq”'  4-em’  4-oq’ 4-qW  4-en’  4-os  4-eq  4-ot  4-éw  4-ow  4-éx”'  4-ow’  4-é 4-1k 4-1k”' 4-1k’ 4-Im  4-o 4-OXW 4-OXW 4-of 4-ç.W  4-k  4-n  CVCVC  OTHER  k”’iuèx k”’üceh??  k”tlc’ k”ñk” k’”üp’ k”üy k”üz  lix li  4-op  CVCC  kW1y k”’ól  lap’ laq” law lá liq’ us lI lix 11w lIf ljfw lük”' léw  4-ok 4-ok”‘’ 4-ok’  CCVC  +k’iw  léle 7 1c’i’ 1óq”’i” lohéc’ lopoxw lopél lopix lo’ay file” lIsék  láa”x lep’oq’iks losopik 7 li’mI bIte  4-a”  4-ánec’  4-aq”‘’owt  4-éwt 4-otk”’ 9 +iz  +áoy 4-okép +om?k +oPoqW 4-otox 4-owéy 4-oyok”'  +àfáse 9 4-e”kép +o”pnte” 4-oq’éw’s 4-oq”'”tm 4-owéy’st  4-oyoq”'  +ü”qIm’  4-oc”’e9  4-céck  4-of op  4-ép  4-iq’ 4-jqW  4-lw  4-Ix 4-1 4-Iy  154  sbo1  c,cv  eve  CVC  CCVC  CVCC  CVCVC  OTHER  k”ek  k’émn k’eq’ k’ok” k’omk k’Iqt 9 k’iy k’ámk’ k’üwx k’iix”n’  k’àx”üs k’émox 9 Ic’epé k’oko k’okom’ k’ok”om’ Ic’om’oq Ic’opoq Icoqwe9 Ic’oq”’u7 Ic’oq’om Ic’oyIn k’oyIy’  Ic’áqne 9 Ic’ék’ye keqwu hI 9 Ic’e k’en’i k’o1én’tx” k’oné1tx”’ koqaqwne k’oq’ésk’ow Ic’o”qe’  Co 4cont’  +Iz 4-I 4-Ok’ 4-uc, 4-urn’  +üq’  Ic’  k’o  k’eh k’e+ Ic’ok” kokv k’ol Ic’ol’  Ic’om’ k’op k’oq k’oq’ koqN Ic’os k’ox  k’ál’ k’á+ k’6k ?c’ék’  Ic’Ok” k’On Ic’ép Ic’Oq Ic’éq’ Ic’eq”’ k’Ow k’éx Ic’éx’’  koyoqW Ic’ozéh Ic’oz’üt  k’éy k’éz k’Ik”  k’Ik’ k’o k’ay k’oz  9 k’u  Ic”iIc’èp k’Im’oq Ic’Inüx” Ic’Iq’e’  Ic’in k’Iq k’Iq’ k’iq”  kiXW+  kjqW  kjyoqW  k’Iw’ k’ix  k’uile’  k’i k’ük’  k’üp k’üs Ic’üy’  m  me  mal mel mac’ mok  maJ’ maq mat máS  máq”m  mát méc’x mék’p  155  méc’i 9 mé4-üs mOah meloq’”  máwè mék’è mémoyt méyè  symbol  C,CV Co  m cont’  CVC  CVC  ma+ mak’  mac’ méc’  man  mék”’  moq’’  mé+  moq’ mos mat max moe’ may may’  mék’ mém men méx  mac  mk’ 9 ?? mi  CCVC  CVCC  CVCVC  OTHER  mén’x meq’m métk’’ meyx moyc mim’x  macaz’ moc9k’ moc’ax  méyq”e9 mlèk”óze9 mac’éy(?)  moc’ox’”  moc’u+t  mo!akV moJám mak’aq’” mak’us monIp moqa’ masin matCi may’ax mayéw’ mazéc mc’eh mIc”’ay mits 1 miye’ mClè mC+èc  moktwar 1 u 9 moko mak”y’éc’n’ mok’qi’ mane mastén c’xn’ 6 max mlm mzèm mIce?q 9 miktC mómos mCk”lac mCmlaqs mCq”’è  naqaz né’èk” 9 nèyi napél naq’ém’ a 9 ni nIhe’  lp 7 nac nék”è 9 nén’eke néwze nace nahzik”e namIm(è)4niwé nCzélè  mülx mütm müyx  mé méy’ méy mik’ mix” mi”  mIy’ móc moe’ mól mCc’ milk’ milq mCs mily mCy’  m’  n  ---  no  m’an  ne 7 nah nam’ nan nap  nax  náq’  z 9 na  naq”  nst  né? née né(h) nak nék’ néq”  nCk”e9 nCyeh  