UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Binding-theoretic analysis of Navajo possessor YI- Horseherder, Nicole 1998

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

Item Metadata


831-ubc_1998-0473.pdf [ 2.16MB ]
JSON: 831-1.0088584.json
JSON-LD: 831-1.0088584-ld.json
RDF/XML (Pretty): 831-1.0088584-rdf.xml
RDF/JSON: 831-1.0088584-rdf.json
Turtle: 831-1.0088584-turtle.txt
N-Triples: 831-1.0088584-rdf-ntriples.txt
Original Record: 831-1.0088584-source.json
Full Text

Full Text

A BINDING-THEORETIC ANALYSIS OF NAVAJO POSSESSOR YI-by NICOLE HORSEHERDER B.A., University of Arizona, 1993 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Linguistics) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October 1998 © Nicole Horseherder,1998 In p resen t ing this thesis in partial fu l f i lment of t h e requ i rements fo r an advanced d e g r e e at t h e Univers i ty of Brit ish C o l u m b i a , I agree that t h e Library shall make it f reely available f o r re ference and study. I fu r ther agree that permiss ion for extensive c o p y i n g o f th is thesis f o r scholar ly pu rposes may b e g ran ted b y t h e head of m y d e p a r t m e n t or by his o r her representat ives. It is u n d e r s t o o d that c o p y i n g or pub l i ca t i on of this thesis fo r f inancial gain shall n o t be a l lowed w i t h o u t my w r i t t e n pe rmiss ion . D e p a r t m e n t o f ^ j UJlfrfic^ T h e Univers i ty o f Brit ish C o l u m b i a Vancouver , Canada Date < W DE-6 (2/88) Abstract This thesis examines possessor yi- in Navajo (Southern Athabaskan). Previous analyses deal with,y/- mainly as a prefix attached to verbs and post-positions; in contrast to prior work, this thesis analyzes yi- as a possessor prefix (attached to nouns). I propose that possessor^/- is a disjoint anaphor (DA), as originally proposed by Saxon (1984a, 1986,1995) for its cognate in Dogrib (Northern Athabaskan). As a disjoint anaphor it must have a local A-antecedent from which it is disjoint in reference. I show that,y/- must also have an A'-antecedent with which it is obligatorily coreferent. I interpret the binding behavior of yi- in terms of (Aouns1 1985) theory of Generalized Binding. I claim that since it must simultaneously satisfy condition ( as an A-anaphor) and C (since it must be A-free), yi- must crucially have two antecedents: an A - antecedent with which it is coreferent, and an A-antecedent from which it is disjoint in reference. I show that for this relation to be licit, both antecedents (A' and A) must also agree in phi-features with yi-, which is inherently specified as third person singular. ii Table of Contents Abstract ii Table of Contents iii Acknowledgements iv Introduction 1 0.0. Introduction 1 0.1. Goals and outline of thesis 2 0.2. Organization of thesis 3 Chapter 1 5 1.0. Outline of Navajo 5 1.1. Syntax 5 1.2. Data and Methodology 7 Chapter 2: Literature 12 2.0. Introduction 12 2.1. Previous analysis ofyi- 13 2.1.1. Passive 13 2.1.2. Inverse 14 2.1.3. Grammatical Relation mapping 20 2.1.4. Pronoun vs. Agreement marker 27 2.2. Unified generalization 29 Chapter 3: Proposal 30 3.0. Introduction 30 3.1. Possessive yi- must have an Antecedent 31 3.2. Licensing condition #1: yi- must be A'-bound 32 3.3. Licensing condition #2: yi- cannot be A-bound 34 3.4. Yi- must be A-free in any domain 38 3.5. Summary 40 3.6. Yi- as a Disjoint Anaphor 42 3.7. Conclusion 42 iii Chapter 4: Possessive Anaphora and lst/2nd person 44 4.0. Introduction 44 4.1. Problem 45 4.1.1. 1st and 2nd person disallow anaphora 45 4.1.2. Plural antecedents disallow anaphora 47 4.1.3. Yi- cannot have a lst,2nd or plural antecedent 48 4.1.4. Generalization 49 4.2. Solution 50 4.2.1. Sub-solution: yi- is inherently 3rd person singular 50 4.2.2. Summary of problem 51 4.2.3. Solution: Disjoint antecedent 53 4.3. Conclusion 55 Chapter 5: Comparison with Dogrib (Northern Athabaskan) 56 5.0. Introduction 56 5.1. Disjoint Anaphors (DA) as Non-subject Possessors 56 5.2. Subject Possessor 57 5.3. Summary 57 Chapter 6: Conclusion 59 6.0. Summary of Thesis 59 6.1. Implication for future research 59 References 62 iv Acknowledgement I'm thankful to my grandmother (Alice Begay Nez) who gave us every reason to speak with purpose. She told me that our sacred language came from White Shell Woman, and that no matter where I am as long as I speak my language, the gods would always recognize me. Thanks to the consultants, who spent many hour discussing language, art, customs, and thought with me. Special thanks to Robert Young and William Morgan Sr. Their dedication and work on the Navajo language continues to be recognized by linguists and the Navajo people today. I am eternally in debt to; Henry Davis, without his encouragement, and push none of this would have ever made sense to me, Hamida Demirdache, for her enthusiasm and careful guidance, Leslie Saxon, whose work inspired me. I'm grateful to Rose-Marie Dechaine, who has supported me in so many ways right from the beginning and . also Martina Wiltschko, who was always there at the right time. I would like to thank participants of the Navajo Language Academy; Ken Hale, Margaret Speas, Carlota Smith, Ted Fernald, MaryAnn Willie, Ellavina Perkins, Paul Platero, Lorene Legah, Alyse Neundorf, Linda Platero. I would like to acknowledge Eloise Jelinek who sent me on the road to linguistics. A special thanks to fellow graduate students Tomio Hirose, Eleanor Blain, Elizabeth Currie, Darin Howe, Nike Ola, Sandra Lai, Maxine Baptise, Leora Bar-el. Also the students in the Navajo Field Methods class (W1997); I was motivated by their enthusiasm. To my mother Linda Henley and aunties Edith Simonson, and Lorraine Herder for there language ability and humour. Both of them, never hesitated to lead me back to traditional medicines and healing ways, when I stumbled and needed help; no one should ever be without aunties like mine. To my late grandfather, Hataii Nez, 'Tall Singer', (Horseherder's Son The First) who never compromised his ways. The rest of my family, for their humour and stories. And finally to all those that came before, that never gave up the struggles so that I could be here today. This thesis is for Jessica, my nieces, nephews; the next generation of speakers. v 0.0. Introduction In Navajo there are two 3rd person possessor prefixes, yi- and hi-. The best documentation of the usage of possessor yi- and bi- is in the work of Young & Morgan (Y&M)(1980, 1987). Both^/- and bi- attach to a noun base, to refer to the possessor of the noun (la and b). l)a. bichiih, 'his nose' bima, 'his mother' b. yichjjh, 'his nose' yima, 'his mother' Bi- as a possessive pronoun exhibits the ability to refer independently. The definition of'yi- seems more complex. Yi- as a possessor prefix is described by Young & Morgan (Y&M)(1987) as a prefix attaching itself to noun bases, and "serves to distinguish between two nouns either of which might otherwise be construed as the possessor". In other words through the use ofyi- or bi-, Navajo disambiguates between two third person possessors. The following data from Y & M (1987 p.9) are a minimal pair that illustrate the use of yi and bi: l)a. shinaai shilnaa'aash yimasani yii yaa nat'aash l.poss.brother l.poss.cousin 3-poss.grandmother 3.with 3.for 2.du.returned 'my older brother; took my cousinj back to hisj grandmother' b. shinaai shiinaa'aash bimasani yii yaa nat'aash l.poss.brother l.poss.cousin 3-poss.grandmother 3.with 3.for 2.du.returned . 