A comparative grammar of Xinkan

This dissertation is a comparative grammar of the four known Xinkan languages of southeastern Guatemala (Guazacapán, Chiquimulilla, Jumaytepeque, and Yupiltepeque). The goal of this grammar is twofold: to provide a thorough description of the Xinkan languages and to reconstruct Proto-Xinkan from which these four languages developed. Xinkan languages currently are represented by only three people, all of whom are second language users of the language. This grammar begins with an introduction to the language family, the past research on Xinkan languages, and the goals behind the descriptions. In addition to this introduction to the language, a typological overview is included which highlights and outlines the interesting typological phenonmena in the languages with specific references to sections within the grammar for a detailed analysis of each part of the language.

A COMPARATIVE GRAMMAR OF XINKAN by Christopher Rogers A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Linguistics The University of Utah December 2010 Copyright © Christopher Rogers 2010 All Rights Reserved Th e Uni v e r s i t y o f Ut a h Gr a dua t e S cho o l STATEMENT OF DISSERTATION APPROVAL The dissertation of Christopher Rogers has been approved by the following supervisory committee members: Lyle Campbell , Chair 10/27/2010 Date Approved MaryAnn Christison , Member 10/27/2010 Date Approved Rachel Hayes-Harb , Member 10/27/2010 Date Approved John Robertson , Member 10/27/2010 Date Approved Andrew Garrett , Member Date Approved and by Edward Rubin , Chair of the Department of Linguistics and by Charles A. Wight, Dean of The Graduate School. ABSTRACT This dissertation is a comparative grammar of the four known Xinkan languages of southeastern Guatemala (Guazacapán, Chiquimulilla, Jumaytepeque, and Yupiltepeque). The goal of this grammar is twofold: to provide a thorough description of the Xinkan languages and to reconstruct Proto-Xinkan from which these four languages developed. Xinkan languages currently are represented by only three people, all of whom are second language users of the language. This grammar begins with an introduction to the language family, the past research on Xinkan languages, and the goals behind the descriptions. In addition to this introduction to the language, a typological overview is included which highlights and outlines the interesting typological phenonmena in the languages with specific references to sections within the grammar for a detailed analysis of each part of the language. After these preliminary chapters, the grammar continues with a discussion of the phonological patterns and the reconstruction of the Proto-Xinkan phonological system. These chapters include a description of the Xinkan vowel harmony patterns, the glottalized consonant patterns, and the possible reconstruction of a glottalized fricative. These chapters conclude with a proposed family tree of the four Xinkan languages. Following this is a discussion of the morphology of the Xinkan languages compared side by side; this Chapter concludes with an analysis of the reconstruction of relevant aspects of Proto-Xinkan morphology. In these chapters all the grammatical categories and iv morphological processes known to be exhibited in Xinkan are surveyed, including nouns, adjectives, verbs and verb classes, inchoative derivations, valency changing operations, and nominal and verbal particles. Similarly, this grammar also includes a description of the main syntactic properties of the Xinkan languages and a reconstruction of the syntactic patterns hypothesized to be relevant in Proto-Xinkan. The description of the syntax focuses on clauses with and without verbs, existential clauses, imperatives, noun phrases, adverb phrases, relative clauses, complement clauses and conjoined phrases and clauses. This grammar is dedicated to the Xinkan community of Guatemala; especially to the last speakers of this language and those involved in the revitalization efforts. Thank you. CONTENTS ABSTRACT ....................................................................................................................... iii LIST OF TABLES ............................................................................................................. xi LIST OF ABBREVIATIONS .......................................................................................... xiv ACKNOWLEDGEMENTS ........................................................................................... xviii Chapter 1 INTRODUCTION ........................................................................................................... 1 1.1 Modern Xinkan ......................................................................................................... 2 1.2 Past work with Xinkan ............................................................................................. 6 1.3 Xinkan community ................................................................................................. 11 1.4 Xinkan linguistic affiliations .................................................................................. 19 1.5 Organization of the grammar ................................................................................. 22 1.5.1 Commentary on the reconstruction of Proto-Xinkan ....................................... 24 2 TYPOLOGICAL OVERVIEW ..................................................................................... 32 2.1 Basic sentence/clause word order ........................................................................... 39 2.2 Noun phrase word order ......................................................................................... 41 2.3 Relational noun or preposition phrases .................................................................. 43 2.4 Relative clause word order ..................................................................................... 44 3 PHONOLOGY ............................................................................................................... 46 3.1 Vowels .................................................................................................................... 47 3.1.1 Vowel length .................................................................................................... 49 3.1.1.1 Lexical specification ................................................................................... 49 3.1.1.2 Vowel length alternation ............................................................................ 54 3.1.1.2.1 Vowel length and verb transitivity. ...................................................... 55 3.1.1.2.2 Vowel length in Jumaytepeque verbal nouns. ...................................... 57 3.1.1.2.3 Vowel length with de-verbalizing suffixes .......................................... 58 3.1.1.2.4 Vowel length in plural noun formation ................................................ 60 vii 3.1.2 Vowel Harmony ............................................................................................... 61 3.2 Consonants ............................................................................................................. 70 3.2.1 Guazacapán ...................................................................................................... 71 3.2.2 Chiquimulilla .................................................................................................... 74 3.2.3 Jumaytepeque ................................................................................................... 75 3.2.4 Yupiltepeque .................................................................................................... 76 3.2.5 Consonant distribution ..................................................................................... 78 3.2.5.1 Word-initial consonants ............................................................................. 78 3.2.5.2 Word-medial consonants ............................................................................ 80 3.2.5.3 Word-final consonants ............................................................................... 83 3.2.5.4 Consonant clusters ...................................................................................... 86 3.3 Phonological Alternations ...................................................................................... 95 3.3.1 Glottalization .................................................................................................... 96 3.3.1.1 Glottalization processes ............................................................................ 109 3.3.2 Voicing of stop following a nasal ................................................................... 115 3.3.3 Nasal assimilation ........................................................................................... 118 3.3.4 Lenition to /h/ ................................................................................................. 120 3.3.5 Vowel raising ................................................................................................. 122 3.3.6 Glottal-stop epenthesis ................................................................................... 125 3.3.7 Consonant dissimilation ................................................................................. 126 3.3.8 Consonant deletion ......................................................................................... 128 3.3.8.1 Guazacapán consonant deletion ............................................................... 128 3.3.8.2 Other consonant deletion .......................................................................... 130 3.4 Syllable structure .................................................................................................. 130 3.5 Stress .................................................................................................................... 134 3.6 Orthography .......................................................................................................... 136 4 HISTORICAL PHONOLOGY .................................................................................... 139 4.1 Introduction .......................................................................................................... 139 4.2 Proto-Xinkan phonological reconstruction .......................................................... 141 4.2.1Consonants ...................................................................................................... 143 4.2.1.1 Labials ...................................................................................................... 143 4.2.1.2 Alveolars .................................................................................................. 146 4.2.1.3 Alveo-palatals ........................................................................................... 156 4.2.1.4 Velars ........................................................................................................ 159 4.2.1.5 Glottals ..................................................................................................... 160 4.2.2 Vowel changes ............................................................................................... 161 4.3 Summary of sound changes .................................................................................. 168 4.4 Subgrouping ......................................................................................................... 170 4.5 Conclusion ............................................................................................................ 173 5 MORPHOLOGY ......................................................................................................... 175 5.1 Nouns and adjectives ............................................................................................ 175 5.1.1 Nouns .............................................................................................................. 176 viii 5.1.1.1 Noun possession ....................................................................................... 176 5.1.1.1.1 Inalienable possession ........................................................................ 177 5.1.1.1.2 Alienable possession .......................................................................... 181 5.1.1.1.3 Inalienable/alienable possession. ........................................................ 184 5.1.1.1.4 Mass nouns. ........................................................................................ 186 5.1.1.1.5 Genitive construction ......................................................................... 186 5.1.1.2 Plural formation ........................................................................................ 190 5.1.1.3 Diminutive noun formations .................................................................... 193 5.1.2 Adjectives ....................................................................................................... 196 5.1.3 Other modifiers ............................................................................................... 204 5.1.3.1 Articles ..................................................................................................... 205 5.1.3.2 Demonstratives ......................................................................................... 207 5.1.3.3 Quantifiers ................................................................................................ 212 5.1.3.3.1 Numerals. ........................................................................................... 212 5.1.3.3.2 Abstract quantifiers ............................................................................ 215 5.1.4 Noun Phrases .................................................................................................. 218 5.2 Pronouns ............................................................................................................... 219 5.2.1 Independent personal pronouns ...................................................................... 220 5.2.2 Dependent pronouns ....................................................................................... 223 5.2.2.1 Pronominal prefixes ................................................................................. 224 5.2.2.2 Pronominal suffixes .................................................................................. 228 5.3 Verbs .................................................................................................................... 231 5.3.1 Verb classes and transitivity ........................................................................... 231 5.3.2 Verbal inflection ............................................................................................. 241 5.3.2.1 Subject-verb agreement ............................................................................ 241 5.3.2.2 Aspect ....................................................................................................... 244 5.3.2.3 Imperative inflection ................................................................................ 248 5.3.2.4 Tense ........................................................................................................ 253 5.3.2.4.1 Past tense. ........................................................................................... 253 5.3.2.4.2 Future perfect tense. ........................................................................... 255 5.3.2.5 Voice ........................................................................................................ 257 5.3.2.6 Mood and modality .................................................................................. 259 5.4 Relational nouns ................................................................................................... 