nés  new  new’ néx flexw n1kV nIk’ nI4nIqV njz  nc’ flilk\V  156  symbol  C,CV  CVC  n cont’  n’  p  CVC  CCVC  CVCC  CVCVC  OTHER  put  pác”9 pet pétk” zm 6 p p1q pi9 pi9  9 paqwu  nuim nus  n’o  paz pen pok pokAI pol 4 po poq  páh pal’ páq’ páq pay pá pék”’  poqW  pé+  piys  po+á  pos pot  pek’ pen  püyt  pok’oq”’ 9 pone  pé p 9 l’we (e) 9 péke pénck pen’ze pétole’ pétk”è péy’è pocok+ pok+én poq”l’Ic  pow  pep  puys  poséx  posn’u+  pox” pox”’ poy  péq” pés pet  poték p010k”' potoq  poz’ow’x” ps 7 u 9 pI pIkcè  poy’  pew  p0tiw  pk’  péx”'  powén  polant  pm  pé  pi’  péy  poy’ox poy’ok”'  pu’  péy’  pozoq”'  pol’üx”e9 poayeks potékw+  pul  p1k p1k’  pozén  poz’ow’x”  pozoq”'  pst  t 9 pu  pe4’uis péyèq pe”èc 9 peyè poloc’ poliw  pIq pjqW  poz’é+  p1w pic  pték pkih plOXw  pozu’  p_1 pjW  pól po pó”'  pisui+ pIlb pish+  pac  pücIn  p6k”’ p64p6k’ pen  pütéy  pCs put pCXW pCy pCy’  157  pIy  symbol  C,CV  CVC  CVC  p’ek’ p’oc’ p’ok’ p’k’ p’l p’o+ p’on p’q  p’éc’ p’ék” p’ék”’ p’él p’ém p’éq’ p’es p’éw  pqW p’as  CCVC  CVCC  CVCVC  OThER  pén’I’  p’én’èq p’éyq’ p’lc’ p’oyoq  yen 9 p’é p’éy’+e(h) 9 p’oc’z’e p’ük”le’  qá1x’’  p’Ic’ pjqW  p’aw’  p’Is  p’  p’Iw  p’x”  p’ix  p’y’  p’ix’”  p’f  p’I p’u+ p’ük’ p’üm  q  ---  qc  qáy  qc’  qaz  qáyt  qal qáca’  qape nIm 9 qe  ql  qol  qayx’’  qàyém’  q+mmn  qo+  q+  qapc  qéwos  qowéw’+  qm  qom  qéck  qèmüt  qIqèc’kn  qp  qom’  qeyoq  qos  qap  qow qoy qoz  qaw qic qic’ q14qik’ qix”'  qoláq’ qolex” qolil qonoc’ qomut qow’éw qIqèk’  q”em  qI  q’  ---  q’o 9 q’oc’ q’ol q’om q’on’ q’op  q’ác’ q’áw q’á q’áy’ q’áz q’é+  q’os  q’éx”'  q’ow  q’Ik’ q’Ip’  q’amIn q’aq’ek”' q’aw’è q’apox” q’ém’ès 1 q’épe’ q’olox”' q’alo q’ollp’  q’ow’  q’iw’  q’o+ow  q’owéqè+  q’oy’  qjxW  q’omok’  q’yü’è  q’I  q’omo”  q’Iy  q’oto  q’ox”  q’am’n q’áwm q’Iy’m q’iyt  q’otuk” q’owIs  158  q’álq’ec q’áq’me’ q’áyonoc q’áyx? q’lze’ q’1èn’oy q’o+ax”'il q’onix”'èc’k’ q’opk”oleh  symbol  C,CV  CVC  CVC  CCVC  CVCC  q’ cont’  CVCVC  OTHER  q’ow’4q’oyá q’yq’ q’ylq’ q’yo  q’yi q’Ile’ q’ic’i? qW  qWoc  qwam  qW9ep  qwint  qWon(W  qWajxWe9  qwm qWon  qW  qw7ez  qwisp  qwse9  qwasqwy  qway  qw9it  qWjy+  qWte4  xw 9 qwayele  qwon  qwaz  qW9ut  qwyaw  qwaneyqwa9  qwas  qWec  qWxWop  qwyc  qwoseqw  qWozem  q”aléwe(9)  qWow  q”én’  qWgy  qWey  qW4j  qWy  qWey  qWoyey+  qwz  qwez  qWjlxqn  qWW  qWjc  qwoqweski9  qWu9  qwIc  qwoxwole9  qwIl  qWtjxa9  qwj1 q”in  qWik qwIm  qwIw qWjxW qWjW qwoz qWu9 qWu+ qWap qWny  qW  9 0 qW  qWW  qwom  qW  qW96y  qws  qW  qWgrn  qwa9q  qwe+t  qwgsew  qWcyes9  qWw+  qwoxaq  qwwm  qWw  qw6l  qwuls  qWoze9  qWoysqne  qWoy  qwe+  qWuys  qwuke9  qwjyye  qWy  qWIc  qwune9  qWistm.  