'my older brother took my cousinj back to hiss grandmother'[Y&M 1987] 1 The data in la/b) is a straightforward example of how the Navajo possessive prefix distinguish between two potential possessors with the use of yi, and bi-. In a) yi-prefixed to the NP 'grandmother' refers to the cousin as the possessor of the grandmother. In b) bi- refers to brother as the possessor of grandmother. Young and Morgan began their work on the Navajo language in 1937, and as a result produced among other publications, the most comprehensive work to date, The Colloquial Dictionary (1980, 1987) and Analytical Lexicon of Navaio (1991) with Sally Midgette. Other descriptions of possessory/- and bi- are in Father Berard Haile's book Learning Navajo (1941). In his book, Father Haile documented yi as a possessor prefix, used primarily on kinship terms. However it is clear from Young and Morgan (1987) that yi is used with more than just kinship terms, including animals and proper names such as 'Mr. Little'. Although extensive recent work has been done on Navajo yi- and bi- prefixed to verbs and post-positions, description of possessive yi- and bi- are limited to these earlier publications. 0.1. Goals and outline of thesis The goal of this thesis is to formalize Young and Morgan's observations on the 3rd person possessor prefix yi-. This thesis proposes that Navajo yi-, unlike bi-, has the properties of both a pronoun and an anaphor. The main difference between the two prefixes is that bi- behaves like a pronoun, while yU is an anaphor which must have an A'-antecedent with which it is co-indexed. Not only musty/'- be an A'-anaphor, but 2 it must also be an A-type Disjoint Anaphor (DA), following Saxon's (1986) analysis for its cognate in Dogrib (Northern Athabaskan)): that is, it must have an A-antecedent from which it must be disjoint in reference. Hence, I propose that Navajo yi- is subject to both Conditions A and C of Generalized Binding Theory (Aoun 1985), while bi- as pronoun is subject to condition B. 0.2. Organization of thesis The thesis is organized as follows: Chapter 1 provides a brief outline of Navajo syntax, and a discussion of the Y & M data on possessor^/'- and bi-. Included in Chapter 1 is information on consultants. Chapter 2 discusses previous studies on yi-(yi- and bi-). This chapter sets out necessary background by summarizing previous analysis of yi-. These studies have primarily focused on yi as the third person object of a verb, as opposed to possessor /^', which attaches to nouns. Hale (1973) analyzes verbal yi-lbi- in terms of a process of Subject Object Inversion (SOI). This very important study lays the groundwork for later analyses of yi- and bi-. Willie (1991) argues for the pronominal argument hypothesis (PAH) by claiming thatyi- and bi- are pronominal arguments, and that overt NP's are adjuncts co-indexed with the arguments. Speas (1988) analyzes bi- as a pronoun, and yi- as an agreement marker. Other work onyi- and bi- includes Platero (1974,1982), and Perkihs(1978). Chapter 3 introduces and motivates Navajo yi- as a Disjoint Anaphor (DA), the main proposal of the thesis. Original to this research is the proposal that.y/- is subject to both conditions A and C of Generalized Binding (Aoun 1985). Chapter 4 is a 3 discussion of 1st and 2nd person antecedents. I show that binding is disallowed by antecedents that are not specified for 3rd person singular; more specifically, yi- must have two antecedents and both must agree in phi-features. Chapter 5 compares the Navajo disjoint anaphor (DA) to disjoint anaphora in the Northern Athabaskan language Dogrib (Saxon 1984, 1986, 1995). In the final chapter (6) I conclude by summarizing the thesis and discuss briefly the implications of this research in terms of verbal yi-. 4 Chapter 1 1.0. Outline of Navajo Navajo, a Southern Athabaskan language, is spoken in the Southwest United States. The Navajo Nation extends into three states: what are now Northern Arizona, Western New Mexico and Southern Utah. The Nation's enrollment is more than 200,000. The Navajo Language like all native (First Nations) languages, is endangered. There are fewer speakers in each generation of Navajos. A short grammatical sketch of the language is given in Sec. 1.1. 1.1. Syntax Navajo may be described as a radical head marking language. In head marking languages, grammatical relations are marked on the head rather than on a dependent of the head, (la and b) illustrate head marking, where subject and object are marked on the verb by pronominal affixes. Note there are no overt nominal arguments. l)a. shich'id lS.O+2S.S+stem 'You are scratching me' b. nishch'id 2S.O+lS.S+stem T am scratching you' [Y&M 1987p,64] 5 In Navajo, pronominal affixes are obligatory, and lexical DPs may be freely omitted. (2a/b) illustrate sentences with overt DP's. (2c) is a typical sentence where overt DP's are omitted. 2)a. kii bima 'ayoi 'ayo'ni Kii poss.mother 3.3.loves Kii loves his mother b. bima 'ayoi 'ayo'ni poss.mother 3.3.loves He loves his mother c. 'ayoi 'ayo'ni 3.3. loves He/she loves him/her [Parsons, Speas 1994 p.l5n.36] Word order in Navajo is quite rigidly SOV. In other words, DP's are not freely ordered, so a sentence such as (3 a) does not have the same interpretation as (3 b): 3)a. Mela Dugi yiyiiitsa S O V Mela Dugi 3.3.saw Mela saw Dugi *Dugi saw Mela b. Dugi Mela yiyiiitsa S O V Dugi Mela 3.3.saw Dugi saw Mela *Mela saw Dugi As a head final language, Navajo is characterized by post-positions rather than prepositions. 6 4)a. ashkii at'eed yichT yalti' boy girl 3.to 3. S.talking 'The boy is talking to the girl' [Perkins 1978p.l04] 1.2. Data and methodology The first step in this research was to establish the consultant's judgement of possessive yi- and bi-. What is important is that a speaker have consistent judgements. If a speaker is not consistent or is unsure of the usage of yi- versus bi-, then complex data would be more problematic. I initially re-elicited the following data from Young and Morgan (1987), in order to establish a speaker's knowledge of the usage of yi- and bi. This re-elicitation raised many interesting discussions about speaker intuitions. These discussions continued to be crucial throughout the rest of the study. For example there was a consensus that context was important: speakers would consistently offer some context to the data in question. While bi- could be potentially ambiguous, yi- was consistently judged to be unambiguous. It was surprising, however, that the youngest speaker in the study (17 years) could not distinguish the use of yi- versus bi-. In (5) there are 5 minimal pairs that show the usage of yi- versus bi-. Contra Haile (1941) and Perkins (PC 1989), the data show thatyi- may refer not only the kinship terms (1, 3,4) but also to animals (2) and proper names (5). The complete set of minimal pairs in Y & M (1987:p9) contains two more sets of sentences with yi-lbi-attached to animals and one with an inanimate noun 'house1. This confirms that yi is not necessarily limited in the type of noun bases it can attach to. 7 5)l)a. shinaai shiinaa'aash yimasani yii yaa nat'aash l.poss.brother l.poss.cousin 3-poss.grandmother 3.with 3.for .3.du.returned 'my older brother; took my cousinj back to hisj grandmother' b. shinaai shiinaa'aash bimasani yii yaa nat'aash l.poss.brother l.poss.cousin 3-poss.grandmother 3.with 3.for 3.du.returned 'my older brother; took my cousinj back to hisi grandmother' 5)2)a. gaagi dibe yazhi yichpi neiniitash crow sheep little 3.poss.nose Rep.3.peck 'the crow is pecking on the lamb's nose' b. gaagi dibe yazhi bich#h neiniitash crow sheep little 3.poss.nose Rep.3.peck 'the lamb is tapping on the crow's nose' 5)3)a. shideezhi bimasani yighandi yii nat'aazh l.poss.sister 3.poss.grandmother 3.poss.house 3.with 3.du.return 'My younger sister; took her grandmother to herj home' b. shideezhi bimasani bighandi yii nat'aazh l.poss.sister 3.poss.grandmother 3.poss.house 3.with 3.du.return 'My younger sister; took her grandmother to her; home' 5)4)a. shinaai shimasani yiyeel ya neinigi l.poss.brother l.poss.grandmother 3.poss.bundle 3.for 3.3.returned (inanimate) 'My older brother carried my grandmother's bundle back for her' b. shinaai shimasani biyeel ya neinigi l.poss.brother l.poss.grandmother 3.poss.bundle 3.for 3.3.returned (inanimate) i. 