261 5.5 Verbal particles ..................................................................................................... 263 5.5.1 p'e/p'eh directional ......................................................................................... 264 5.5.2 wa Optative ..................................................................................................... 267 5.5.3 Negative imperative particle in Guazacapán .................................................. 268 5.5.4 Verbs taken from Spanish .............................................................................. 269 5.6. Nominal particles ................................................................................................ 270 5.6.1 kumu ‗as‘ ........................................................................................................ 270 5.6.2 ti'i- / t'i- direct object ..................................................................................... 271 5.6.3 ‘i- reflexive in Guazacapán ............................................................................ 273 5.6.4 ki' .................................................................................................................... 274 5.6.5 kiki-/kih ........................................................................................................... 274 5.6 Question words ..................................................................................................... 276 5.7 Conjunctions ......................................................................................................... 276 ix 5.8 Derivational morphology ..................................................................................... 276 5.8.1 Derivations affecting noun roots .................................................................... 277 5.8.2 Derivations affecting adjective roots .............................................................. 283 5.8.3 Verbal derivation ............................................................................................ 290 5.8.3.1 Valency changes ....................................................................................... 290 5.8.3.1.1 Antipassive formation ........................................................................ 290 5.8.3.1.2 Voice changes ..................................................................................... 295 5.8.3.1.2 Increasing. .......................................................................................... 303 5.8.3.2 Verb nominals ...................................................................................... 310 5.8.3.2.1. Abstract verbal nouns ........................................................................ 311 5.8.3.2.2 Agent nouns . ................................................................................... 314 5.8.3.2.3 Patient noun. ....................................................................................... 316 5.8.3.2.4 Instrumental noun. .............................................................................. 318 5.8.3.3 Participles ................................................................................................. 320 6 HISTORICAL MORPHOLOGY ................................................................................ 324 6.1 Pronouns and pronominal affixes ......................................................................... 325 6.1.1 Pronouns ......................................................................................................... 325 6.1.2 Pronominal affixes ......................................................................................... 327 6.2 Bound morphology ............................................................................................... 328 7 SYNTAX ..................................................................................................................... 337 7.1 Syntactic alignment .............................................................................................. 338 7.1.1 Grammatical relations .................................................................................... 339 7.1.2 Semantic relations .......................................................................................... 345 7.1.3 Antipassive and verb agreement ..................................................................... 346 7.2 Simple sentence formation ................................................................................... 347 7.2.1 Sentences with verbs ...................................................................................... 347 7.2.2 Copular sentences ........................................................................................... 354 7.2.2.1 Temporal duration .................................................................................... 355 7.2.2.2 The zero copula ........................................................................................ 362 7.2.2.3 The copula uk'a ........................................................................................ 367 7.2.2.4 Existence .................................................................................................. 368 7.2.2.5 Equational construction ............................................................................ 371 7.3 Question formation ............................................................................................... 372 7.3.1 Yes/no questions ............................................................................................. 373 7.3.2 Content questions (wh-questions) .................................................................. 377 7.4 Preposing .............................................................................................................. 381 7.5 Negation ............................................................................................................... 383 7.6 Complex sentence formation ................................................................................ 387 7.6.1 Conjoined clauses ........................................................................................... 387 7.6.2 Serial verb constructions ................................................................................ 390 7.6.3 Relative clauses .............................................................................................. 394 7.6.4 Complement clauses ....................................................................................... 398 x 7.6.5 Adverbial clauses ........................................................................................... 401 7.6.6 Conditional clauses ......................................................................................... 402 8 HISTORICAL SYNTAX ............................................................................................ 406 8.1Overview of syntactic reconstruction .................................................................... 406 8.2 Xinkan syntactic reconstruction ........................................................................... 409 8.2.1 Syntactic alignment ........................................................................................ 409 8.2.2 Verb classes .................................................................................................... 410 8.2.3 Word order ..................................................................................................... 411 8.2.4 Nominal syntax reconstruction ....................................................................... 416 APPENDIX ..................................................................................................................... 418 REFERENCES ............................................................................................................... 435 LIST OF TABLES Table 1. Xinkan vowel inventory ................................................................................................ 48 2. Guazacapán consonants ................................................................................................ 72 3. Chiquimulilla consonants.............................................................................................. 75 4. Jumaytepeque consonants ............................................................................................. 77 5. Yupiltepeque consonants .............................................................................................. 77 6. Proto-Xinkan vowels .................................................................................................. 142 7. Proto-Xinkan consonants ............................................................................................ 142 8. Xinkan numerals ......................................................................................................... 213 9. Guazacapán pronouns ................................................................................................. 220 10. Chiquimulilla pronouns ............................................................................................ 221 11. Jumaytepeque pronous .............................................................................................. 222 12. Yupiltepeuqe pronouns ............................................................................................. 223 13 Y s as a (1908) ............................................... 223 14. Guazacapán pronominal prefixes for nouns and transitive verbs ............................. 224 xii 15. Guazacapán intransitive verb prefixes ...................................................................... 225 16. Chiquimulilla pronominal prefixes for nouns and transitive verbs .......................... 226 17. Chiquimulilla intransitive verb prefixes ................................................................... 226 18. Jumaytepeque pronominal prefixes for nouns and transitive verbs ......................... 226 19. Jumaytepeque intranstitive verb prefixes ................................................................. 227 20. Yupiltepeque pronominal prefixes ........................................................................... 227 21. Guazacapán nominal suffixes .................................................................................. 228 22. Guazacapán verbal suffixes ...................................................................................... 228 23. Chiquimulilla nominal suffixes................................................................................ 229 24. Chiquimulilla verbal suffixes ................................................................................... 229 25. Jumaytepeque nominal suffixes ............................................................................... 229 26. Jumaytepeque verbal suffixes .................................................................................. 230 27. Yupiltepeque nominal suffixes ................................................................................ 230 28. Yupiltepeque verbal suffixes ................................................................................... 230 29. Xinkan relational nouns ............................................................................................ 263 30. Xinkan question words ............................................................................................. 276 31. Xinkan pronominal reconstruction ........................................................................... 326 32. Xinkan prefix reconstruction .................................................................................... 327 33. Xinkan verb suffixes reconstruction ........................................................................ 328 xiii 34. Xinkan nominal suffixes reconstruction ................................................................... 329 35. Xinkan verb conjugation reconstruction .................................................................. 330 36. Xinkan verb class reconstruction ............................................................................. 331 37. Valency Changes Reconstruction ............................................................................ 331 38. Xinkan participle reconstruction .............................................................................. 332 39. Causative suffixes reconstruction ............................................................................ 333 40. Xinkan epistemic modal reconstruction ................................................................... 334 41. Xinkan verbal noun reconstruction .......................................................................... 336 42. Xinkan syntactic patterns reconstruction ................................................................. 410 43. Xinkan verb classes reconstruction ........................................................................... 411 44. Xinkan word order reconstruction ........................................................................... 411 45. Xinkan noun phrase reconstruction ......................................................................... 413 46. Verbless sentence pattern reconstruction ................................................................. 413 47. Complex sentence pattern reconstruction ................................................................ 414 48. Xinkan noun possession pattern reconstruction ....................................................... 