qWW  qWIl  qwxw  11 qWj  qWun  qwuc  qwut  qWu+  qW9itxw  qwum qWum qWup  qwut qWuy  159  symbol  s  C,CV  ---  CVC  CVC  CCVC  CVCC  CVCVC  OTHER  sew  sé 9 s6c’ sék sém sen sep sew sew’ sex sex”' se  stá  sent séw’t séykV 9 sek sem’t seye sot 9 soz sCtu sCx”t sC”'m’  sé’iàq séme 9 sépon sohéw solok sok’ok”' sonèy’ soyop sayoq’  sélè lIs 7 se seyè sopoill sox’’let soywen’x”' jj9 5 sitist 9 sj4-c’u 9 sisqe  sisèk”  sisq’t  suyp  sic’om  sii”est  so’y sol sok sok’ sok” som’ sop soq soq” saq’ sgqW  se”'  sow  séy  sipoc’  sCwle  sow’  séy’  9 siqe  süest  sox  sic’  sCsek”'  sox”'  sik”'  so”'  sIl  say  sik’  soy’  sip  sac  sip’ sIq slq’ sIq”'  SIW SiXw  si sIw sIc sIc”’ sCc sCc’ sCk’  sum sup, sCq”' sCXw sCy  sCy’  áw’  ayn  áwas  ol’pix  ipi’ac ik Iy  ulàpsCs  160  symbol  C,CV  CVC  CVC  CCVC  CVCC  CVCVC  OTHER  t  t  tern tay tak’” tal tol’ tol tan tap taq taq”' tax tax”' tax  tá ta tah tok” ta+ tarn’ tap toq” tow’ ta toy”' toy tl tz tIk”' tIq”' tIq”' tix lix”' tiic t’ly  t9ix” tüt  ték+ ’ 4 tay’x’ trn4 tinx tüt tuiwn’ turn’x tuwx  tékarn teyih talax takéy’  tákt tak’’uJ’s ték”’x”’e téxkan teylirne méytés 6 t twon ászeh 7 ta iistk 9 ta tarnlIkeh tam4-Ik tamnéc’ tom’é!t toaw’t tay’mét tik”olus tkktI tü+kIst  t’éx”'  1 t’áqe’  t’aq”Ine 9 t’Iyne  waq’t wék’k’ wénx wéw’+ wéyc wéyk’ woc? wIyx wux”t  wáaz’ wéc’eh walirn woloq”' wolo wolIk’ wol’oc wok’ok’ wornéx wosak’ woton woyot waza ‘cvuq”ol  wéwlè wéykon wèw’y’ok wit 9 we wew’qt woc’mét wolk’ze wal’Iqt wotérntk nürnt 9 wI wIne’x”' wIy’e wiwe  toy toz taf tof”' t$  tin  tak”i  tomix”' tom’os tom’ax tawac’ towék tawép tawIt taxéc’ tax”ap  to tü turn’ tüp tüq tüq”' tüz  w  t’oq”'  t’á t’éw t’ük”'  wok’ wal won wop woq’ wow wax’’ wo wo”' way’ woz waf  wák”' wá wá+ wá wác” wáz wék’ wén wén’ wet wéy wéy’ wik wIl wj+  ex 9 w  161  symbol  C,CV Ca  VC  w cont’  CVC  CVC  CVCC  CVCVC  OTHER  xèc’an xalaV’ xaJix xamen’ xawé+ xayak xayal xixe’ xixix  xaxalélux” ixe 7 y’am’  wIs wjy’ wit! witpv  w,  x  x”  xa  ---  Xe 9 xac xac’ xak xa4xak’ xam xan’ xap xap’ xat xaw xay xaz  xé4xék’ xép’ xét xéw’ xéx xéy xéz