'My older brother carried my grandmother's bundle back for her' *(potentially ambiguous) ii.'My grandmother carried my brothers bundle back for him' 5)5)a. hastiin tso hastiin yazhi yileechaa'i yits'aa' yiskah man big man small 3.poss.dog 3.away.from 3.3.shot 'Mr. Big shot Mr. Little's dog.' 8 b. hastiin tso hastiin yazhi bileechaa'i yits'^a' yiskah man big man small 3.poss.dog 3.away.from 3.3.shot 'Mr. Big shot Mr. Little's dog.' [Y&M 1987 p.9] I will show in this study how distinct yi- and bi- are in disambiguating potential possessors in NP. This study is particularly important because it documents the appropriate use of yi- as opposed to bi- amongst all the speakers. In the next generation the number of speakers who can differentiate between ayi- sentence and a bi- sentence may be zero. The consultants in this study are from various backgrounds. All of them speak Navajo as their first language and continue to use it as a primary language. The oldest speaker speaks no English at all. All examples in the text not specifically attributed to a particular author have been checked with one or more of these speakers. Consultants: Alice (Begay) Nez Born about 1907, dzii nitsaa. Married to Hatalhie Nez about 1925. Continues to maintain a small flock of sheep, a farm and various winter and summer camps for horses and cattle. Typically she walks about 2-3 miles a week and is constantly working on a new weaving project, usually saddle blankets. Grandma Alice told me she had an opportunity to go to school at about age 14 or 15, but was never interested. 9 Edith Simonson Born about 1949 in Tse binit'aahotso. Educated to about grade 8 from Gallup Jr.High School, Gallup, New Mexico. Today she helps maintain the family sheep, cattle, horses, a cornfield and various winter and summer camps. She has two adult children and 5 grandchildren. Four of five grandchildren are bilingual. She is an accomplished header and weaver. Linda L.Henley Born about 1947 in Tse binit'aahotso. Linda has a B. A in Education and an M.A. in Special Education. Currently she works for Pinon Elementary School in Pinon, Arizona. She returns to mothers residence (about 15 miles N.W. of Pinon) regularly so that she can help with farm, and livestock, and keep her mother company. She has two children and one grandchild. I also interviewed the following younger speaker in order to ascertain whether the yi-Ibi- alternation described in this thesis was being learnt by the younger generation of Navajo Speakers. Valencia Herder Born in 1982 in Tuba City Arizona. Valencia is a student at Flagstaff High School, Flagstaff Arizona. She returns home to her mother's home regularly (about 120 miles N.E. of Flagstaff). She was raised with a strong Navajo language background and maintains a high level of fluency; at what level is yet to be determined. It is clear she understands complex conversation with minimal 10 clarification. She is interested in becoming literate in Navajo. Valencia's hobbies include beading and creating dance regalia. The following speakers and speaker-linguists have had a tremendous influence on the data and analysis used in this research: Mary Ann Willie, Ellavina Tsosie Perkins, Lorene B. Legah, Lorraine Herder, Robert Young, Ken Hale. 11 Chapter 2 Previous Analyses of the Yi-/Bi- Alternation 2.0. Introduction Yi- and bi- are generally described as 3rd person pronominal prefixes. Both prefixes may attach to a verb (la and b), post-position (2a and b), or noun base (3a and b)1. Verb l)a. shiyaazh shash yinalzid l.poss.son bear 3.3.fears 'My son fears bears' b. shiyaazh shash binalzid l.poss.son bear 3.3.fears 'My son, the bear is afraid of him' Post-position 2)a. 'awee' tsask'eh yikaa' sida baby bed 3.on 3.sit (animate) 'The baby is sitting on the bed' 1 Data such as the following in (ia) illustrate multiple instances of yi-, where yi- is attached to a verb, a post-position, and a noun. Although this is typical of yi, multiple instances of A;'-are not possible, (i.b)isan attempt to attach bi- to the verb, post-position, and noun, and the result is ungrammatical. (i.e) illustrates bi-attached to both the noun, and the post-position, and still the sentence is odd at best, (i.d) is a grammatical sentence with bi- attached only to the noun. i)a. hastiin tso hastiin yazhi yile6chaa'i yits'44' yiztai man.Big man.Little 3.poss.dog 3.from 3.3.Pf.kick 'Mr. Big kicked Mr. Little's dog. [adapted from Y&M 1987 p9] b. hastiin tso hastiin yazzie bileecha'i bits'aa biztai c. hastiin tso hastiin yazzie bileecha'i bits'aa yiztal d. hastiin tso hastiin yazzie bileecha'i yits'aa yiztai 'Mr. Big kicked Mr. Little's dog* 12 b. 'asaa' tse bikaa' si'a pot rock 3.on 3.set (inanimate) 'The pot is set on top of the rock' [Perkins 1973, p. 119:31] Noun 3)a. shideezhi bimasani yighandi yii nat'aazh l.poss.sister 3.poss.grandmother 3.poss.home 3.with 3.du.Pl.returned 'My younger sister took her grandmother to her home (sisters home)' b. shideezhi bimasani bighandi yii nat'aazh l.poss.sister 3.poss.grandmother 3.poss.home 3.with 3.du.Pl.returned 'My younger sister took her grandmother to her home (grandmother's home)' [Y&M 1987 p9] 2.1. Previous Analyses of Yi- (yi-/bi-) 2.1.1 Passivization The alternation of the third person pronouns yi-lbi- has been the focus of many studies in Navajo syntax. Yi-lbi- was described as an active/passive alternation in Reichard (1951). In passivization, the subject gets demoted and the object gets promoted. One implication of this is that an active transitive sentence becomes intransitive when passivized. Passivizaton triggers demotion of the subject to an adjunct and promotion of the object to subject, hence the change in word order from SOV to OSV. In the data below (4a) is analysed as active while (4b) is passive. 4)a. 'ashkii at'eed yiyiiitsa boy girl 3-3.Pf.see ACTIVE 'The boy saw the girl' note that i(d) has exactly the same interpretation as i(a). Since this issue is orthogonal to the main topic of the thesis, I leave it for future research. 13 b. at'eed 'ashkii biiitsa girl boy 3-3.Pf.see PASSIVE 'The girl was seen by the boy This description of yi-lbi- was quickly abandoned, since an independent detransitivized passive construction was discovered in Navajo, whereas the yUlbi- alternation shows no evidence of detransitivization. Later yi-lbi- was analyzed as obviative/proximate by Hale (1968). The NP closest to the verb is obviative (4a) if yi- is used, and the NP closest to the verb is proximate if bi-is used, as in (4b). This analysis was only a precursor to Hale's later analysis of Subject-Object Inversion (SOI), which linguists continue to refer to as the 'landmark' analysis of yi-lbi-. 2.1.2. Inverse In Inversion no demotion takes place. Instead the change in structure is due to the fact that the language is discourse sensitive (i.e. topic versus non-topic). Inversion often uses a nominal hierarchy, which has also been claimed to exist in Navajo. Hale Hale (1973) analyzes the yi-lbi- alternation in terms of Subject/Object Inversion (SOI). Simply, stated, this involves the inversion of noun phrases, as well as a change in the third person object prefix on the verb, from yi- to bi-. Although the syntactic rule is similar to the passive construction, yi-lbi- constructions are transitive. In other 14 words, (4b) should be interpreted as either 'the boy saw the girl' or 'the girl, the boy saw her' instead of 'the girl was seen by the boy'. This can also be seen in the following where (5a) follows the basic SOV word order and (5b) shows OSV word order. 