416 LIST OF ABBREVIATIONS * reconstructed form ** ungrammatical form % uncommon, but grammatical < comes from > changes into <…> orthographic representation of an orginal source  becomes 1PL first person plural 1SG first person singular 2PL second person plural 2SG second person singular 3PL third person plural 3SG third person singular ADJ adjective AGR agreement AGT agent noun ANTIP antipassive ART article C consonant xv Ch Chiquimulilla Xinka CAUS causative COMP complementizer CON conditional CONJ conjunction CONTR contrastive emphasis DEM demonstrative DEP dependent DIM diminutive DIR direct object DUR durative (temporal duration) EPIST espistemic modal EQUAL equational EXCL exclusive EXIST existence FORM formal FUT future FUTEXP future expectation G Guazacapán Xinka GEN genitive GENPOSS indirect object possession IDUR imperfective durative (temporal duration) IMPV imperative xvi INC inchoative INCL inclusive INFORM informal INSTR instrument noun INTR intransitivizer IPERF imperfective IRR irrealis IV intransitive verb lit. literally J Jumaytepeque Xinka LOC locative N noun NEG.IMPV negative.imperative NOM/ACC nominative-accuative O object OPT optional PERF perfective PL plural PNT patient noun POSS possessive PRED predicate PRES.PART present participle PST past xvii QP question particle REFL reflexive REL relativizer S subject SG singular Sp. Spanish SUBJ subject TV transitive verb UNACC unaccusative UNERG unergative V verb; vowel VN verbal noun VP verb phrase Y Yupiltepeque Xinka ACKNOWLEDGEMENTS Many people and organizations must be acknowledged as aiding and supporting the writing of this grammar. First, this grammar would not be possible without the permission by both Terrence Kaufman and Lyle Campbell to use, analyze and interpret their unpublished field notes on Xinkan; they have both been very helpful. Of course, I could also not have completed this work without the support of the Xinkan community in Guatemala. I specifically would like to thank each of the last speakers of Xinkan in allowing me to record them and for teaching me about their language and culture; thank you Carlos Méndez, Juan Santos Benito, Nicolás Vásquez Hernández, Felix Hernandez, Herlindo Pablo Garcia, Pablo Esquite Garcia (fallecido), Raymundo Hernández Godínez, Angel Vasquez, a a , a s. s a a a, a a a a s X a a s a a ; his hospitality and interest in seeing this work completed are greatly appreciated. I would also like to thank the Parlamento del Pueblo Xinkan de Guatemala (PAPXIG) and all of the members of this organization for support on and participating in revitalization workshops, and giving me feedback on the grammatical sketch which served as a preliminar X a a a , a s s a a Z a a a s , s a a a , a , and Ever Benito for helping me get to know the Xinkan speakers and accompanying me on my visits. xix In the U.S., Lyle Campbell has been a dedicated committee chair and has guided this research thoughtfully and helpfully. I owe much of my current thinking about Xinkan and linguistics in general to him. I would also like to acknowledge the other members of my graduate committee: Mauricio Mixco, Rachel Hayes-Harb, MaryAnn Christison, John Robertson, and Andrew Garrett for their helpful feedback and cooperation. Their feedback has been invaluable in the writing and revisions of this grammar. Of course, all errors are mine alone. I also want to acknowledge the entire faculty in the Department of Linguistics at the University of Utah who has taught courses that have informed this grammar. Lastly, I want to thank Sarah, Alexis, Jacob, Joseph, and Elizabeth for putting up with my dedication to writing this grammar, where for days (and once for six weeks) they saw very little of me. Language documentation and revitalization is also about the support that we get at home. CHAPTER 1 INTRODUCTION This work is a reference grammar of Xinkan, a language family of four closely related languages situated in southeastern Guatemala. The goals which have been set to guide the descriptions and examples in this grammar are twofold: first, to prepare a reasonably comprehensive review of the synchronic grammar of each of the four Xinkan languages, and second to hypothesize through the comparative method the paths of diachronic development for each language from the common linguistic ancestor, Proto- Xinkan. The result of these objectives is a reasonably extensive description of the Xinkan languages, synchronically and diachronically. It is hoped that this work will serve those interested in language reconstruction and history, especially in Guatemala and in Mesoamerica. As such, it will be beneficial to the investigations of Central American languages. It is further hoped that this grammar will be beneficial to those people for whom a Xinkan languages is their heritage language, by giving them a sense of their language history, and providing them with at least some of the tools necessary to carry on the revitalization efforts currently underway. 2 1.1 Modern Xinkan Xinkan is a small, language family of southeastern Guatemala located in the department of Santa Rosa, having no known external genetic affiliations. There has been a long, but sporadic, history of minor descriptive work aimed at languages of this family. In most of this previous research, Xinkan has not been recognized as a family of languages, but rather thought to be a single language, which has been referred to as Xinka, Xinca, Sinca or Szinca. This grammar provides evidence that the Xinkan language varieties should be treated distinctly as independent languages. Interestingly, however, the origin of the term Xinka and its variants is unknown; it does not seem to derive from any known word of Xinkan origin1. This language family is comprised of four clearly related varieties: Guazacapán Xinka, Chiquimulilla Xinka, Jumaytepeque Xinka, and Yupiltepeque Xinka. These names given to the varieties of Xinkan correspond to local town names. In the remainder of this grammar the names taken from a s a a a -X a‖ s iate the four languages - this is in following with common practice in the literature. In some regards these varieties are so close diachronically that some might be tempted to consider them different dialects. However, the differences in the grammars of 1 It can be speculated that the name Xinka is derived from a neighboring Uto-aztecan language: Pipil. In this latter language xinka x=/š/ a s ‗ s, s, s ‘ However, there is no evidence to support this speculation. Nevertheless, the Xinkan languages have borrowed several terms from Pipil, indicating a span of contact between the speakers of the two language groups. 3 each of the languages are significant enough (especially in their morphology) to demonstrate that the four are indeed separate languages (see Sasche 2010:47-49 for research in support of the idea that these Xinkan languages might be a single language; see also Chapters 5, 6, and 8, here, for a discussion of the diachronic development in the Xinkan languages indicating their status as separate languages). The family diversified into unique languages despite the fact that they occupy a relatively small geographical region of Guatemala (see below). The different Xinkan varieties are mutually unintelligible among the speakers of the different varieties (Terry Kaufman and Lyle Campbell, p.c.). Of these four languages, Yupiltepeque became extinct shortly after the turn of the last century; some time close after 1908. Chiquimulilla has recently become extinct; the last fully fluent speakers survived to the late 1970s. Sachse (2010:58) on reporting about her personal fieldwork affirms that semispeakers of this variety of Xinkan were living as late as 2000-2003; however, it is not clear how the competency of these speakers is to be as Sa s ‘s , a a s s s a s to the Guazacapán community and not the Chiquimulilla community. Jumaytepeque is also essentially gone; there is one very elderly and infirm second-language speaker and one rememberer of the language who can recall a handful of vocabulary items. Lastly, there are two (semi)speakers of Guazacapán and three remembers. One of these semispeakers is quite efficient at remembering vocabulary and small amounts of grammar (amazingly, since it has been approximately 40+ years since he would have spoken Xinkan); he too learned this language as a second language. In effect, there are no viable, fully competent speakers of any of the languages today. 4 For some scholars, the terms extinction, rememberers, and second-language speakers are problematic and perhaps derogatory. However, their use here is not intended to indicate a negative judgment about speaker competence; rather these terms are employed as means of clearly classifying the vitality of the Xinkan languages. Remembers are those older members of the Xinkan community who are themselves not speakers of any of the Xinkan languages, but who remember a relatively recent historical time when the Xinkan languages were used in public spheres of communication. They are unable to produce novel sentences in any of the Xinkan languages but often make comments a a a s s as, ‗X a s a s sa X a Y‘ T s community members have memorized some lexical items and a few idiomatic expressions which they gleaned from native speakers of the languages. In a few of these cases this information was gathered from elder family members or community members who were able to use at least this much of the language. Second-language speaker (semispeaker), similarly, refers to community members who can on occasion produce novel utterance, but whose grammatical competence in the Xinkan language of his or her community is limited at best. The two community members of Guazacapán who are semispeakers never completely learned the language and are unable to use certain linguistic structures and patterns essential to the full a a a a T s s a s s - a s a s‖ as a a 1989 s members learned what they know of Guazacapán as a second-language while interacting with peers in informal social situations. Furthermore, this period of language acquisition in the lives of these community members was accentuated by an extreme environment of 5 linguistic intolerance where Xinkan speakers were discouraged from using their native language. These second-language speakers now play an important role in the community and they are important to this grammar, though they are limited in what they can remember and what they have learned. This is not to imply that the data from these second-language speakers are in any way less relevant to the study of Xinkan. These kinds of data show how the language is actually being used and provides information on the process of language attrition and obsolescence. However, the generalizations provided from the second-language data are different facts than those need for a comprehensive grammar of Xinkan and the grammatical competence of native speakers. Fortunately, however, reasonably extensive fieldwork with the last fully fluent and competent speakers of the three then surviving Xinkan languages was conducted in the 1970s by Terrence Kaufman and Lyle Campbell. Their fieldwork notes provide documentation for the grammar of the Xinkan languages as it was spoken at that time. However, this information was not published and made available. A large percent of the data in this grammar comes from these unpublished materials, recently databased and analyzed as part of the Xinkan project at the Center for American Indian Languages at the University of Utah.2 2 This Project was funded by a grant from the National Science Foundation entitled -X a , a ‘: T a a a D a Projects to Completion ‖ a a a a a a Indigineous Languages of Latin Aamerica (AILLA), and soon an online database will be online, hosted by the Center for American Indian Languages at the University of Utah. 6 1.2 Past work with Xinkan The oldest description of a Xinkan language describes Guazacapán (though see Sachse 2010: 76-77 for a different view) and was written by a priest named Manuel Maldonado Matos around the year 1770, entitled El Arte de la Lengua Szinca. His method of description followed the latinate model with the seeming goal of indicating how closely it resembled Latin as a putative ideal. The Latin model used to describe this Xinkan language resulted in some serious limitations in the usefulness of this grammar. Specifically, this grammar does provide some excellent information, but omits some ss a a s a a ‘s a a , a s a a a s s s academic circles. For example, this grammar makes no explicit mention of some of the most typologically interesting characteristics of Xinkan: vowel length, vowel harmony, glottalized resonants, and ejective consonants, and little specifically about the morphology or syntax. One can glean morphological evidence about the language based on the word-and-paradigm descriptions in the grammar (for example possessive vs. non-possessive pronominal prefixes) but the grammar itself does not provide any direct morphological analysis. That is not to say that the grammar is completely without use; rather it has merits in what it does provide, not in what it does not. For example, while ignoring many of the typologically unique features of the language, it does provide a measure of understanding about how a fluent Spanish speaker in the 1700s viewed the linguistic structures of this Xinkan language. That is, although it is largely incomplete in its descriptions, it is thorough enough that some information about the linguistic structures of Xinkan can be gleaned by the careful reader. Sachse (2001) is a reprinting of the original manuscript with a short historical introduction; this makes the often difficult reading of the text 7 easier T s a a ‘s a a s a s s a y and type, whereas the original was written in hand and is, at time, difficult to interpret. However, this reproduction is not widely available and completely unavailable to the Guatemalan community. The original 1770 manuscript is held in a private library collection in the United States, at the Tozzer Library of Harvard University. Furthermore, due to its orthography and old terminology, it would be necessary for a linguist to interpret this colonial grammar philologically in order for it to be made useful at all to present-day learners of Xinkan languages. Fortunately, Sachse (2010) has recently completed a careful philological description and analysis of this colonial grammar. The original colonial manuscript contains a grammatical description of approximately 108 handwritten pages and a 1300 item vocabulary list. Eustorjio Calderon (1908) provided a brief comparative description of Yupiltepeque and Chiquimulilla, with a few brief passing notes on other possible varieties of Xinkan. Similar to the Arte, described above, this is a good resource but it suffers from significant omissions. Specifically, this description is seventy seven pages long, of which only twenty four are on the grammar, of both Chiquimulilla and Yupiltepeque, given largely after the latinate model with listed paradigms and no analysis and is comprised mostly of a description of the phonology of the two languages. The majority of this grammatical description contains a vocabulary list of the two languages with greetings in Yupiltepeque, without mentioning the typologically interesting aspects of Xinkan grammar. Furthermore, the phonological description is written in a pre-modern Spanish-based orthography that leaves unrepresented many phonological contrasts in the languages, misses large amounts of the phonological system, and leaves 8 open many questions of how to interpret the sounds of extinct Yupiltepeque. There is very little that can be understood about syntax or complex clauses from this grammar (though what can be understood is included in this dissertation). It is useful, however, because both of the languages included in the description are now extinct and it provides most of the information that exists on Yupiltepeque Xinka. A description of Xinkan based solely on this grammar would be unsuccessful, unless the sole objective of the description is to highlight a few isolated words without much regard for authentic linguistic competence. Moreover, this grammatical sketch is only available outside of Guatemala to university libraries with access to the collections where it is stored. Walter Lehmann (1920) re- s s a ‘s 1908 description, with a few corrections to the orthography caused by typographical errors. Inclu a ‘s nformation, Lehmann provides some historical anecdotes and assumptions about the development of the Xinkan culture. Additionally, word lists from the field notes of at least one other researcher who worked with speakers of Yupiltepeque, were included by Lehmann (see Gavarrete and Valdez 1868), in an attempt to provide all then extant information. This work is especially important to the Xinkan community because it contains essentially all of the information known to exist about the Yupiltepeque language ( a a s ‗ s‘ a ‘s s Unfortunately, most of the community members are either unaware of this publication or do not have access to it. It is not even very accessible to scholars accept unless they are affiliated with university libraries which have a copy of the work. , a ‘s s as a a a a : s s to the Xinkan community because it is written entirely in German, and not Spanish 9 (though some earlier works are reproduced, a ‘s s , with the Spanish of the original sources). This makes this work linguistically inaccessible, as well as physically inaccessible. Otto Schumann (1967) attempted to write a grammatical sketch of Guazacapán. This work suffers from a number of inconsistencies, and has been the cause of much confusion in the Xinkan community. This is the case in part because the emphasis is on the ethnographic characteristics of Xinkan culture more than on the language. Unfortunately, this emphasis led Schumann to gloss over complex portions of the grammar and to use descriptions which are contradictory and misleading. While this grammar sketch is slightly more accessible because it