xic xic’ xIk xik’ xl! xik’ xin xis xjt xiw’ xiy xiy’ xlz  xit  xen’x  xWa9  xák”  x”ux”  xVosl  x”èsit  x”anitmt  x’ac  x”áJ  7us t x’  x”’iyl xWu9t  x’’óse  esitl’àc t x’  x”a’iy  x’’itx”èc’eh  x”uy’t  x”acak  x”ak”’  x”itc’  x”al  x”itl  x’’ac’ak’  x”’aJ  x”itl  x”ac’aq  x”am  x”ém  x’a1ók’  x’’an  x”en’  x”alak”  x”ép  x”ayak”  x’’ap’  x”es  ip’is t x’  x”aq”  xWlc x’lkW xWlk  xixVek  x”as  x”at  x”ityak”  x”itleh  x”ay x”az  x”it  x”us  162  symbol  C,CV Co  CVC  Xw  CVC  CCVC  CVCC  CVCVC  OTHER  ast  án’ih àtoy’ cocoq oc’éc’  áy’wili c6k’oyt pléw’ye owelk” owéyx”  xáy  oc’oq’  oyopt  Ay’ áz éc  ploq o1éx”’ onoq’ onáq’  py’tném xozontak iépeh  XWul x’vum  cont’  x”up xwus x”üy  xwu+  cac’ oc ok ok’ cam ol ok’ om’ on’ rap op’ aq  ál ák’ áq’ ás áw’  ék” él 4c Ic’  apeh cotoq coték owéy’  coq’ os cow’  oy ay’  cin  oyex’’  Iq CIW’ ‘iy  cozén ozüm Ic’oq’  oz’  Iyaq’  4toq  xwoc  xWac  x”ayt  xWa9es  x”onak’oy  x”ac’ x”’al  x’Vah  x’”ast  x”ác’eh  x”oyiq  xVal  *9ms  x”ac’oq’  x”oy’lom  x”ok’oq’  x”oy’qe  x”am  xwan  xVaq’’  ?‘ow x”oy  x”áy  ?icyüs x”óx”lèméy’è x”üèk’  x”ay’ ?vaz  xWac  xm cwjc cwjc,  xWil xW1k  wIs cwocw  163  symbol  C,CV  CVC  CVC  CCVC  CVCC  CVCVC  OTHER  xwuc, cont’  x”ük’ x”üp wus  x”’üy  y  ---  yo’y  yém  yéc’ü  yécut  yak’  yew  ykém’  yó”aq  yam’  yew’  ymét  yq  yé  ywin  yéy  yói9  yew’ yox  yo yW yocW  s 9 yu  yüweli  yIq yjqW  yj  yjçW yjm  yip, yüq” y4qW  y’ali  y’  y’em  y’olép  y’64-oy’t  zolok” zok’oq’ zom’én zonok zonix”' zosow’ zox”ép zoyek”' zix”eh züx”eh  zolk”C zowést zix”eh  y’en y’ey  z  zè  9 ze zo  záq zác  zoh zok” zol  zéc zé zé zé+ zék’ zen zéw zéw’ zéx  zok’ zam’ zoq’ zos zos zow zow’ zox zox”' zo zoy  zéy zIk zIq” zIw  zo  zIx  zoc”'  zj  ZiXw  ZCC  ut 9 z  zelt 9 zom’  zux”  ZéXW  zCm zen’ zCq  164  symbol  C,CV  CVC  CVC  CVCC  CCVC  CVCVC  OTHER  711S zixw  Z  cont’  it:iy  z’  z’a--  các’  c cac cal Sac’ ck’ ci  ca4dan dan das Saw’ da Say daz  5Wal SWa+ S’’as Sway  c’il’  cal’px  cak’ can cap Sa cay dc’  dIl 51k’ SIm dIn’ dis 51w  59n’  d’’us  dWóyt  SW9al 5w9Iy d’Il’  d’yIxm  1 çW  dWay  dWoI  1.The theta only occurs with this particular form. perhaps it is a Haqomelein form. 2.Ir) is not a  Thompsor and Thonçson (1990b) note this is Spuzziin form,  N4-e’kepmx phoneme and in this form it is an English loan.  16S  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items