5)a. i j i ' dzaaneez yiztai horse mule 3-3.Pf.kick DIRECT (SOV) 'the horse kicked the mule' b. dzaaneez 1$' biztaf mule horse 3-3Pf.kick INVERSE (OSV) 'The mule was kicked by the horse' It seems that subject-object inversion is a free option in simple sentences as the following sentences show: 6) a. ieechaa' i mdsi yishxash (SOV) b. mosi ieechaa/i bishxash (OSV) 'The dog bi-t the cat' 7) a. 'azee'iii'ini 'ashkii yighadi'niidlaad (SOV) b. 'ashkii 'azee'iii'ini bighadi'niidlaad (OSV) 'The doctor x-rayed the boy' However, the rule of SOI is not completely unrestricted. Although it yields well formed sentences in (7a and b) above, the following transitive sentences show that the rule is not entirely optional. In (8a and b) only they/'- form is grammatical. 8)a 'ashkii to dilchxoshi yoodl44' (SOV) 'The boy drank the soda pop' b. Ieechaa'i ieets'aa' yiinaaad (SOV) 'The dog is licking the plate' 15 Hale points out that the inverted sentences corresponding to (8 a and b) above (with bi-object marker) are unacceptable. The ungrammatical sentences are in (9a and b) below: 9) a *to dilchxoshi 'ashkii boodk|4' (OSV) b. *ieets'aa' ieechaa'i biinaaad (OSV) At this point there are two possifo'-lities as to what restricts inversion. The condition might have to do with either (i), the noun phrases, or (ii), the verb. Hale concludes that inversion has to do with the noun phrases. Based on the data, transitive sentences with inanimate objects ('pop, plate' (8a and b)) cannot undergo inversion. It turns out that there are more conditions to the syntactic rule. The following data illustrate that in cases where the logical object is animate, the inverted word order is grammatical. The following data show the inverted order with the bi- prefix: 10) a. shini' sa biisxj , (OSV) 'Old age killed my horse' b. dibe to 'abiii'eel (OSV) 'The water swept the sheep off In these cases, it is the SOV word order with _yz'- that is unacceptable. The unacceptable sentences are illustrated in (1 la and b) below: 11) a. *s4 shilij' yiyiisxj (SOV) 16 'Old age killed my horse' b. *to dibe 'ayiii'eel (SOV) 'The water swept the sheep off However the following sentences, using the same verbs as in (11 a and b), are grammatical. Hale points out that this implies that it is not the verb which is responsible for this behavior. 12)a. ndshdoitsoh bjjh yiyiisxj (SOV) 'The mountain lion killed the deer', b. to tsin 'ayuTeel (SOV) -'The water swept the stick off. It appears that the condition on inversion has to do with animacy. That is, inversion is not felicitous if the object is inanimate, and it is not required if the object is animate (with the exception of 1 la and b). To account for the ungrammatically of (1 la and b) above, if the logical subject (or agent) is inanimate, and the logical object (patient) is animate, then we must assume that SOI is obligatory. These generalizations are summarized in table (13) below: 13) Subject animate animate inanimate inanimate Object animate inanimate inanimate animate Inversion optional blocked blocked obligatory [Hale 1973p.303] 17 When both NP's are animate than inversion (bi-) is optional. If the subject NP is animate and the object NP is inanimate, then inversion is blocked. If both NP's are inanimate than inversion is also blocked. If the subject NP is inanimate and the object NP is animate then inversion is obligatory. Animacy and Inversion Hale observes that there is a correlation between animacy and inversion. However, compare the generalizations summarized in (13) with the following data. According to the table, only the data in (14) below should be possible with the inverted (bi-) form, that is, if we consider nominals such as 'lightning' and 'rain' to be inanimate. Instead, nominals such as 'lightning' and 'rain' behave as equal in rank with animates, i.e. people. In the data below either the direct form (SOV) in (15) or the inverted forms (14) are acceptable. 14) a. fu' ' i i 'ni ' biisxi (OSV) 'Lightning kill the horse' b. shiye' niits^ nabjstiee' (OSV) 'Rain wet my son' 15) a. ' i i 'ni ' I i i ' yiyiisx'i (SOV) 'Lightning killed the horse' b. niitsa shiye' naistiee' (SOV) 'Rain wet my son' [Hale 1973] The data thus far show that the conditions on inversion cannot be formulated in absolute terms. Hale concludes that the conditions may not be syntactic or semantic 18 but could be an aspect of Navajo thought, therefore beyond the control of non-native speakers of Navajo. Hale's systematic analysis of Xheyi-lbi- alternation continues to be one of the most referenced works on yi- and bi-. A Note on Noun Hierarchies and Animacy It is apparent that noun-type (i.e. animate versus inanimate) has something to do with the possibilities of inversion. Hale's work, amongst others, has led many researchers to postulate various animacy hierarchies (Frishberg 1972, Salego 1977, Wetherspoon 1977). In this type of hierarchy, nouns are ranked according to power, intelligence, movement etc. For example, Creamer (1974) postulates eight levels of rank, as illustrated in the following: Group 1: human beings where infants are of lower rank than their elders, and high potency forces such as lightning are ranked equal to man. Group 2: large and medium-sized animals with noted intelligence Group 3: medium-sized animals Group 4: small animals Group 5: insects Group 6: natural forces Group 7: plants and inanimate objects Group 8: abstractions Other similar ranking systems have been proposed, but the fact is that not one hierarchy can be agreed upon. This is because tests of native speakers have been inconclusive. This way of explaining the yi-/bi- alternation does not always predict the correct prefix. 19 I hypothesize that agentivity/control, and not an animacy hierarchy, is the key to . explaining the yi-/bi- alternation. One of the generalizations that Hale (1973p.305) makes is that it might make more sense to rank nominals according to the extent in which that noun is capable of movement, activity, or causation. This would explain why 'lightning and rain' as inanimates rank as equals with animates. This reasoning makes more sense to me as a native speaker. Animacy hierarchies seem to rank entities according to a scale where some beings are inherently superior to others. This has been used to draw conclusions about how Navajos view the world. However, to my mind this view does not correspond to Navajo thinking. A more appropriate way of understanding the rankings is to view them relative to individual situations. For example, as 'nilts^', rain', water is capable of movement, activity and causation, whereas as 'to, water as in a puddle' it is not. Ranking an entity according to its abi-lity to act on another or cause something is a fact of life: lightning can kill, and bees do sting. In this view a rigid noun hierarchy is eliminated, and each event is evaluated according to the agents and patients involved, in other words in terms of what actually happened instead of what someone else thinks can happen. I think this view coincides with Navajo thought rather than the notion that men rank higher than babys, sheep after horses etc2. Ranking smells to me of imperialism. 2.1.3. Grammatical Relations (GR) Mapping Perkins In conversation, sentences with two overt DPs are rare and awkward in Navajo. Therefore it should be noted that data with two DPs are independently somewhat infelicitous, further complicating the issue of animacy ranking between two DPs in the same sentence. 