was written in Spanish, it offers next to nothing apart from a short sketch and glossary. It also omits the typological characteristics present in Xinkan languages. Also, it was never published, rather was a licentiate thesis in Mexico, unavailable to all but a few who were able to find a copy when it was written. Between 1972 and 1979 Terrence Kaufman and Lyle Campbell worked with speakers of the three then living Xinkan languages. Their work has been the most exhaustive and provides information about all of the typologically interesting features of Xinkan languages. Their field work concentrated on words, morphology and grammar of the languages spoken in Chiquimulilla, Guazacapán, and Jumaytepeque (Yupiltepeque had already become extinct before this time). They organized the information gathered on slip cards in file boxes, later arranged to reveal comparisons among these three languages. However, this information has not yet been published and so is not available to members of the Xinkan community, though a practical grammatical sketch has been 10 completed (but unpublished) for the community with information taken from their field notes (see Rogers 2008) and the community leaders have recently been given digital copies of all these field notes. A comparative dictionary compiled from these notes will be completed in the near future and made available to the community, to scholars, and to the public generally over the internet. However, these field notes contain some obscure linguistic terminology, which impedes their usefulness. Lastly, Frauke Sachse (2004) wrote a grammatical sketch of Xinkan which was based largely on the speech of one of the second-language speakers of Guazacapán Xinkan and published as Chiquimulilla Xinka (a completely different language). Actually, Sachse (2004: 17) thinks that the difference between the Xinkan languages is trivial and consequently considers there to be only one Xinkan language: It should be noted that until now it has been thought that different languages were used in the towns of Guazacapán, Chiquimulilla, and Yupiltepeque. While indeed lexical differences and differences of pronunciation have been noted in the different towns, it must be understood that these [differences] in most cases are optional differences, and it can be observed that similarities and correspondences prevail, that is that [the Xinkan languages] belong to a single base. Following from this it was decided here to give a description of the Xinkan language in general and explain the differences in the places where they occur and appear necessarily (translation mine, CR).3 3 -S a as a a a s sa a s s a a a , Chiquimulilla, y Yupiltepeque se encontraban diferentes idiomas xinkas. s s a a as x a s as a a as variantes de los pueblos, se tiene que entender estas en muchos casos como divergencias facultativas, y se puede ver que las semejanzas y correspondencias pre a , s s a a s a a s a a s a , a a s 11 Consequently, one of the Xinkan governing bodies (see section 1.3) assumed the differences between these Xinkan varieties were not valid and labeled the grammar as being of Chiquimulilla Xinka. This grammar might have proven very useful to the community, but often confuses the information, has numerous inaccuracies, and is incomplete. While making mention of some of the more important features of the Xinkan languages, treating two of the languages as a single language has caused problems for the community by complicating the achievement of their revitalization goals. Furthermore this grammar is replete with errors in linguistic transcription and consequently distorts the phonological and morphological systems of Xinkan. Other than these somewhat larger descriptions just mentioned, relatively few academic articles have been published about the Xinkan languages. The articles that have been published provide initial explanation of the typological characteristics of these languages; however, like some of the grammars just described, they are all physically and linguistically removed from the language community. They are for the most part written in English and published in academic journals or books outside of Guatemala (see for example, Sapper 1904, Stoll 1886 and 1958, Rambo 1965, Campbell 1979, and Termer 1944). 1.3 Xinkan community Historically the Xinkan people came into contact with people from the Old World with the invasion of Guatemala by Pedro de Alvarado in about 1524 (see Termer 1948 a x a a x a s as s s a a sa a ‖ 12 and Sasche 2010: 36). Present-day members of Xinkan communities occupy the same territory on the Pacific coast of Guatemala in the departments of Santa Rosa, Jutiapa, and Jalapa as their ancestors did at the time of the Spanish invasion. However, initial research on place names indicates that the Xinkan speakers probably occupied a larger territory in the distant past (see Campbell 1978 and Sachse 2010: 42-7). While exact figures of Xinkan speakers at the time of the invasion or of current ethnically Xinkan individuals are difficult to pinpoint with any degree of certainty, there has been research conducted to this end. Guatemalan census numbers indicate that there are between 200 and 200,000 speake s -X a‖ a ; wever, this probably refers only to people who self- as a -X a‖ a , a X a a a s a language of their community (Kaufman and Campbell p.c.). Schumann (1967:11) claims that were 19,505 inhabitants of the towns where Xinka was spoken, though the number of fluent speakers is not mentioned. McArthur (1966) indicated that the number of speakers then was less than 200 and confined only to the older generation, and that Spanish is preferred over Xinka. Saville (1918:1) claimed in writing his survey of the language that there were only 5,000 speakers of Xinka. Calderon (1908:6) said that there were 7,500 speakers in 1890 scattered around the department of Santa Rosa. Maldonado (1770), the earliest record of a Xinkan language, does not provide information on the number of speakers in that era. What this information shows, despite the discrepancies, is that over the last 100 years there has been a sharp decline in the number of native speakers of Xinkan languages (see Sachse 2010:35-8 for a good overview of historical census figures of the Xinkan population). 13 A number of unsubstantiated hypotheses also indicate the Xinkan speakers pre-date the Mayan and Aztec cultures. For example, Brinton (1885:1) cites Stoll in saying a , -an investigation of their language might throw a new light on the migrations of the ancient inhabitants of that region,‖ and Brinton continues saying, - a s as s for believing that previous to the arrival of the Quiches and Cakchiquels on the plains of Guatemala that region was occupied by this nation‖ though he does not say what those reason are. Similarly Calderon (1908:6) considers the Xinkan people to be the original inhabitants of Guatemala dating to times before the Maya-Quiché and Aztec invasions. These claims are not proven and must be accounted as speculation until further research on place names and language contact within the area can be made. The map in Figure 1 shows the Xinkan region (circled in black) in relation to Guatemala and the rest of Mesoamerica. The map in Figure 2 indicates the four towns corresponding to the four Xinkan languages described in this grammar. Guazacapán lies at the intersection of the Guatemalan highland region and the Pacific Coastal Plains. Chiquimulilla is approximately 5 kilometers to the east, Jumaytepeque 35 kilometers to the north, and Yupiltepeque 69 kilometers to the northeast. In the Pre-Classic era (2000 B.C. - 250 A.D.) this area was on the trade route that connected Mesoamerica and lower Central America (Sharer 2006: 190). This area was a prime area for agriculture, trade, and considered one of the first regions of settlement in Mesoamerica (Sharer 2006: 220). However, it is true that a number of different ethnic groups occupied this region of Mesoamerica (Sharer 2006: 236), and the Xinkans represent only one of them. Interestingly, however, this is one of the least studied areas of Guatemala (and Mesoamerica in general) both ethnographically and 14 archaeologically (Nash 1967, Vogt 1969, Olson 1991:404, Estrada Belli and Kosakowsky 1996:29, Ichon and Grignon 1998:327). The Xinkan community does not form a unified and autonomous entity. That is, there has never been, since the time the Xinkas became known in the days of the Spanish invasion, a geographical or political unity among the Xinkan communities. In fact, it is only a consequence of empirical research that groups these languages together. Consequently, there is not a form of centralized government (other than the national Guatemalan government). However, recently, members of the Xinkan communities have organized themselves for the purpose of revalorization of the Xinkan languages and Figure 1. Map of Guatemala and Xinkan Region Guatemala 15 T s a a s a ‗ ass s‘ un by a couple dozen young adults who are descendants of the Xinkas but who know very little about their heritage. These young people meet regularly, with the few remaining second-language speakers and rememberers, attempting to get as much information from them as possible. They have mobilized to meet once a week on Sunday, to discuss the language and make goals that will help them reach their objectives. Unfortunately, however, and despite the efforts of a local Guatemala linguist, they are not aware of the limitations of the two kinds of speakers (i.e.,, rememberers and second-language speakers) and are confused by the often contradictory evidence provided. Furthermore, none of the Xinkan youth has training in Figure 2. Map of Xinkan Towns Represented in the Grammar Guazacapán Chiquimulilla Jumaytepeque Yupiltepeque Santa Rosa Jutiapa 16 linguistics, language documentation, or language revitalization (although one has recently started a degree towards this end). Consequently, they are unable to make informed decisions about the structures of the languages and how to proceed in meeting their objectives. This group of young people travels from numerous towns scattered across the region, including from Chiquimulilla, Guazacapán, Jumaytepeque, and Yupiltepeque, to meet every Sunday and work together on their revitalization goals. They have access to some of the information that has been gathered in the past but have encountered a number of difficulties in organizing their efforts. For example, until recently these young people did not know that there were four Xinkan languages and assumed that all the information they possessed represented a single language.4 In reality they have scraps of data from each of the four languages and have encountered a number of seeming contradictions. These contradictions have caused confusion in the progress about learning these languages. Additionally, Xinkan has a number of sounds, structures, and patterns that are foreign to Spanish speakers (e.g.,, glottalized consonants, the high central vowel, and verb classes; see sections 3.2, 3.1and 5.3.1 and 5.3.2 respectively), and while these 4 Separating a language from a dialect is, of course, largely an empirical categorization and so these young adults should not be expected to make the same classifications. However, the often different lexical items and morphological operations in the languages must be understood for what they are, for any second-language learner, otherwise there are contradictions and confusion. 17 linguistic elements can be learned, the Xinkan youth are daunted by them and avoid producing them, because they have not been trained or taught to use them. In order to help them gain the political attention they needed and to organize the Xinkan community legitimately, others members of the community have become involved, and two governing organizations have emerged. The Council of the Xinkan People of Guatemala (COPXIG) was formed and set out to organize Xinkan peoples. Officers were named and a skeleton structure was put into place that was to unite the Xinkan area and the heirs of the Xinkan culture. A local linguist (a Kaqchiqel (Mayan) speaker with training in sociolinguistics) has become allied with the movement and has aided the community with the formation of goals and objectives, though his efforts have been limited by the lack of training of the community and the lack of linguistic resources. Unfortunately, internal divisions within COPXIG caused a schism within the Council. The schism was based on governance policy and political power. The COPXIG organization called for representatives from each of four Xinkan towns, who would have a voice in decisions of the Council. The first chairman was thought to have abused his power by concerning COPXIG with more of the national politics than with local concerns. Consequently some of the officers wanted to reorganize the Council. When they were unable to do so, those opposed to COPXIG formed their own organization: the Parliament of the Xinkan People of Guatemala (PAPXIG), and it is this organization which is currently most interested in revitalization. The local linguist is now allied with PAPXIG. The Xinkan community has made one of their central objectives the revitalization of Xinkan. For them, this means that they are interested in revitalizing the language and 18 culture and understanding its historical roots. They have achieved national recognition in Guatemala, and have begun an elementary school program about Xinkan. This program focuses on the training of children and teachers in the Xinkan languages. This program, however, is limited in that it focuses on teaching about the language and culture rather than teaching kids to become fluent speakers. The community has attempted to get funding to meet their objectives and was marginally successful; however COPXIG in a a a s X ‘s Consequently, both organizations are left with volunteer resources only. The community and the young adults are very anxious to succeed, but lack adequate direction. In the classes each Sunday, one of the Council members teaches the class about some of the linguistic structures of Xinkan based on Sachse (2004). The s ss a a s s ‗ a a ‘ a a While this is encouraging to the learners, they are learning only a small portion of the Xinkan language (Guazacapán, actually). The phonetic description of the language makes no mention of the glottalized consonants and the morphological description seems exotic and impractical. In order to sort out these problems the community needs access to all of the available data and training on how to use it. This training would necessarily mean that they learn how to understand the data and how to use it to teach others. While none of the people involved in COPXIG or PAPXIG are native speakers of any Xinkan language, they are all descendants of people who were. The effort of the community to revitalize and learn about their language(s) and culture is hampered by the extreme state of endangerment of the Xinkan languages. As mentioned above, only two one of the four Xinkan languages still has any kind of speaker (Guazacapán and 19 Jumaytepeque) and those who count are the three second-language speakers mentioned above who learned the language fairly fluently as a second language over 40 years ago, and who have had little opportunity to practice the language within the last four decades. The second-langauge speaker in Jumaytepeque, in fact, is much less fluent in Xinkan than the speakers in Guazacapán, and is both elderly and infirm. Furthermore both of the second-language speakers in Guazacapán are quite elderly; the youngest being 84 and the other is 88t. Consequently they lack the stamina needed for prolonged work on the language in the form of interviews, field work, or simple conversations. 1.4 Xinkan linguistic affiliations Xinkan is not genetically related to any other language or language family; however, a number of hypotheses have been proposed attempting to group Xinkan with other known languages. Xinka, when it was thought to be a single language, was claimed to be related to Lenca (see Lehmann 1920: 727,767 probably relying on Brinton 1885:96), but this was refuted by Campbell (1978: 602-3, 1979:961). Note that in his proposal of genetic affiliation between Xinkan and Lencan, Lehamann also proposed to group other languages such as, Chontal (a language of Oaxaca Mexico, also called Tequistlatec), Chumash, Seri, Mohave, Hokan, and Jicaque. None of these latter proposals was ever taken seriously, though the Xinca-Lenca hypothesis was repeated widely. Also it was proposed to form a family with the so-called Alagüilac language (Campbell 1972, 1979). This proposal was, rather, one possible hypothesis and should be c s a a a s a a a s but has left no other trace of linguistic evidence. The hypothesis that includes it as being related to Xinkan is probably one more of economy, than actual evidence, a a 20 a a a a a a a a a s a a a as s s s a a a a a a X a languages being discussed here, though also possibly an independent a s a a a as , a (1908: 6) identified the X a a a s a a s -Y , Y , Y ‖ s s a s identified with a language spoken in the Northwest area of Mexico. It is a a a a s , a s sa , though it is possible he was referring to a group of languages in Guererro Mexico which were frequently discussed in colonial times, but are now considered extinct, and sometimes ass a T a a a , s s ss s a ‘s a s s , a ‘s s s s a s a s s s -Y ‖ a -Y ‖, a a a s a Y s in Guatemala. There is no linguistic reason to suppose this proposal has merit. Although there is no evidence suggesting genetic relationships to other languages, what is conclusive is that Xinka has borrowed words from a number of neighboring languages, particularly Mayan (Campbell 1972, 1978:603; Kaufman 1977:67). In Calderon (1908) and with the same information repeated again in Lehmann (1920), two additional varieties were identified as part of the Xinkan family; those of Sinacantán and Jutiapa. It is clear from the wordlists and the scant grammatical information in these sources that these were definitely Xinkan. However, because the information extant on these varieties is scant and because this information indicates a close affinity with Yupiltepeque, it is assumed that both these are the same as or are varieties of Yupiltepeque Xinka. They will be treated as one language here with the 21 caveat that it is unknown how mutually intelligible these varieties may have been with one another. Furthermore, Calderon (1908:5) claims that at the time of his investigations a a a s s a, - , s unsual and peculiar aspect of Chiquimulilla, and possibly the only such example in all of Guatemala, is that the barrio of the North Plaza has one unique language and the other South Plaza barrio has an entirely different language, distinct from the former.‖ (translation mine).5 No information is known about the difference between these two reported languages. As mentioned, the linguistic information on Yupiltepeque comes from essentially two sources: Lehmann‘s (1920) X a a a s a a ‘s 1908 grammatical sketch of Chiquimulilla with comparisons to Yupiltepeque. In his survey of the entire former Xinkan territory in search of potential surviving speakers, Lyle Campbell was able to elicit a handful of Yupiltepeque vocabulary from remembers, though there is no way to validate their information reliably because Yupiltepeque had become extinct around 50 years before. Consequently, it may be painfully obvious that there are large gaps of information on Yupiltepeque. There is nothing that can be done about this now, but the information that is available is very relevant to an accurate reconstruction of Proto-Xinkan. 5 - s x a a a, s Guatemala, es que el barrio de la plaza Norte tiene una lengua especial y el otro barrio de a a a S , a a a s a a a‖ 22 The languages which are closest (geographically) to the Xinkan languages a a a a : a K‘ a a s , ‘ ‘ a a : Cholan) to the far northeast, and Pipil (Uto-Aztecan: Nahua branch) to the immediate s as Ka a a : K‘ a s a s a , s a with the speakers of these languages is evidenced by the loanwords from each in the Xinkan languages. For example the word wünak (Guazacapán), winak (Chiquimulilla) ‗ , s ‘ a a a a Y , s a from Mayan winaq ‗ s ‘ T a a s s x -Zoquean loanwords in Xinkan, which are diffused throughout the majority of the Mesoamerican linguistic area (Campbell, Kaufman, and Smith-Stark 1986). Other languages may have once been in this geographical area and influenced the Xinkan languages, but any suggested contact is speculative as there is now no evidence of such contact. The archeological site at Chalchuapa, El Salvador, is directly to the west and adjacent to the Xinkan region. This site was Poqomam speaking at the time of the Spanish invasion, though Poqomam reached here very late, and the site is associated by some with speakers of the Cholan- Tzeltalan branch of Mayan (Campbell 1978). This site is considered one of the two largest Pre-Classic architectural sites; La Blanca is the second (see Sharer 2006:193). 1.5 Organization of the grammar The grammar is organized in eight chapters and an appendix intended to survey the complete synchronic and diachronic descriptions of the Xinkan languages. Chapter 2 provides a typological overview of the Xinkan languages. Chapter 3 surveys the phonology of Xinkan, where both the phonological inventory and segment distribution are discussed. Chapter 4 gives the sound correspondences of the Xinkan languages and 23 reconstructs a possible Proto-Xinkan phonological inventory. Chapter 5 describes at length the morphology of the Xinkan languages. Chapter 6 provides a reconstruction of the functions of this morphological system in Proto-Xinkan. Chapter 7 details the syntactic patterns, and Chapter 8 uses these patterns to reconstruct the surface syntactic patterns in Proto-Xinkan. Throughout the grammar examples are cited from a database created from the unpublished fieldnotes from the 1970s. In this database example sentences and lexical items are given a unique identifier indicating the year the form was recorded, the specific language, and the speaker providing the information. However, these unique identifiers have not been included in this version of the grammar due to two factors: the inaccessibility of the database itself and the planned changes to these unique identifiers in the published version of the database. Because the database is archived in a format that is inaccessible to those without the necessary software program, the database is not readily available to the community and other linguists. In the planned comparative dictionary which is to be published from this database, the format will be altered and changed to a more accessible one. In this planned change of formats the unique indentifiers in the database will be replaced by new identifiers that match the needs of the comparative dictionary better. In the published version of this grammar the location of each example sentence within the dictionary will be included. The citation information is not included here because there is little practical use in including them, as they will be changed, and because without the supporting resources this information would seem overly complicated. 24 1.5.1 Commentary on the reconstruction of Proto-Xinkan A few general comments about the historical aspects of this grammar are in order before proceeding. While the reconstruction of Proto-Xinkan is fairly straightforward, there are at least two sources of complication. First, there is sometimes great variation between speakers of a language and across individual speaker utterances with regard to the production of sounds, especially glottalized consonants. While variation is normal for all languages, due to the state of these languages, it is now impossible to check, or re-check, any of the data presented herein, thereby making the discussion of some patterns and constructions necessarily vague. Second, the quality and extent of the historical records often leave uncertainties in the exact specifications of sounds, meanings, and functions, especially in the case of Yupiltepeque. Neither of these complications creates insurmonuntable problems for the reconstruction of Proto-Xinkan grammar, but they are noteworthy because they can limit the amount of information that can be reconstructed. In particular, they limit the role Yupiltepeque can play, though this might have been greater if the sources were more extensive and more reliable. In order to overcome these difficulties as much as possible an extensive philological analysis and comparison of Calerdón (1908) and Lehmann (1920) has been completed. The information about Yupiltepeque in this grammar comes from that analysis, and is kept, where needed, in the orthography of the original sources. There are problems with using the Yupiltepeque Xinka data for reconstruction. Specifically, there is a problem of relating the Yupiltepeque Xinkan data to the other three Xinkan languages. For example, the data found in both Calderón (1908) and Lehmann (1920) are presented in a prescientific nonstandard orthography (written before the advent of modern phonetics and the phonemic principle), and consequently it is 25 difficult to know the phonetic value of the graphemes. This is especially true of words with glottalized sounds in them; for example in both sources there is no distinction made between /k/ an / ‘/, s < > a, u, o). This failure to distinguish these contrasts in the Yupiltepeque Xinka data results in difficulties for determining how to compare Yupilepeque forms with those from the other languages where the contrast between plain and glottalized consonants is clearly distinguished. This means that the Yupiltepeque data provide no reliable witness concerning glottalization in Proto-Xinkan, but these forms can be useful to highlight the place and manner of articulation of the consonants under consideration. A similar problem is found with long and short vowels, which are contrastive in Xinkan languages, but which are not distinguished in the Yupiltepeque sources; both are indicated in the sources in the same way, with a single vowel. Furthermore, Calderón (1908) lists a few words which begin with <b> but which, judging from their s s a a s, a a a s / / / ‘/ Here it is not possible to know whether this is a mistaken recording of /p/, which we suspect, or whether something else is going on that is now unclear to us. Consequently, due to these limitations of the orthography, it is difficult to give a complete depiction of the subgrouping of the fa a Y ‘s s This issue can be seen in the case of // and /l/ in Xinkan. Proto-Xinkan had a voiceless lateral approximant // which has changed to /l/ in Jumaytepeque in all environments, and is retained as [] in both Guazacapán and Chiquimulilla Xinka. The Yupiltepeque data represents this sound variably as <jl> and <lj> after the low vowel, a, and as <l> s S a , s s / ‘/ as a / ‘/ a a < > 26 Yupiltepeque. Taken at face value, it would seem Yupiltepeque and Jumaytepeque might be grouped in a single subgroup as descendents of an intermediate common ancestor, itself a daughter of Proto-Xinkan, because they share an innovation. This is based on the changes * > [l] in Jumaytepeque and partially in Yupiltepeque, and *l' > [t'] in both these languages, assuming here that the corresponding Yupiltepeque <t> in these words as a / ‘/ However, this is uncertain; it can only be tentative at best, and the subgroup would need to be supported by other evidence. This matter is considered again in section 4.4. Yupiltepeque is not included in all of the correspondence sets in the following section because of missing information, but where there is information, it has been included. There are other internal considerations involving all of the Xinkan languages, generally, which affect the outcome of the reconstruction of Proto-Xinkan phonology specifically. By internal I mean those morphological and phonological processes of Xinkan that limit the use of consonants and vowels. While in general these do not cause serious problems for the reconstruction, they are worth mentioning here as they are commented on throughout this chapter. For example, there is a unique pattern of vowel harmony in Xinkan based on the height of the vowels in a word. That is, co-occurrence of vowels within a morpheme or word is restricted to those vowels which have similar height vowel phonetically (see section 3.1.2). Often the changes from Proto-Xinkan to one of its daughter languages appear to be constrained by this limitation. The vowel harmony process and its effects are discussed more in depth below. Additionally, speakers of the Xinkan languages often vary in their use of glottalization (see Chapter 4). Campbell and Muntzel (1989) show that in at least one 27 speaker of Jumaytepeque this is due to imperfect learning as a second language speaker. However, Dorian (1993) and Lass (1993) both agree that even though it may be the case that contracting languages are changed due to the imperfect learning of some speakers, there might also be a more general external or internal influence on the language which has more regular results than the claim from imperfect learning. For example a (second-language) speaker might tend to simplify the phonology of a language due to imperfect learning; this simplification is usually from marked elements to unmarked ones (or from more difficult ones to less difficult ones in some meaningful definition of difficulty). However, a common internal change in the phonology of any language is from more marked to less marked. Consequently, changes such as the simplification of linguistic elements cannot be considered to result from one or the other explanations in isolation since both are possible at any time in the history of a language. It is difficult in most cases of language change involving moribund or obsolescing languages to ascertain when a change toward simplification is caused by imperfect learning or natural internal changes or both (see Campbell and Muntzel 1989 for a discussion on multiple causation in changes in obsolescent languages). This issue is brought up throughout this work when a reconstruction requires it. A good example of s ‗ a sa ‘ is found in examples of sound correspondences where one of the Xinkan languages does not pattern with the others and the aberrant sound is not motivated by any apparent linguistic phenomenon. For example, often the reflexes for a given proto-sound, say *p', are glottalized in two languages but not in the third. The absence of glottalization is not predictable, i.e., it does not occur in general across the entire deviant language, which would indicate a 28 sound change. Rather it occurs randomly in individual lexical items which make up cognate sets. Critically, however, there is never a specific identifiable linguistic phonetic condition for this change. That is not to say that there are no conditions motivating the change, but that any conditions are idiosyncratic in that they reflect the style, preference, or competence of the individual speaker and not general linguistic conditions. For example, it could be that the speaker(s) did not glottalize a consonant because of difficulty in articulatory production. In such a case the motivation for change is a sociolinguistic question and might indicate a broader shift from *C' to /C/ due to the latter explanations rather than phonetic environments. In order to address the explanations for these differences of pronunciation