20 Perkins (1978) adopts Hale's (1973) analysis (with some modifications). However, she points out that the principles set forth by SOI do not account for more complex sentences such as those with indirect objects or post-positions, as the following examples illustrate: (16)a. Jaan hastiin asdz§ yeiniltj John man woman 3-3.Pf.give(AN) 'John gave the man the woman' b. Hastiin jaan asdz4 beiniifj man John woman 3-3.Pf.give(AN) 'John gave the man the woman' SOI would wrongly predict in (16a) that NP2 is the object and in (16b) NP1 is the object. In both cases, however, they are the indirect object. Perkins rewrites the principles from the point of view of the assignment of grammatical relations: i.e., where NPs are assigned the relation of subject, object and indirect object. She sets out new principles to deal with the yi-/bi- alternation. For full details, see Perkins (1978:psl 11-137). With these principles, Perkins can handle data with indirect objects; this seems to be an improvement over SOI. However, these principles fail to account for many other types of data. In data with post-positionals, SOI seems to make the correct predictions, as seen in the following sentences: 21 17)a. ashMi at'eed yichT yaiti' boy girl to 3.Pf.talk 'The boy is talking to the girl' b. at'eed ashkii bich' j ' yaiti' girl boy to 3.Pf.talk 'The boy is talking to the girl' However, as it turns out, SOI doesn't always make the correct prediction with post-positionals. Perkins points out that SOI incorrectly predicts the yi- form in (18b). 18)a. awee'tse . *bi -kaa'tsits'aa'bii si'4 yi baby rock yi/*bi- top box with 3.set(IN) 'The baby is in the box (set) on the rock' b. asaa' tse *yi -kaa' tsits'aa' bii si'4 bi pot rock *ydbi- top box with 3.set(IN) 'The pot is in the box set on the rock' To deal with this problem, Perkins proposes another principle to deal specifically with post-positions of location and direction. Possessive phrases also display the yi-lbi- alternation. Perkins claims that>7- forms uniformly mark the first NP as the subject and bi- forms uniformly mark the second NP as the subject, whether they occur on verbs, post-positions, or nouns. This principle is adequate for the following examples: 19)a. Hastiin Baa' yilij' yizloh man Baa' Poss.horse 3-3.Pf.rope 'The man roped Baa's horse' 22 b. Hastiin Baa' bilii ' yizloh man Baa' Poss.horse 3-3.Pf.rope 'Baa' roped the man's horse' However, as Perkins points out, it fails to account for the following data: 20)a. *Jaan Baa' yima yishxash John Baa'Poss.mother 3-3.Pf..bit 'John bit Baa's mother' b. Jaan Baa' bima yishxash John Baa' Poss.mother 3-3.Pf.bi.t 'John bit Baa's mother' [Perkins 1978] In the cases above the principle incorrectly predicts that the second NP is the subject in (20b); therefore Perkins concludes that "kinship possession is an exception" to the principles. Now if these principles are applied only to yi- on verbs (yz'-V) then they make the correct predictions: the first NP is the subject in both (20a and b). However, they make the wrong predictions for possessives, as Perkins herself discovered. Instead we must assume that possessive yi- and bi-, although related to yi- and bi- on verbs, are used differently. Specifically, a possessor must be potentially ambiguous between two NPs for yi- to be relevant (Y&M 1987p9). This is not the case in the data in (20a and b); in other words, if (20a) were grammatical it would have the same interpretation as (20b). 23 Platero Like Perkins, Platero (1974, 1982) discusses the yi-lbi- alternation in terms of the interpretation of grammatical relations (IGR). In his analysis, yi- and bi- assign grammatical functions to the sentence arguments. As shown below, if the prefix \syi-, then NP b is Subj and NP a is the Obj, wheras the reverse order is applicable for bi-. 21)a. dzaaneez yiztai horse mule 3-3.Pf.kick 'the horse kicked the mule' NP b NP a y/'-verb IGR: [s Subj a g Obj p a t y/'-verb] (b)dzaaneez f#' biztai mule horse 3-3Pf.kick 'The mule was kicked by the horse' N P b N P a 6/'-verb [s Obj pat Subj a g Z>/-verb] Willie Willie (1991) bases her analysis on Jelinek's (1984) Pronominal Argument Hypothesis, which I briefly introduce here. The Pronominal Argument Hypothesis (PAH) is based on the empirical observation that the following two properties are often correlated cross-linguistically: i) obligatory rich inflection ii) optionally realized arguments (i.e. full DP's) According to the PAH, these are analyzed theoretically as follows: a) inflectional affixes are arguments b) full DP's are adjuncts coindexed with the pronominal arguments 24 The following trees illustrate the difference between a Lexial Argument language like English and a Pronominal Argument language (PAL); Following the PAH, Willie proposes that (yi-lbi-) are in argument position in examples like (23). 23) yiyiiitsa 3.3.saw 'S/he saw him/her' What yi- and bi- do is trigger the different mapping rules that specify how semantic roles are assigned to the pronominal argument positions. Willie explains that nominal expressions are added to the structure not for grammatical reasons, but for discourse reasons. When there are overt nominals, then the subject and object NP are adjoined to the sentence and co-indexed with the "pronominal arguments". 25 Willie explains the yi-lbi- alternation in terms of a Direct and Inverse voice alternation. According to Willie, the closest NP to the verb is associated with the closest pronominal argument position to the verb. In the following sentence in (24a) yi, is assigned the agent theta role and is therefore the Subject pronominal argument, while the Object argument is assigned patient theta role. In (24b) with bi-, the theta assignment is reversed. 24)a. i#' dzaaneez yiztai horse mule 3-3.kicked 'The horse kicked the mule' b. dzaaneez i%' biztai mule horse INV.3-3.kicked ' The mule was kicked by the horse' [Willie (1991): 71 ] In the yi- form the inner nominal adjunct (dzaaneez, 'mule') is coindexed with the internal argument, and the outer nominal "by default" (Hi, 'horse) is coindexed with the external argument. The word order in (24a) is SOV, hence the Direct.construction. In (24b) the NP occurring nearest the bi- is the agent adjunct of that sentence, and the other nominal is a patient adjunct. Willie concludes that in both voice forms the immediately preceding NP is an adjunct to the internal argument, regardless of theta role assignment. These nominals are in adjunct position, not in argument position. The noun does not have grammatical relations independently of the pronominal argument, that is nouns are adjuncts co-indexed with the "incorporated pronoun". According to Willie, the inverse bi-, appears when there is more than one third person argument. The difference between Direct/Inverse and Active/Passive is that yi-lbi-26 constructions are active sentences, so in other words the inverse is like a passive in that the subject has a patient theta role, but unlike a passive in that it has both agent and patient arguments, and thus remains a fully transitive construction. Therefore the yi-Ibi- alternation is not an Active/Passive construction. 2.1.4. Pronoun v.s. Agreement Speas. Different to any other analysis is Speas' (1991) proposal that bi- is an incorporated pronoun andyi- is an agreement marker. She proposes that bi- is a pronoun which occupies the object position at D-Structure and is incorporated into the verb. Then the NP directly preceding the verb must be the subject. Accordingly, the first NP is dislocated, and associated with the /3/-pronbun. This is what happens in an English sentence like (i): i) 'John, Mary saw him'. Under this proposal, the sentences in (25a,b and c (i)) have the structure in (ii): 25)a. i. 'ashkii 'at'eed biiits^ ii. 'ashkiij ['at'eed bi; -iitsa] boy girl 3.- saw 'The boy;, the girl saw himj' b. i. biiits^ i i . [pro bi -iitsa] 3. -saw 'Him/her, s/he saw' 27 c. i. 