adequately there seem to be two competing possible solutions. The first would be to assume, as is common when doing historical linguistics, that languages tend to change in the direction from more marked to less marked, in other words, to assume the change is caused by internal motivations of simplification. Following this, it would be appropriate to reconstruct *C', postulating a change to a plain consonant in the one language lacking the glottalized sound. The result is merely a language-internal change identified in a few lexical items; since it is not a general change in the language, it is assumed that it indicates the natural variation in language that can lead to subsequent more global changes. However, the other option is to acknowledge the external influence (i.e., social and cultural influences) on this language and the imperfect learning of the surviving speakers coupled with knowledge of similar problems in the reconstruction of other language families. This acknowledgement might show that it is, sometimes, the least 29 common segment in a correspondence set that is the most conservative and that all other languages being considered underwent the given change. For example, in Romance almost all the modern languages (e.g., Spanish, Italian, French, Portuguese, etc.) change *k to some sort of fricative or affricate, except Sardinian. Nevertheless, *k is reconstructed and not some fricative or affricate because of what is assumed to be plausible in the direction of sound change. In the Xinkan example where all but one language has a glottalized consonant in a given correspondence set, this would mean that it might be appropriate to reconstruct the nonglottalized consonant, *C, and posit changes in the other languages. A necessary note would be that the change has proceeded in the direction form less-marked to more-marked. Fortunately, issues such as the reconstruction of glottalized consonants are not overly problematic; in most cases the reconstruction of Proto-Xinkan sounds is transparent and straightforward. In the situations in which transparency is not the case, careful examination of the data leads to a clear solution. Note, however, that both of these foregoing reconstructions involving glottalized consonants might be an accurate representation of the historical development of the Xinkan languages. However, the current state of Guazacapán as spoken by the handful of second-language speakers severely reduces the number of occurrences of glottalized consonants (though not all of them), arguing for a general pattern in the direction of deglottalization (i.e.,, ‘ > a s a speakers overgeneralized the glottalization due to imperfect learning. In that case it might be considered a general change in the language - highly unlikely, though, since a a s a s > ‘ s a 30 environment of the change. Fortunately, the instances of overgeneralization of glottalization are isolated to one or two speakers and are sporadic occurrences, not regular, and therefore cannot be considered generally relevant for the individual language or language family. This means that unless there is strong evidence to the contrary, it is assumed that in a correspondence set, the most common sound across the related languages is a direct descendent from the proto-sound. Furthermore, the reconstruction of the glottalized consonants is made more difficult because of their role in Xinkan morphology. Specifically, verbs can be inflected for either perfective or imperfective aspects. In the imperfective, the rightmost consonant of the stem is glottalized, and in the perfective it is not. If the root has an underlying glottalized consonant, then it remains unchanged, glottalized in both aspects. There is a problem in that the extant documentation often records verbs in only one or the other of the two aspects, but not both. This means that if the morphology was the same in the past as it is now, it would be safe to assume that there was an alternation between glottalized and plain variants of the consonants in the aspectual changes in verb roots . In these cases, *C' is always reconstructed for the imperfective aspect. With these caveats, there are at least three ways that the reconstruction of Proto- Xinkan relates to general issues of historical linguistics: the direction of sound change (traditionally believed to be predominantly from marked to less marked), the effects of language contact, and implications of the viability of the language to appropriate reconstruction. These issues are discussed in this order, with the language specific issues being dealt with first, followed by the reconstruction of the Xinkan phonology, and lastly 31 a general discussion relating this reconstruction to relevant issues of historical linguistics in general. The data presented throughout this grammar come from many different sources. The most useful are the unpublished field notes of Lyle Campbell and Terrence Kaufman, now prepared in database format (Rogers, Kaufman, Campbell, and Palosaari 2008, unpublished). This database contains information on three of the Xinkan languages (Guazacapán, Chiquimulilla, and Jumaytepeque). This information coupled with the information gathered from my own fieldwork with the last speakers of two of these languages constitutes the data upon which the reconstruction presented below is based. As mentioned earlier, the fourth Xinkan language, Yupiltepeque, is extinct and very poorly attested, but a useful pre-scientific grammatical sketch and vocabulary are found in Calderón (1908) and reprinted with a few corrections in Lehmann (1920), which also reproduces the other scant materials on this language from other; all of the Yupiltepeque data come from these sources. As a preliminary step in presenting the grammatical information of Xinkan, and as an organizational and topical outline of the following chapters, the next chapter provides a brief overview of the typologically significant traits of Xinkan. CHAPTER 2 TYPOLOGICAL OVERVIEW The purpose of this chapter is to provide an overview of the typological characteristics of Xinkan languages. The discussion in this chapter serves as background for the more in-depth development of the linguistic patterns in later chapters. Another objective of this chapter is to indicate how structures and patterns in the Xinkan languages compare to other languages, both in the Mesoamerican linguistic area and in the world in general. The discussion of phonological features of Xinkan languages has been left largely for Chapter 3, but it can be said here that these languages are phonologically unique, or unusual, in that they have the following characteristics: 1. Glottalized obstruents and glottalized resonants (section 3.2). 2. Vowel harmony based on height and centrality of the vowel inventory (section 3.1.2). 3. Allomorphic alternation involving glottalized consonants (section 3.3.1). 4. a a a a a [ s‘] a a [ s] s 3.2). 5. Phonological alternations between voiceless fricatives [s] and [š] on the a a [ s‘] s 3 2 4 2 a 3 3 1 33 6. Predictable stress on the vowel after the right-most consonant of the root (section 3.5). Xinkan languages are moderately polysynthetic, with a large class of suffixes and a smaller class of prefixes. Furthermore, it is a characteristic of these languages that multiple suffixes can be attached to a single root, but that only a single prefix is ever allowed. The prefixes are always person agreement markers when on verbs in the imperfective aspect and personal possession markers when used in conjunction with nouns. Additionally, prefixes and suffixes may be used on the same root at the same time. That is, a root can have a prefix as well as one or more suffixes, though there is not requirement for it to have both. In some regards, Xinkan languages generally have some fusional characteristics as well as agglutinative one but it cannot be suggested that Xinkan coincides with the prototype of either extreme. More specifically, some morphemes, such as the verb agreement markers (see section 5.3.2.1), can mark several grammatical meanings and functions simultaneously - that is, they are portmanteau morphemes. For example, in the case of the verb agreement markers, they mark both person and number and the aspect of the verbal action. However, most unbound (free) morphemes are not portmanteau morphemes. The following morphological processes are observed in the Xinkan languages. The section of discussion of each process is listed at the very end of each item. 1. Prefixation - small set (see sections 5.1.1.1.2 and 5.2.2.1) 2. Suffixation - large set ( see sections 5.1.1.1.1, 5.1.1.2-4, 5.1.2.1, 5.2.2.2, 5.3.2.1, 5.3.3.1-2, and 5.4 for examples) 34 3. Stem modification - restricted to verbal aspectual inflections and voice derivations (see sections 5.3.2.2 and 5.3.2.4). There are nine lexical classes, or wor classes, exhibited in the grammar of the Xinkan languages. These classes are defined both morphologically and syntactically. Morphologically, each lexical class is defined by the set of bound suffixes they bear and/or by the morphological processes and operations that they can undergo. Syntactically each grammatical category is delimited by its place in the linear order within a phrase or a clause. The relavant lexical classes are nouns, including pronouns, (section 5.1.1), adjectives (section 5.1.2), relational nouns (section 5.4), quantifiers (section 5.1.3.3), determiners (sections 5.1.3.1 and 5.1.3.2), adverbs (Chapter 5 throughout), verbs (section 5.3), verbal particles (see section 5.5), and verbal auxiliaries (see section Chapter 5 throughout). Xinkan languages are nominative-accusative in that verbal agreement is the same for the agents of transitive verbs and the subjects of intransitive verbs; while there is no verbal agreement for objects and transitive verbs. There is no nominal case system in the Xinkan languages which means verbal concord is the only relevant way of determining nominal argument alignment. Examples of subject-verb agreement are seen in (1) where the verbal marker that is corefential with the grammatical subject is given in bold. (1) Syntactic alignment a. ima-n' Hwan nen' tell-1SG.PERF.TV Juan I ‗ a ‘ 35 b. ün-im'a Hwan nen' 1SG.IPERF.TV-tell Juan I ‗ a ‘ c. ün-tik'i-lha' nen' 1SG.PERF.IV-sleep-UNERG I ‗ s ‘ d. ün-apla-‘ nen' 1SG.PERF.IV-bathe-UNACC I ‗ a ‘ These examples show that regarding verbal agreement subjects of transitive verbs (1a and 1b) and subjects of intransitive verbs (1c and 1d) are treated the same, to the exclusion of the objects of transitive verbs. Specifically, these are the only nominal arguments of a verb that require a person affix agreeing in person and number to be used on the verbs. Consequently, Xinkan languages can be categorized as nominative-accusative. Moreover, Xinkan languages exhibit three verb classes which are distinct from the syntactic alignment just described. In this regard they are similar, but by no means identical, to languages such as Cupeño (Uto-Aztecan, Hill 1969: 350-5). More specifically, Xinkan verbs are divided into three classes based on semantic properties and these classes are indicated overtly in the morphology of the language. The first class of s a ‗ a ‘ ‗ - a ‘ sense that there is no overt 36 morphological marking on them. All the transitive verbs in the Xinkan lexicon belong to this class as exemplified in (1a) and (1b) above. The second and third class of verbs are exclusively intransitive. The second class is the unergative intransitive verbs which indicate that the subject has a measure of control or volition in the action of the verb. The third class is the unaccusative intransitive verbs which indicate that the subject does not have control or volition in the action of verb. These two classes of intransitive verbs are indicated overtly through suffixes on the verb root. The suffixes used for unergative verbs are -lha' [-ɬaʔ] (Guazacapán and Chiquimulilla), -la' [- aʔ] (Jumaytepeque), and <-l > in (Yupiltepeque). In contrast the suffix used for unaccusative verbs is -‘ [-ʔ] Guazacapán, Chiquimulilla, and Jumaytepeque; no data on this suffix are available in the Yupiltepeque data, but there are some indications, such as stress placement on cognate verbs that indicates this might have been the case in this languages as well (see section 5.3.1 for more discussion). Some examples of intransitive verb class morphology are given in (2) and (3) with the class suffixes indicated in bold. Note also that these class suffixes are only used in the perfective aspect of intransitive verbs (see the appendix for a set of full verb paradigms). (2) Xinkan unergative verbs a. iw'a-lha' ‗ as ‘ Guazacapán b. ipla-lha' ‗ a ‘ Chiquimulilla c. pümü-la' ‗ a ‘ Jumaytepeque d. <saprikil > ‗ s ‘ Yupiltepeque (Calderon 1908:19) 37 (3) Xinkan unaccusative verbs a. yolhna-ʔ ‗ s ‘ Guazacapán b. maaxi-ʔ ‗ ‘ Chiquimulilla c. hürami-ʔ ‗ ‘ Jumaytepeque As indicated in (1) above subject-verb concord is always indicated through prefixation for intransitive verbs (not shown in examples (2) and (3)). The verb co-referential agreement indicates which nominal argument is the grammatical subject of the verb (in the case of intransitive verbs, of course there is only one nominal argument). On the other hand, the intransitive verb class suffixes indicate whether the grammatical subject is the logical subject (agent) or the logical object (patient). Thus the unergative suffixes in (2) mean that the grammatical subject is the same as the logical subject (agent) of the action, the doer, or controller, of the action. The unaccusative suffix in (3) means that the grammatical subject is the same as the logical object (patient) of the action, the undergoer, or noncontroller, of the action. Every intransitive verb must have one of these two suffixes. Section 5.3.1 discusses intransitive verbs in more detail Furthermore, there is a secondary alignment pattern with one derived verb type: the antipassive. This is especially interesting because Xinkan languages exhibit nominative-accusative alignment and it is uncommon for an antipassive construction to be used with this language type, as it is most commonly found with ergative languages (though there are exceptions, and Xinkan is one of these). Subjects of antipassive verbs are treated like transitive objects, patients of the action (P), in that they do no require verb agreement markers. This is shown in the examples in (4) where the suffix ‘-k'i' is used 38 to derive the antipassive verb form. The contrast is indicated in bold, where the transitive verb form has a subject agreement suffix while the antipassive verb form does not. (4) Guazacapán antipassive derivation a. wüüxa-n' p'awawa' shake.out.PERF-1SG.PERF.TV rag ‗ s a ‘ b. wüüxa-k'i nen' shake.out-ANTIP I ‗ s a ‘ (omitted object) Other than the verb agreement markers just mentioned, linear order does provide some indication of the grammatical relations between verbs and nominal arguments. Grammatical objects of transitive verbs always immediately follow a verb and do not have any overt morphological marking. The position of grammatical subjects can vary but most commonly they follow the grammatical objects, VOS. In (5) the object is bold. (5) ima-y nen' taata-n' tell-3SG.PERF.TV I father-1SG.POSS ‗ a ‘ 39 Indirect objects and any other obliques can be placed either between the direct object and the subject or after the subject. (6) nuk'a-y' k'alh map'u taata-ka' Hwan give.PERF-3SG.PERF.TV a tortilla father-2SG.POSS Juan ‗ a a a a a‘ (7) nuk'a-y' kalh map'u Hwan taata-ka' give.PERF.-3SG.PERF.TV Juan a tortilla father-2SG.POSS ‗ a a a a a‘ Sections 5.3.1, 5.3.2.4, and 7.1 treat grammatical alignment more fully. 