'at'eed biiltsa ii. ['at'eed bi -iitsa] girl 3-3.saw 'Him/her, the girl saw' 1 [Speas 1990 p261:102] Speas claims that (25 a) (which is a sentence with a iz'-verb and two overt NPs) is like left dislocation in English (see ex.(i)). The leftmost NP must be coindexed with the pronoun bi-, in order to be licensed. In (25b) bi- occurs without an overt NP. In this case bi-, the overt pronoun, receives an emphatic interpretation . In (25 c) the overt NP is the subject of the sentence while bi- is the object. The following data in (26a - c) show that as an agreement marker, yi- does not itself occupy an argument position. Therefore, the DP with which it is associated is in argument position. When there is no DP, we assume the associated argument is pro. 26) a. 'ashkii 'at'eed yiyiiitsa boy girl 3.3.saw 'The boy saw the girl' b. yiyiiitsa 3.3.saw 'S/he saw him/her' c. 'at'eed yiyiiits4 girl 3.3.saw 'S/he saw the girl' Speas' analysis maintains that >7- and bi- have a different syntactic status, as opposed to the other analyses, which all assume thatj/'- and bi- are syntactically parallel. 28 2.2. Towards a unified generalization In this thesis, I will take a different perspective on the problem of how to unify the various environments in which the yi-lbi- alternation occurs. Instead of beginning with verb yi- and attempting to generalize to nouns, I will begin by providing the first detailed analysis of nominal yi-, which will be presented in chapters 3 and 4. I will return briefly to a comparison of possessive yi- with verby/'- in chapter 6. 29 Chapter 3 Proposal 3.0. Introduction I propose thaty/'- in Navajo.is a disjoint anaphor, as originally proposed by Saxon for its cognate in Dogrib (Northern Athabaskan) (see Saxon 1984, 1985, 1986). As a disjoint anaphor, I will show that yi- must have a local A-antecedent from which it is obligatorily disjoint in reference. I will also show, however thaty/'- must have a second A'-antecedent with which it is obligatorily coreferent. I will interpret the binding behavior of yi- in terms of Aoun's (1985) theory of Generalized Binding. More specifically I make the proposal thaty/'- must simultaneously satisfy Condition A (since it is an anaphor) and Condition C (since it must be A-free). I further claim that the anaphoric properties of the possessive pronoun yi- provide an argument that lexical DPs are in A-position, i.e. they are not adjuncts, as opposed to the claim of (amongst others) Jelinek (1984), Willie (1991). The properties of Navajo possessory/'- thus supply important evidence against the Pronominal Argument Hypothesis (PAH) Jelinek (1984). In contrast, I show that overt DPs in possessor positions occupy A'-positions, either as SPEC DP or adjuncts to DP 1. 1 Although I use DP to refer to nominal, noun phrases etc, I remain neutral between the NP and DP analyses. 30 3.1. Possessive j t - must have an antecedent In the following data, note the contrast between (la) and (lb). The only difference between (la), which is ungrammatical (if out of context), and (lb)2, which is grammatical, is the absense vs. presence of an antecedent, Joe, for >>/'-. Therefore the generalization is that yi- cannot stand alone: it must be co-indexed with a nominal antecedent. *(la) yi-lii' ya yizloh j>/-horse 3.for 3-3.Pf.rope 'hei roped hisk horse' (lb) Joe yi-lij' yizloh _yz'-horse 3-3.Pf.rope 'hef roped Joe'sj horse' The data in (2a) and (2b) establishes that this antecedent can be provided by the discourse context. (2b) is a possible response to (2a). (2a) haayit'eego Frank Joe yika'elwod Q.in.what.way Frank Joe 3.for/after.run 'How did Frankf help Joej' (2b) yi-lft' yaa yizloh _y/'-horse 3.for 3-3.Pf.rope 'hef roped hisj horse for him/ 2 the interpretation [Joe; roped [hiSj horse]] is possible for some speakers. However this reading is hard to get. This is in contrast to Dogrib, where the reading is readily available (see chapter 5). I leave this issue for further research. 31 3.2. Licensing Condition #1: yi- must be A'-bound How do we reconcile (J) and (2)7 We must force yi- to be co-indexed with Joe in order to explain the ungrammaticality of (la) but at the same time allow this antecedent to be provided by the discourse context to explain the grammaticality of (2b). This can be achieved if we make the assumption thaty/'- is obligatorily A'-bound and adopt Huang's (1984) proposal that discourse topics (e.g. Joe and Frank in (2a)) are syntactically represented as IP adjuncts A'-binding a pronominal, as shown in (2c): (2)c. discourse topic = prof yi-f Under this proposal: (la) is ungrammatical because yi- is not bound, (lb) is grammatical because yi- is A'-bound by the possessor Joe, and finally. (2b) is also grammatical because the discourse topic Joe is represented by a null topic, TOP, A'-binding yi-. 32 The structure of (lb) and ( 2 b ) are given below: (lb) Joe yi-Kj' yizloh IP prof ( 2 b ) yi-W yaa yizloh yi-j horseh It is important to note that possessors must be in A'-positions (Spec DP or adjunct to DP) for this analysis to go through, as shown above. 33 3.3. Licensing Condition #2: possessive j / - cannot be A-bound Let us turn to (2b), and look at all of its possible interpretations which are listed in (3). Context repeated: (2a) haayit'eego Frank Joe yika'elwod 'How did Frankf help Joe;' (2b) yi-Ej' yaa yizloh _y/'-horse 3.for 3-3.Pf.rope 'hef roped hisj horse for him/ (3) a.*'hej (Joe) roped hisj horse for himf', b* 'hef (Frank) roped hisf horse for him/ c. *'hef roped hisj horse for himjc' d. 'hef roped hisk horse for him/ (3a) is an infelicitous response to the question in (2a). In the same way, 'He (Joe) roped his (Joe's ) horse for him (Frank)', is not an appropriate response to, 'How did Frank help Joe?'. (3b) on the other hand is a plausible answer to the question (2a). Why then is (3b) bad? I propose that (3b) is ungrammatical because yi- (the possessor) is A-bound by prof, as shown in the tree diagram below: 34 (*3b) 'hef (Frank) roped hisf horse for himj' pro D yi-f horseh (3c) is bad, like (3a); it is an infelicitous response to the question in (2a). (3d) is impossible unless another third person is introduced. The easiest way to get this reading is by answering,' Kody yi-lft' yaa yizloh' to the question in (2a); in the case the interpretation 'Frank roped Kody's horse for Joe' is available. Without mentioning 'Kody' overtly in the response, the reading is ruled out altogether. In short the ability to get this type of interpretation depends heavily on the availability of appropriate discourse antecedents. Now let us turn again to (lb) and consider all its logically possible interpretations, which are listed in (4). (lb) Joe yi-lft' yizloh y/'-horse 3-3 .Pf. rope 'hef roped Joe'sj horse' 3 5 (4)a. 'hef roped Joe'sj horse' *b. 'Joej roped hisf horse' *c. 'Joej roped hisj horse (=hej roped Joe'sj horse) (lb) can have only one possible interpretation, (4a), which is licit since yi- is A'-bound by Joe. This is shown in the tree below. (4a) 'Hef roped Joe'sj horse' D yh NP horseh (lb) cannot have the interpretation in (4b), since yi- would be unbound, and yi- must be A'-bound. This is illustrated in the following tree: 36 (*4b) 'Joej roped hisf horse' IP agrj agrh rope A D NP yi-t horseh Crucially, (lb) cannot have the interpretation in (4c), although yi- in (4c) is A'-bound by TOP. Why is this interpretation ungrammatical? Because yi- is A-bound by the subject Joe. This point is illustrated in the tree in (4c). (*4c)'Joej roped hisj horse. IP agrj agrh rope A D NP yi-j horseh 37 Hence, I conclude that>>/- cannot be A-bound and must be A'-bound. Note: the proposal that (4c) and in particular (3b) are ruled out because yi- cannot be A-bound (by either pro in (3b) or Joe in (4c)) implies that lexical DPs (e.g. 'his horse') are in A-position. 3.4. yi- must be A-free in any domain Up to now, we have seen that>7- is like an anaphor since it has to have an antecedent, more precisely an A'-binder. However yi- must also be A-free. My next question is: mustyi- be locally A-free, like a pronoun (that is, in its governing category the minimal domain containing^/'-), ox A-free everywhere, like an R-expression?. Consider the following sentence, where yi- is in a complement clause: (5) Paul yi-i#' yiyiiitsa nizin Paulj/'-horse 3.saw.3 3.thinks/wants 'Hei thinks Paulk saw his*i/*k/j horse'. Yi- must be disjoint from both the subject of the main clause and the subject of the subordinate clause. Note that there must be a topic present in the discourse to license yi-, as in (2) above. In contrast, the English sentence in (6) is ambiguous and can mean either: (6a) 'Hei thinks Paulk saw hisk horse'. (Paul's horse) (b) 'Hei thinks Paulk saw hisi horse'. (c) 'Hei thinks Paulk saw hisj horse' 38 However, the Navajo sentence in (5) cannot have either meaning, (6a) or (6b). Instead (5) can only mean 'He thinks Paul saw someone's horse', where all DPs must have different indices, as shown in (6c). The fact that 'he' (matrix subject) cannot bind yi-shows thaty/'- in this complement clause is like an R-expression : it must be free everywhere. We have looked at data with they/'- in a complement clause. Now let us look at yi- in • * 3 an adjunct clause. Examine the following question-answer pair : (7)a. Haa'ii biniina Dugi Mela bich'i' bahachi' What reason.for Dugi Mela 3.towards. 3.angry 'Why is Mela angry with Dugi?' b. yi-leechaa'i yineez'j' -igii biniina bich'i' bahachi' 3.poss.dog that.3-3.stole-clitic reason.for 3.towards 3.angry i) 'She is angry because he stole her dog.' Literal translation: ii) 'He stole her dog that is why she is angry with him.' The coreference relationship beween pro (she) andy/'- (her) dog in (7b) might be construed as an A-binding relationship. However consider the literal translation of (7b) in (ii), and let us assume that the structure of (7b) is exactly as in its literal translation. Then (7b) will have the following structure: 3 note that the response with^i- in (7b) was induced;by<the OSV (bi-) pattern in the question in (7a). 39 (8) ^DJ?^ pro.i towards j / - m dog agr.iagrdstole K V P agr.mangry Notice that there is no A-binding relationship between 'her dog' and 'she', since pron does not c-command v/'-m. 3.5. Summary So far I have proposed that ayi- must satisfy conjointly conditions A and C of the Binding Theory. I derive the 2 licensing conditions governing yi- from the idea that yi- is an A'-anaphor and the proposal that as an A'-anaphor >v- must satisfy conjointly conditions A and C, of the binding theory, within the Generalized Binding approach of Aoun (1985). An informal statement of Generalized Binding is given below, together with the Navajo version I employ for Navajo. 40 Generalized Binding (where X = A or A') A. An X-anaphor must be X-bound in its domain C. R-expressions must be A-free Navaio Version An A'-anaphor must be A'-bound in its domain. R-expressions must be A-free. The A'-anaphor yi- must satisfy condition C, which requires that it be A-free, thus capturing licensing condition #2. It must also satisfy condition A which it does by taking the A ' value for X in A. Note that the only way it can satisfy conjointly condition A and C is if X = A ' . Thus yi- is an A'-anaphor which must be A'-bound in its domain, thereby satisfying licensing condition #1 . The analysis is dependent on the assumption that DPs are in A-position, not in adjunct position. Yi- cannot be A-bound by the matrix null (pronominal) subject, as shown schematically below: 'hei roped h i s^ horse' To capture this generalization we must assume that yi-ffi, 'his horse' is in argument position. Note, however, that the possessor must be an adjunct to DP, in order to allow the possessor to bind yi- in DP. 41 3.6. Yi- as a disjoint anaphor We have seen thaty;'- is an A'-anaphor, and therefore must have an A'-antecedent and be A-free. However, there is another condition on yz- that we have not yet addressed. As shown in (2 and 7) yi- must have two antecedents; this does not follow from the analysis presented so far. (9a) Hayit'eego Joe hooghan-di nadza How Joe home-enclitic 3.Pf.return 'How did Joe get home?' (b) * yiljj* hooghan-di bii nadza 3.poss.horse home-enclitic 3.with 3.Pf.return 'He rode home' (literally) 'His horse rode him home' As we seen in (9) above, yi- is ungrammatical even though (i) there is an A'-antecedent (Joe) and (ii) yz'- is A-free (since it occurs on a DP in subject position). Therefore, we need a third condition. Following Saxon (1984a. 1986,1995) I propose thaty/'- is a disjoint anaphor as well as an A'-anaphor. That is, it must have two antecedents: an A'-antecedent with which it corefers, and crucially an A-antecedent from which it is disjoint. In chapter 4,1 will further explore the implications of yi- as a disjoint anaphor. 3.7. Conclusion The 3 r d person prefix yi- in Navajo can attach itself to verbs, post-positions and nouns. In fact it is typical to see multiple instance of yi- in a single sentence. Yi- has been described and analyzed mainly in its role as a verbal prefix, and its function as a 42 prefix on nouns is less well known. In this chapter, I have presented a systematic analysis of possessory;-. 43 Chapter 4 Possessive Anaphora and 1st and 2nd person 4.0. Introduction We have already established that the 3rd person possessor prefix yi- is a disjoint anaphor subject to both Conditions A and C of Generalized Binding Theory (Aoun 1985). Yi- must satisfy condition A (since it is an A anaphor) and C ( since it must be A-free). Yi- cannot independently refer; it must be co-indexed with an overt A -antecedent, hence it is subject to condition A. As a disjoint anaphor^/- must also have an A-antecedent from which is is disjoint. Yi- can be properly bound by an antecedent that is provided by the discourse, which we represent as base generated in an A -postion (Topic). So if there is no context, i.e. no overt or discourse antecedent, then yi- is unbound and thereby yields an ungrammatical sentence, such as the following, which is repeated from Chapter 3. 1) . *yilft' yizloh 'he roped his horse1 In contrast, the context in (2) provides two overt third person DPs that license^/'-. Yi-is bound by an antecedent with which it can be co-indexed (grandfather) and one from which it must be disjoint (brother), and the sentence is grammatical. 2) . shitsili shichei yiKi' ya yizloh l.poss.brother l.poss.grandfather 3.poss.horse 3.for 3.3.Pf.rope 'My younger brother roped my grandfathers horse for him' 44 The following pattern summarizes the argument thus far: yi- is co-indexed with the nearest c-commanding DP in an A 1 position and is necessarily disjoint from a c-commanding DP. Note in an A-position, as shown in the schema below: A 3rd; Y'i A 3rdj 4.1. Problem 4.1.1. 1st + 2nd person disallows anaphora A potential problem arises in data where antecedents are provided by the discourse and yi- is still ungrammatical. The analysis predicts that (3 and 4) should be grammatical. Yi- in (3 and 4) has a plausible antecedent (Joe), yet given the context in (3 a and 4a), (3b and 4b) are still bad1 3)a. Hayit'eego Joe bika'eshwod 4)a. Hayit'eego Joe bika'iinilwod How Joe 3-l.Pf.run.after(help) Joe 3-2.Pf.run.after(help) 'How did I help Joe' 'How did you help Joe' 1 To be systematic, another possible pattern should be mentioned where the 1st person is the A'-antecedent for yi- and the 3rd person is the disjoint A-antecedent. The data in the following show that this is not a possible structure in Navajo. Instead the sentence Mela saw me' has pattern similar to (3 and 4) above. (i)a. hayit'eego mela shika'eelwod (i)b. * yilii' yizloh How Mela l-3.Pf.run.after(help) 3.poss.horse 3-3Pf.rope 'How did Mela help me' 'she roped my horse' (ii) shiKi' yizloh 'she roped my horse' In (i) yi- would have to have a 1st person A'-antecedent; however, it is inherently third person as shown in section 4.2.1. In (ii), I show a grammatical sentence with a first person possessor 'shi'\ this case is simply irrelevant to the issue under discussion. 