2.1 Basic sentence/clause word order The basic word order of a transitive clause in Xinkan is V(erb) O(bject) S(ubject). However, there is some varaiation. Specifically, nominal arguments of a predicate can be preposed to positions in front of the verb phrase. While this is common for subject nominals, it is quite rare, though possible, for grammatical objects. A preposed nominal argument is always modified by the definite article. This means that the available data on Xinka indicates that an indefinite noun phrase can not be preposed. Examples of sentences from Guazacapán are given below; see section 7.2 for further examples in this and the other Xinkan languages. 40 (8) opo-y' palh uuts'i ayaalha VOS break.PERF-3SG.PERF.TV now nixtamal mujer ‗The women broke the x a a ‘ (9) na ayaalha man ton'ohe-y' Hwan SVO the woman that trick.PERF-3SG.PERF.TV Juan ‗That woman tricked (lied to) Juan‘ (10) ün-mük'a-lha' huurak xa waya' VS 1SG.PERF.IV-work-UNERG man in corn.field ‗T a ‘ (11) Ø-apla-‘ na pwerto-h maku VS 3SG.PERF.IV-open-UNACC the door-3SG.POSS house ‗T s as ‘ (12) na nüma-k hü' teena' pulhpu hooro-y' SOV the eat-INSTR this much dirt have.PERF-3SG.PERF.TV ‗T s a a a ‘ , was dirty) A number of other observations about word order can be made about specific phrases (e.g., noun phrases) or clause types (e.g., relative clauses). In the next few sections these other word orders are surveyed. 41 2.2 Noun phrase word order This section briefly surveys the noun phrase in Xinkan languages, though section 5.1.4 discusses this type of phrase in more detail. When used with adjectives, the order is that the noun always follows the adjective. In other words, it is always ADJECTIVE-NOUN (AN). (13) üran huurak ADJECTIVE-NOUN big man ‗ , - , a ‘ Determiners exhibit two patterns dependent upon the type of determiner being used in the noun phrase. The articles always precede the noun these modify, thus exhibiting the pattern ART-NOUN. Demonstratives on the other hand always follow the noun they modify, exhibiting the pattern NOUN-DEM. Lastly, in a genitive phrase the possessed nominal always precedes the possessor; the pattern is therefore NOUN-GENITIVE. (14) na uw'i-h kaxkax NOUN-GENITIVE the meat-3SG.POSS gopher ‗T ‘s s / a ‘ (15) xurumu man NOUN-DEMONSTRATIVE young.man that ‗T a a ‘ 42 (16) na waya' ARTICLE-NOUN the corn.field ‗ ‘ Possession is further divided semantically into alienable and inalienable possession. Possessive pronominal prefixes are used for alienable possession, while possessive pronominal suffixes indicate inalienable possession (see section 5.2.2.1 and 5.1.1.1). Overall, then, a general pattern of MODIFIER-HEAD is exhibited for nouns being modified by either adjectives or by articles. However, a pattern of HEAD-MODIFIER is exhibited for nouns modified by either demonstrative determiners or in the genitive constructions. Note that the construction ART-N-DEM is the only one available in Xinkan languages. T s s s a a a s, s a K‘ a a a s, a as a demonstrative flanking. (17) na huurak man the man that ‗ a a ‘ (18) na waya' hü' the corn.field this ‗ s ‘ 43 2.3 Relational noun or preposition phrases Xinkan languages, like most other languages in the Mesoamerican linguistic area (see Campbell, Kaufman, and Smith-Stark 1986:545), have relational nouns. These s x ss - a a a s, [a ] s a n root and ss ss a a x s‖ (19) üül'ü-n' behind-1SG.POSS ‗ ‘ (20) neelha-h for-3SG.POSS ‗ ‘ (21) par'a-ka below-2SG.POSS ‗ ‘ (22) xa-h in-3SG.POSS ‗ ‘ 44 These nouns, when not possessed, can be used as prepositions in the Xinkan languages (see section 5.4). The order of the constituents in these preposition phrases is PREP-NOUN. (23) xa maku in house ‗ s ‘ (24) hina' ay'aalha with woman ‗ a ‘ 2.4 Relative clause word order There are a few options in forming relative clauses, but these options all have to do with the choice of relativizer rather than word order (see section 7.5.3). In all relative clause strategies the order is HEAD-NOUN RELATIVIZER RELATIVE-CLAUSE (N-REL). (25) na nen' hooro-n' machiiti ke küwa-ha-y' nen' the I have.PERF-1SG.PERF.TV machete that borrow.CAUS.PERF-3SG.PERF.TV I ‗ a a a ‘ 45 (26) talhma hü' kuy tur'a-n' road this FUT take.PERF-1SG.PERF.TV ‗T s s a a a ‘6 (27) hin hünü-n' huurak na ka-taayi-' hina' no know.PERF-1SG.PERF.TV man that 2SG.PERF.IV-came-UNACC with ‗ ‘ a a a ‘ This chapter is meant as a preliminary overview of the information presented in the rest of this grammar. Cross-referenced sections have been indicated to point the reader to a more full discussion of the relevant aspects of Xinkan grammar. 6 This example might not be indicate a relative clause but merely be a focused or a s , ‗ s a ‘, s s , a s constituents must be modified by the definite article, and since this example does not exhibit it, it is assumed that it is a relative clause rather than a variation in basic word order. See section 7.5.3 for more discussion. CHAPTER 3 PHONOLOGY The purpose of this chapter is to describe the phonology of the Xinkan languages. In particular, the articulatory properties of the vowels and consonants are surveyed together with their distributional patterns. Additionally syllable structure, stress assignment, and phonological processes, such as (de)glottalization, are described. This chapter concludes with a description of the practical orthography used for Xinkan languages. Because the discussion of sounds is based on abstract sound patterns, all the examples in this chapter are provided in the International Phonetic Alphabet (slightly modified by using Americanist phonetic symbols in order to accommodate Xinkan). Examples in subsequent chapters will be presented exclusively in the orthography presented at the end of this chapter, with representation in IPA given only when beneficial for clarity. It is one of the main goals of this grammar that the information is presented in a way that is accessible to the Xinkan community, as well as the scholarly linguistic community. This is the motivating factor for utilizing the practical orthography and limiting the number of examples written in the IPA. This is standard practice in language documentation wherever the practical orthography provides an adequate means 47 of representing the phonemes of the languages. Phonemic representations are given in a a a s ‗/…/‘ a s a s a s a a s ‗[…]‘ 3.1 Vowels T a s x s a X a a a s: / , ɨ, , , , a/ a a these vowels have long and short contrasts in the phonologies of these languages. The long and short contrasts are phonemic in that they effect a change in meaning in words; however, vowel length is also a part of the morphological processes and is discussed in section 5.3.2.2. Vowels in Xinkan languages have unique distributional properties which restrict vowel co-occurrence within a word. Specifically, these languages exhibit patterns that can be classed as vowel harmony, where there are restrictions on what groups of vowels are allowed to occur with one another in any given word. In Xinkan, these restrictions are based primarily ‗ ‘ s as as a s within the vowel space. Table 1 provides the vowel inventory of all four Xinkan languages in IPA representation. 7 The description of Xinkan vowels in Table 1 follows closely the framework for phonetic description in Ladefoged (1997) and Ladefoged and Maddieson (1996). The descriptive terms in the first column and the first row indicate the general phonetic articulatory features commonly used to describe vowels. Specifically, this table shows 7 All back vowels are also redundantly round. This conforms to generalized patterns of vowels cross-linguistically: non-low back vowels tend to be round (see Ladefoged and Maddieson 1996:290, Crothers 1978:97, Fant 1973:186, and Ladefoged 2006:181). 48 Table 1. Xinkan vowel inventory Front Central Back High Short i ɨ u Long ii ɨɨ uu Mid Short e o Long ee oo Low Short a Long aa that Xinkan languages have a vowel inventory of six vowels which have both long and short equivalents (long vowels written double, for example ii for [i:]). As mentioned above, and discussed in section 3.1.1, vowel length is phonemically contrastive, though there are also morphophonological processes which change underlyingly short vowels into long vowels in certain contexts. Table 1 also shows that there is no vowel in the central mid space, indicating that there is no contrastive /ə/, a fact which is expected and common cross-linguistically with vowel inventories of six vowels (see Maddieson 1984 and 1997; Crothers 1978). While it is not clear cross-linguistically why this organization should hold it has been suggested that it is due to the optimization of the vowel space for needed linguistic contrasts (see Crothers 1978 and de Boer 2001). Vowels tend to be maximally distant from one another a , a - ax a ‖ The mid central vowels, in other languages, are not as distinct articulatorily from several of the other vowels in their inventories, not as linguistically distant, as the high central vowel might is in Xinkan. If this suggestion is true it can be tentatively suggested that in Xinkan this allows the high-central vowel /ɨ/ to have more perceptual vowel space which in turn might appear to affect the distribution of vowels (see the discussion of vowel harmony in section 3.1.2). 49 3.1.1 Vowel length As mentioned in the last section, vowel length is phonemically contrastive in the Xinkan languages for all vowels (as seen in Table 1). Long vowels can be lexically specified (underlying) or they can be the result of phonological processes (e.g., vowels are lengthened in unaccusative verb forms, when the agent noun suffix is added to the verb stem, and in the plural formation of noun roots ending with a vowel. In Jumaytepeque, vowel length is also the result of a phonological process that affects vowels in the verbal noun derived from a causative verb.) Underlying long vowels are exhibited in any position within the word except word-finally; while derived vowel length affects the second to last syllable in the root in all situations except the Jumaytepeque verbal noun derivation which affects the final vowel in the root. In this section each of these vowel-length alternations and patterns is surveyed. The phonotactic constraints on vowel length are treated in detail in section 3.4, while the verb morphology is surveyed in section 5.3. As discussed in the introductory chapter, little linguistic information on Yupiltepeque is available and so vowel length in Yupiltepeque is not indicated below. 3.1.1.1 Lexical specification Lexically specified (underlying) vowel length means that the vowel length in a word is not a result of a phonological process, but is rather a phonemically contrastive segment in the underlying word root. Underlying long vowels specified in the lexicon are found in both native Xinkan words and many words borrowed from Spanish. Importantly, however, one of the ways in which the Xinkan languages differ most noticeably from one another is in their vocabularies, and consequently not all of the words in each of the 50 languages have the same vowel length specifications underlyingly. Some examples of words with lexically specified vowel length are given in (28) through (30); the Spanish etymological source is given to the right of a word if it is a loan word. See the comparative Xinkan dictionary (to appear) for other words with inherently long vowels. Some of the words in these examples are similar in form to unaccusative verb constructions (see section 5.3.1) though there is no evidence showing that this is their correct morphosyntactic analysis. It may be that these forms are frozen and no longer have the unaccusative semantic meanings. (28) Examples of Guazacapán words with lexically specified long vowels aaʔu ‗ ‘ aara ‗ ‘ ay'aaɬa ‗ a ‘ č'iipi ‗ as ‘ eeɬe ‗ a ‘ woona ‗ ‘ haama ki' ‗ ‘ haamaʔ ‗ ‘ haaniʔ ‗ ‘, ‗as‘ haar'un ‗ ‘ huurak ‗ a ‘ huuri ‗ s‘ huuši ‗ a ‘ huutak ‗a s‘ huuts'uk ‗ ‘, ‗ ‘ huutuk ‗s ‘ iihuukah ‗ ‘ iimaakah ‗ a ‘ iimookah ‗ s ‘ iipan ‗s a ‘, ‗ ‘ iipemaakuh ‗ s‘ iiti ‗ a ‘ išaapi ‗ ‘ ɨɨɬɨk ‗ , a ‘ ɨɨn'a ‗ a ‘ k'iir'a ‗s a , s ‘ 51 k'iira ‗ ‘ k'iišu ‗ x a ‘ k'oočoʔ ‗ s‘ k'oomo ‗ ‘ k'oosek ‗ , s‘ k'ooso ‗s a x‘ k'oošo ‗ s‘ k'ooto ‗ a s‘ k'oots'ay ‗ a ‘ k'oroor'o ‗reed‘ k'uunuʔ ‗ ‘ k'uuyu ‗ as a a ‘ k'weets'a ‗ a ‘ kaašik ‗ ‘ kaayi ‗s ‘ kiiw'i ‗ a ‘ kɨɨša ‗ a , a ‘ kɨɨw'ɨ ‗s ‘ koolah ‗ a a a ‘ < S a kookoʔ ‗ a a ‘ kooraʔ ‗ ‘ kuukuʔ ‗ , ‘ maaɬek ‗ ‘ maaɬi ‗as ‘ maama ‗ a ‘ meeme ‗ a ‘ miiči ‗ a ‘ miiku ‗s a ‘ mɨɨmɨ ‗s ‘ muur'a ‗ a ‘ muuti- ‗ a ‘ naaɬik ‗ ‘ naana ‗a a ‘ naatɨɨkah ‗ a ‘ naay'ah ‗ s ‘ natɨɨkah ‗ a a ‘ natɨɨy'ah ‗ a s ‘ neeɬa ‗ ‘ neeɬek ‗ s‘ nooya ‗ a a‘ nuunuʔ ‗ ‘ nuuru ‗ s‘ oor'o ‗ ‘ <S s oošo ‗ a , s s‘ seema ‗ s ‘ šaaru ‗s a‘ 52 taata- ‗ a ‘ ts'iin'an'a ‗s ‘ tɨɨm'al ‗ s ‘ weeša ‗ a a‘ wiira ‗ , ‘ (< Sp. huira) yuu ‗ VOC‘ (29) Examples of Chiquimulilla words with lexically specified long vowels aaʔu ‗ ‘ aabuh ‗ !‘ aaluʔ ‗ a a a a‘ aara ‗ ‘ ačiimi ‗ s ss a ‘ akuuša ‗ ‘ < Sp. aguja) boohoo ‗ ‘ č'iipi ‗ a a ‘ č'ɨɨr'ɨk ‗ a ‘ čiiriʔ ‗s ‘ duusi ‗s ‘ <S goona ‗ , a ‘ goošaɬ ‗ ‘ haalak ‗ax ‘ haar'u ‗ ‘ haari ‗ s ‘ haaru ‗s a , s ‘ hɨɨm'a ‗ a ‘ hoor'o ‗ a a ‘ huuri ‗ s‘ huuša ‗ ‘ huušiʔ ‗ a ‘ iipan ‗s a ‘ iiru ‗ ‘ iišuɬ ‗s a a‘ ɨɨkɨ ‗s a ‘ ɨɨl'ɨ ‗ a ‘ ɨɨna ‗ a ‘ k'aamiʔ ‗ ‘ k'iir'a ‗ , ‘ k'iira ‗ ‘ k'iiša ‗ as ‘ k'iišu ‗ a ‘ k'oomo ‗a ‘ k'oošo ‗ a ‘ kooto- ‗ a ‘ 53 k'ooye ‗ s ‘ k'uuyu ‗ a ‘ k'weets'a ‗ a ‘ k'iiw'i ‗ a ‘ kɨɨšɨ ‗ a ‘ kookaɬ ‗s ‘ kuukuʔ ‗ , ‘ ɬuuri ‗ a ‘ maama ‗ a ‘ mačiiti ‗ a ‘ < S a maɬiiɬa ‗ a a ‘ meeme ‗ a ‘ p'uupu ‗ ‘ yaaɬaʔ ‗ , a ‘ (30) Examples of Jumaytepeque words with lexically specified long vowels aaru ‗ ‘ aara ‗ ‘ aayuʔ ‗ a ‘ č'iipi ‗ as ‘ čiibu ‗ a ‘ eela ‗ ‘ haalak ‗ a ‘ haan'ah ‗ ‘ haar'u ‗ ‘ haari ‗ , ‘ huhuuya ‗ a' huuma ‗ a ‘ huuši ‗ a ‘ huuts'i ‗ x a a ‘ ʔiiru ‗ ‘ iišul ‗chigoe‘ iiwa ‗ as ‘ ɨɨkɨ ‗s a ‘ ɨɨna ‗ a ‘ ɨɨt'ɨ ‗ ‘ k'eetan ‗ a ‘ k'iišu ‗ a ‘ k'oočo ‗ ‘ k'oomo ‗ ‘ k'uutu ‗s a a ‘ kaayi ‗s ‘ kiira ‗ ‘ kiiw'i ‗ s ‘ 54 kɨɨwɨ ‗ a ‘ kuuku ‗s a ‘ kuum'i ‗ as ‘ laam'a ‗ a a ‘ laayuʔ ‗ s ‘ leelan ‗ ‘ luuuri ‗ a ‘ maalɨ ‗ ‘ maašin ‗ a ‘ maayi ‗ a ‘ meeʔ ‗ ‘ meen'e ‗ ‘ miiša ‗ a ‘ paaha ‗ , a ‘ paaši ‗s a ‘ tɨɨm'al ‗ a ‘ wiik'i ‗ ‘ woono ‗ ‘ Note the Spanish loanwords with underlying vowel length show that long vowels occur on the vowel which is natively stressed in Spanish. This is a common adaptation strategy in the languages of the Americas. 3.1.1.2 Vowel length alternation Vowel length can also be the result of phonological processes determined by morphological context. That is, vowels can be lengthened in the environment of certain morphological affixes (see Chapter 5 for a discussion of Xinkan morphology). In this situation, either the first vowel or the last vowel in the root may undergo lengthening, depending on linguistic context and on the affix involved. In this section the kinds of alternations in morphological contexts are discussed and exemplified: first, the lengthening of root vowels in the penultimate syllable connected with a change in the transitivity of a verb; next, the lengthening of vowels in the final syllable of the root in conjunction with the agent noun of a derived causative verb in Jumaytepeque; third, 55 vowel lengthening due to the presence of the agent noun suffix, in this situation the first vowel in the root is lengthened in Guazacapán and the last vowel in the root in both Chiquimulilla and Jumaytepeque; and lastly, the lengthening of the last vowel in noun roots when the plural suffix is attached. One phonological change resulting in vowel length is found in the derivation of an unaccusative verb form from its corresponding transitive verb root (see section 5.3.1). However, vowel length is only exhibited in the unaccusative forms of verbs which contain two or more syllables and which have no word-medial consonant clusters, that is, have only a single medial consonant. More specifically, the underlying canonical syllabic shape of a word which is affected by this process is CV1CV (Consonant-Vowel-Consonant-Vowel) and its surface realization is CV1V1 Vʔ W a s a s s a , s most common shape of Xinkan words, making vowel lengthening in this process extremely common. The unaccusative formation of words with other underlying phonological shapes, i.e., CVVCV and VCCV, is discussed in Chapter 5 (see section 5.3.1). In order to form an unaccusative intransitive verb from a transitive verb root the suffix -ʔ ‗UNACCUSATIVE‘ is attached to the end of the stem. When the transitive verb stem is of the phonological shape CVCV the first vowel is lengthened in addition to the use of the unaccusative suffix. That is, the stem is modified from underlying CV1CV to 3.1.1.2.1 Vowel length and verb transitivity. 56 CV1V1CV-ʔ8. The lengthened vowel in the unaccusative form is represented in the orthography and in the analysis below as two adjacent identical vowels. In the examples in (31) through (33) the basic (i.e., underlying) transitive verb is on the left and the derived unaccusative (i.e., surface) form is on the right. (31) Unaccusative formation of transitive verbs: Guazacapán a. ʔima-n  ʔiima-ʔ say.PERF-1SG say.UNACC-UNACC ‗ sa ‘ ‗he was told‘ b. ts'm -ka  ts'm - water.PERF-2SG water.UNACC-UNACC ‗ a ‘ ‗ was a ‘ (32) Unaccusative formation of transitive verbs: Chiquimulilla a. huša  huuša-ʔ blow.PERF blow.UNACC-UNACC ‗ ‘ ‗ as ‘ 8 Note that for other phonological shapes this vowel lengthening process is not required. For example CVCCV is realized as CVCCV-ʔ a V V V s as V V V-ʔ 57 b. ts'aya  ts'aaya-ʔ scale.PERF (a fish) scale.UNACC-UNACC ‗ s a ‘ ‗ as s a ‘ (33) Unaccusative formation of transitive verbs: Jumaytepeque a. kišu  kiišu-ʔ change.PERF change.UNACC-UNACC ‗ a ‘ ‗ a ‘ b. hama  haama-ʔ ripen.PERF ripe.UNACC-UNACC ‗ ‘ ‗ ‘ In Jumaytepeque it is possible to form a verbal noun from a derived causative verb. When this verbal noun is formed, the last vowel of the (underlying) verb root before the causative suffix is lengthened. Importantly, this is true only for Jumaytepeque; the other languages do not have the same vowel lengthening process, although they do have verbal nouns created with cognate affixes. Examples are given in (34). 3.1.1.2.2 Vowel length in Jumaytepeque verbal nouns. 58 (34) Jumaytepeque verbal nouns derived from causative verbs a. ɨy'awa ɨya-ha ɨyaa-ha laugh.at.it.PERF laugh.at.it.PERF-CAUS laugh.at.it.VN-CAUS ‗ a a ‘ ‗ a someone a a ‘ ‗ a someone a a ‘ b. muč'u muč'u-ha muč'uu-ha encoger.PERF encoger.PERF-CAUS encoger.VN-CAUS ‗ s ‘ ‗ a s ‘ ‗ a s ‘ . In addition to changes in vowel length due to verbal inflection (see sections 3.1.1.2.1 and 3.1.1.2.3), vowels are lengthened also when an agent noun suffix is attached to a transitive verb. This suffix causes the last vowel in a verb stem to lengthen in Guazacapán but the first vowel in the verb stem to lengthen in both Chiquimulilla and Jumaytepeque. The agent noun suffix is -ɬa (Guazacapán), -ɬ (Chiquimulilla), or -l (Jumaytepeque). This suffix lengthens the indicated vowel in a verb stem except for in two contexts: (1) when there is already a lexically specified long vowel in specified position in the stem, where the process could be said to apply vacuously, or (2) when there is an intervening suffix between the agent noun suffix and the verb stem. As with the other phonological processes involving vowel length, the lengthening of the vowel in conjunction with the agent nouns suffix is also restricted to verb roots without word medial consonant clusters. This suffix always s a a ‗ [ ] Xs‘, where X is any action of a verb; note that the referent of an agent noun must be animate and volitional. Examples 3.1.1.2.3 Vowel length with de-verbalizing suffixes 59 of the vowel length alternations corresponding to the use of the agent noun suffix are given in (35), (36), and (37). The agent noun suffix is discussed in more detail in section 5.3.2.2.2. (35) Guazacapán vowel lengthening with the agent noun suffix a. mɨka mɨkaa-ɬa work.PERF work-AGT ‗ ‘ ‗ ‘ b. kits'i kits'ii-ɬa roast.PERF roast-AGT ‗ as ‘ ‗ as ‘ (36) Chiquimulilla vowel lengthening with the agent noun suffix a. k'ɨtɨ k'ɨɨtɨ-ɬ measure.PERF measure-AGT ‗ as ‘ ‗ as ‘ i.e., scales or a ruler) b. kawi kaawi-ɬ cry.PERF cry-AGT ‗ ,s ‘ ‗ , s‘ 60 (37) Jumaytepeque vowel lengthening with the agent noun suffix a. yawi yaawi-l make.firewoord.PERF make.firewood-AGT ‗ a ‘ ‗ ‘ b. tutu tuutu-l suck.PERF suck-AGT ‗s ‘ ‗s ‘ . The last case where vowels can be lengthened is in the plural formation of nouns ending in a vowel. In all of the Xinkan languages the last vowel in the nominal root is lengthened when the plural suffix is added. This suffix is -ɬi (Guazacapán and Chiquimulilla) or -li (Jumaytepeque) and plural inflection is discussed in more detail in section 5.1.1.2. (38) Guazacapán plural noun formation miya  miyaa-ɬi hen hen-PL ‗ ‘ ‗ s‘ 3.1.1.2.4 Vowel length in plural noun formation 61 (39) Chiquimulilla plural noun formation ʔiiru  ʔiiruu-ɬi monkey monkey-PL ‗ ‘ ‗ s‘ (40) Jumaytepeque plural noun formation šɨma  šɨmaa-li rat rat-PL ‗ a ‘ ‗ a s‘ 3.1.2 Vowel Harmony The goal of this section is to survey the distribution of vowels within Xinkan words and to show that a process commonly referred to as vowel harmony is exhibited. Vowel harmony is a phonological process which can be exhibited, cross-linguistically, in at least one of two ways: (a) the long-distant assimilation of vowel features, or (b) underlying vowel co-occurrence restrictions (i.e., not involving assimilation). In Xinkan vowel harmony is of the latter type. Specifically, vowels in Xinkan follow strict distributional restrictions according to the articulatory features [HIGH] and [CENTRAL]. That is, vowels are distributed within a word based on the basis of the relative height values of the vowels as indicated in Table 1: high, mid, or low vowels; and on whether they are peripheral to the vowel space or not. Xinkan vowel harmony is partly based on the relative height of the vowels in words and morphemes. In Table 1, three vowel heights are exhibited: high /i/, /u/, /ɨ/, mid /e/ and /o/, and low /a/. The length distinction indicated in the second column in 62 Table 1 does not affect the restriction of vowel distribution and consequently long and short vowels, for the purposes of discussing vowel harmony, may be conflated into six descriptive symbols: high front vowel i, mid front vowel e, high back vowel u, mid back vowel o, high central vowel ɨ, and low central vowel a. In the distribution of vowels, within the same word, all vowels must belong to one of three sets whose vowels can co-occur with each other, but vowels from one set cannot co-occur with vowels of another set in the same word, except neutral /a/, which can occur with any of the three sets. The sets are /i, u, a/, /e, o, a/ and /ɨ, a/. That is, the two central vowels ɨ and a pattern differently from the height distributions restrictions described here. Examples illustrating the distribution of the non-central high vowels /i/ and /u/ is given in the lists in (41), (42), and (43). (41) Distribution of /i/ and /u/ in Guazacapán hiiru ‗ ‘ tu tu ‗s a / ‘ ts'il'i ‗ a s ‘ čiiriʔ ‗s ‘ ts'uu i ‗ a ‘ miya ‗ ‘ tum'ay' ‗ a ‘ ts'am'u ‗ s s‘ pari ‗ a ‘ (42) Distribution of /i/ and /u/ in Chiquimulilla k'isku ‗ ‘ k'usu ‗a a ‘ hiri ‗s a ‘ kiiwiʔ ‗ a ‘ huuri ‗ s‘ piya ‗ a ‘ hun'a ‗ ‘ karumu ‗ ‘ aši ‗ ‘ 63 (43) Distribution of /i/ and /u/ in Jumaytepeque k'iišu ‗ a ‘ huhul ‗ ‘ hiši ‗s ‘ siipi ‗ s k'uusi ‘ , ‘ hiwa ‗ a a , a a ?‘ hum'a ‗ ‘ amu ‗s ‘ k'an'i ‗ ‘ Three facts can be observed from these examples. First, high vowels (/i, u/) occur with other high vowels, while / / is an exception. That is, /i/ and /u/ form a set of high vowels which can co-occur freely with each other but / / is excluded from that set. Second, the set of high vowels {i, u} can co-occur with the low vowel /a/. Third, the linear order within words of the noncentral vowels and the low vowel is not significant to the distribution of vowel; they can appear before or after the vowels of set {i, u}. Next, we turn to the two mid vowels /e/ and /o/, which are exemplified in (44), (45), and (46). (44) Distribution of /e/ and /o/ in Guazacapán šeek'e ‗ s ‘ ter'o ‗ a / ‘ ts'oko ‗ a ‘ k'oosek ‗ a ‘ seema ‗ s ‘ goona ‗ ‘ (45) Distribution of /e/ and /o/ in Chiquimulilla meeme ‗ a ‘ hero ‘s a ‘ hok'o ‗ a ?‘ k'ooye ‗ s ‘ weetan ‗ a s ‘ nooya ‗ a ‘ (46) Distribution of /e/ and /o/ in Jumaytepeque meen'e ‗ ‘ p'en'o ‘ ‘ hon'o ‗ a a ‘ 64 wooče ‗s a ‘ p'eesa ‗ ‘ k'ooyaaya ‗ a s a ‘ Mid vowels occur with less frequency in Xinkan than other vowels. The words in the examples above show that the mid vowels /e/ and /o/ have similar distributional patterns to those described for the noncentral high vowels. However, as is noted below, the linear order of the mid vowels is restricted. Observable in the data are two generalizations; first, the mid vowels /e/ and /o/ form a set where both members of the set can co-occur freely with each other, also freely in terms of linear order. Second, the vowels of this mid vowel set {e, o} can co-occur with the low vowel /a/. This means that /a/ must be allowed to pattern with all of the vowels discussed so far. The difference between the mid vowel set {e, o} and the other set {i, u}, lies in the restrictions on the linear order in which vowels of these sets can occur with the low vowel /a/. Specifically there are no examples in (44), (45), or (46), or anywhere else, where /a/ precedes a member of the set {e, o}, in native words. It should be noted, however, that the mid vowels can follow the low vowel in loan words, as in (47). (47) /e/, /o/ , and /a/ in loan words adoobe ‗a ‘ < S . adobe) paale' ‗ s ‘ < S a The examples in (41) - (47) show that vowels /i, u, e, o/ in Xinkan belong to either the set {i,u} or the set {e,o} with respect to the distributional patterns. The central vowels also play a role in vowel distribution, but in a completely different way. The high 65 central vowel, / /, has specific limitations on its distribution that differ from those seen in the examples above. Consider the list of words containing ɨ in (48), (49), and (50). (48) Examples of the distribution of /ɨ/ in Guazacapán p k' ‗ ‘ ts' r r ' ‗ ‘ ts' m'a ‗ a‘ waw'ya ‗ ‘ a ahɨ ‗ s‘ hɨl'a ‗ ‘ hɨnɨ ‗ a ‘ hɨyɨ ‗ ‘ t'ɨnɨ ‗ a ‘ (49) Examples of the distribution of /ɨ/ in Chiquimulilla č'ɨngɨʔ ‗ a ‘ pɨɨšɨ ‗ a a ‘ k'ɨy'a ‗ a ‘ šangɨ ‗s a ‘ ɨpɨ ‗ s , ‘ ɨɨna ‗ a ‘ ɨm'ɨɬɨ ‗ , a ‘ ɨndɨ ‗s ‘ k'ɨn'ɑ ‗ s ‘ (50) Examples of the distribution of /ɨ/ in Jumaytepeque ahmɨ ‗ ‘ sɨɨm'a ‗ ‘ šɨhɨ ‗ a ‘ hɨn'ɨ ‗ ‘ ɨɨkɨ ‗s a ‘ ɨrsɨ ‗ ‘ lɨsk'ɨw'a ‗ a s ‘ p'ɨn'a ‗ a as ‘ pɨša ‗s ‘ These examples indicate that the high central vowel ɨ can co-occur in roots with a, as exemplified with the other vowels in the inventory. However, what is unique about the distribution of ɨ is that in all the words in (48)-(50), the only possible co-occuring vowels are ɨ and a. This distinction between the high central vowel and all other non-low vowels cannot be based on a restriction to similar height specification characteristics, as 66 in (41)-(47) above, because the high central vowel does not pattern with other vowels with identical height specifications. It is consequently argued that a phonological feature [CENTRAL] is important to the phonological system of Xinkan. Specifically, [-CENTRAL] vowels i, u, e, o are restricted according to height, while the [+CENTRAL] vowel ɨ is restricted according to centrality. It might also be restricted according to height (high central vowels only occur with high central vowels, except for a), but with no mid central vowel in the inventory there is no evidence to suggest that this is a necessary surface description. The phonological characteristics of the low vowel /a/, as exemplified in the examples above, provide evidence for the conclusion that this vowel is neutral. In vowel harmony systems a neutral vowel is often not restricted in the same ways as the other vowels in the inventory. Instead, it is not subject to the phonological limitations relevant to vowel harmony and thus co-occurs with any vowel in the inventory (see Kramer 2003). In Xinkan the low central vowel /a/ is neutral since it does not participate in vowel harmony restrictions on either height or centrality; it is allowed to co-occur with any other vowel in the inventory regardless of the height or centrality specifications of the other vowels involved. Xinkan vowels are specified underlyingly in stems and vowel harmony is a consequence of the underlying vowel co-occurrence restrictions (not the result of assimilation); however, there is a dependent phonological process which alters the articulation of some vowels across morpheme boundaries (i.e., assimilatory). Consider the examples in (51), (52), and (53). 67 (51) Vowel harmony across morphemes boundaries in Guazacapán a. hiiruu- i ts'okoo- e p k'ɨ - i monkey-PL grackle-PL liver-PL ‗ s‘ ‗ a s‘ ‗ s‘ b. ts'iriri-ki oško-k'e pɨɬɨɬɨ-k'i colored-INC rotted-INC smooth-INC ‗ ‘ ‗ ‘ ‗ s ‘ (52) Vowel harmony across morphemes boundaries in Chiquimulilla a. nuun'uu-ɬi onee-ɬe ɨyɨɨ-ɬi mute-PL baby-PL gopher-PL ‗ s‘ ‗ a s‘ ‗ s‘ b. wili-k'i moro-k'e ɨsk'ɨ-k'i naked-PL wet-INC untied-INC ‗ a ‘ ‗ ‘ ‗ ‘ (53) Vowel harmony across morphemes boundaries in Jumaytepeque a. mul'ii-li k'otoroo-le yɨk'ɨɨ-li squirrel-PL parrot-PL dog-PL ‗s s‘ ‗ a s‘ ‗ s‘ 68 b. šuwi-k'i holo-k'e k'ɨɨtɨ-k'i swept-INC pretty-INC cold-INC ‗ s ‘ ‗ ‘ ‗ ‘ Vowel height assimilation across morpheme boundaries as in (51)-(53) is only exhibited when the suffix that is attached contains the high front vowel /i/. This means that, while completely productive, the observations in these examples are limited to three suffixes, -ɬi (Guazacapán and Chiquimulilla), -li a ‗PLURAL‘, -k'i ‗INCHOATIVE/ANTIPASSIVE‘ (all languages), and -yi' ‗3SG.TV‘ (Jumaytepeque). The plural formation is described in section 5.1.1.2, the use of the inchoative/antipassive suffix is detailed in sections 5.1.3 and 5.3.1.1.1, and Jumaytepeque verbal conjugation is detailed in section 5.3. As can be seen in the examples in (51)-(53), the suffix vowel /i/ is realized as [e] when preceded by mid vowels and it is realized as [i] when occurring after any other vowel. Since the high vowel /i/ is the underlying form of these suffixes, it might be argued that the only assimilation occurring changes this high vowel to a mid vowel and that there is no change when the preceding vowel before the suffix is high. This is a common occurrence in languages exhibiting vowel harmony (especially non-assimilatory vowel harmony, as in Xinkan). In Xinkan the vowels of roots are distributed according to co-occurrence restrictions and there is a phonological process triggered only by mid vowels, which affects the suffix vowel, /i/ [e]/ {e,o}+C__. As such, the latter phonological process might be considered a phonological process (constraint) outside, or on the periphery of the vowel harmony restrictions. 69 The possible reasons for the unique behavior of the high central vowel ɨ invite commentary. In (51)-(53) above it was shown that within a word ɨ can only occur with other instances of the same vowel or the low vowel a. The question that might be asked is why the high central vowel behaves individually and is not based on the height restrictions as exemplified with the other vowels. The reason for this unique behavior can be found in the characteristics of high central vowels in vowel inventories of the same size as in Xinkan. It is a common fact that central vowels tend to have a larger vowel space cross-linguistically than other vowels in a given inventory. That is, that the precise perception and articulation of these high central vowels in a six-vowel inventory