45 b. * yitfj' seloh 3.poss.horse 1.3.Pf.rope 'I roped his horse' b. * yifij' siniloh 3.poss.horse 3-2.Pf.rop 'You roped his horse' Compare (3) and (4) with (5), which is grammatical. Note that the only difference between (3) and (4) on the one hand and (5) on the other is that the latter has two third person DPs as potential antecedents, while (3) has a third and a first person antecedent, and (4) has a third and a second person antecedent. 5)a. Mary John yiyiiits4 Mary John 3.3.Pf.see 'Mary saw John' b. yilji' ya yizloh 3.poss.horse 3.for 3.3.Pf.rope 'she roped his horse for him' Why does a first or second person2 antecedent disallow anaphora withy/'-? The data in (3) and (4) meet the requirements for binding: yi- is bound by the nearest c-commanding A'-antecedent, and obligatorily disjoint from another c-commanding DP in an A position. This is a problem because the analysis we've proposed so far will not rule it out; there is an A'-antecedent 'Joe', and an A-antecedent which is disjoint from 'Joe' and the sentence is still bad. The ungrammatical patterns are schematized in the following diagram. Although yi- is provided with a third person antecedent with which it may be* co-indexed, the first 46 person antecedent disallows any binding relationship. The pattern below yields an ungrammatical sentence. A •lstj yi-i A' 3rd; 4.1.2. Plural antecedents disallow anaphora Now compare the data in (6) and (7), which illustrate the difference between plural DP's and singular DP's as possible antecedent for yi. In (6a) the context provides a plural NP, 'at'eeke, 'girls', as well as a third singular DP, 'Mela'. The sentence in (6b) is ungrammatical with the given context in (6a). The data in (7) is familiar, both DP's are third person singular,'Mela', 'Dugi', and this sentence is good. Context: 6)a. at'eeke Mela yiyiiits^ Pl.girl Mela 3.3.Pfsee 'The girls saw Mela' b.* yilft' ya yizloh 3.poss.horse 3.for 3.3.Pf.rope 'They roped her horse for her' Context: 7)a. Mela Dugi yiyiiitsa Mela Dugi 3.3.Pf.see 'Mela saw Dugi' 47 b. yilij' ya yizloh 3.poss.horse 3.for 3.3.Pf.rope 'she roped his horse for him" 4.1.3. Yi- cannot have a first, second or plural antecedent We have seen how a first, second or plural antecedent cannot license yi-. Let us now turn to sentences where only a first, plural and second person antecedent are provided to (8i) (a), (b), and (c) respectively. These cases are bad. Note that in order to force a first or second person to be an antecedent to yi-, I have employed independent pronominal adjuncts, which lead the sentences to be somewhat odd in the first place. The sentences in (ii) are acceptable however, here the possessor agrees with the adjunct pronominal, in contrast to the cases in (i). 8)a. *(i)shL yilii ' seloh c. *(i) n L yW siniloh l.S. 3.poss.horse 1.3.Pf.rope 2.S. 3.poss.horse 2.3.Pf.rope ! L I roped his horse' 'You, you roped his horse' (ii) Shi, shilft' seloh (ii) ni, nilft' siniloh 'I, I roped my horse' 'You, you roped your horse' b.*(i)nihi, yiljj' siidloh l.du.Pl. 3.poss.horse l.du.P1.3.Pf.rope We, we roped his horse' (ii)nihi, nihU^' siidloh 'We, we roped our horse' 48 The data in (8 and 9) are uniformly bad, with or without context, showing thatyi- can only be licensed in a sentence containing two third persons2. This is schematized in the following diagram: *[lst/2nd.]k * 3rdPl; yi-i yi-i 4.1.4. Generalization Recall that^/- must have an antecedent that is in A' position. This antecedent can be provided by the discourse. The problem arises with the data in 4.1.1 ((3 and 4) (3) is repeated below: Here, two plausible antecedents are provided for yi- : a first and a third person, yet the sentence is still ungrammatical. In other words yi- is blocked from being bound. In 4.1.2 (6), a third plural and third person singular antecedent are provided for yi-, and the sentence is also bad. We have also seen in 4.1.3 (8) that a first, second or plural 2 The sentences in (9a,b) and (c) below are ungrammatical. (9a) involves a first person pronominal arguments, (b) involves a plural pronominal, and (c) involves a second person pronominal. Following the proposal for yi-, they can be ruled out for the same reason that (1) ia bad. Yi- is not bound, and yi- must be bound. No Context: (9)a. * yiljj' seloh c. * villi' siniloh 3.poss.horse l-3.Pf.rope 3.poss.horse 2-3.Pf.rope 'I roped his horse' 'You roped his horse' b. * yirji' siidloh 3.poss.horse ldu.Pl.-3.Pf.rope 'We roped his horse' (3 rep.) Hayit'eego Joe bika'eshwod, . . . * yilft' seloh 'How did I help Joe. . . . I roped his horse. 49 nominal cannot itself antecede yz'-. (9) shows that the only grammatical structures are where the A'-antecedent and the disjoint A-antecedent are both third person singular. This generalization is schematized below: 4.2. Solution 4.2.1 Sub-solution: Yi- is inherently 3rd person singular The ungrammaticality of (8) and (9) can be accounted for if we say thaty/'- is inherently third person singular. This means thaty/'- cannot have an antecedent which does not agree with it in phi-features. Data such as (7), which provide a 3rd person antecedent for yi- are grammatical because the antecedents provided agree withy/'-'s phi-features. Therefore the ungrammatical data can be ruled out due to agreement, i.e. yi- does not agree in person with a first or second person antecedent or in number with a plural antecedent. In (10), I repeat (8 a,b, and c) with phi-indices. A' 3rdj yz-i A 3rd; 10)a. *shi_ yilft' seloh *lst i_ yi- (=3) 1L, I roped his horse b. * nihi yil#' siidloh 'We. we roped his horse' c. * rvj_ yilij' siniloh 'You, you roped his horse' y\- (=3) 50 4.2.2. Summary of the Problem However, our claim that>7- is inherently specified for third person singular doesn't adequately explain why a first, second or plural disjoint antecedent should block anaphora with yi-; that is, why a nominal with different features blocks coindexation even when it is referentially distinct. Such is the case with the data in (3) and (4) and *lst/2nd/pluralj The following tree in (11) illustrates the ungrammatical sentence in (3). Even given the context in (a) 'How did I help Joe', the response in (b) 'I roped his horse' is not possible in Navajo with yi- as the possessor prefix. The antecedent provided is in an A'-position, just as required by the licensing condition given in Chapter 3. (ll)a. Hayit'eego Joe bika'eshwod b. *yilji'seloh (6). A 3rd; A 51 In contrast the tree in (12) from (5) illustrates that yi- when provided with two third person DPs as antecedents, is in a licit binding relationship. With the context 'Mela Dugi yiyiiltsq, 'Mela saw Dugi'. . . yilfi' ya yizloh', 'she roped his horse for him', is grammatical. To summarize all the relevant points, the data in (1) and (2) confirm our proposal that yi- must be A'-bound, and that the antecedent must be provided either overtly or in the discourse. This was discussed in detail in chapter 3, where yi is analyzed as a disjoint anaphor subject to both conditions A and C simultaneously. We have also seen thatj>/'- cannot take a plural or first/second person antecedent, but this can be accounted simply if we conclude that yi- is inherently 3rd person singular. Our theory accounts for the fact that yi- must be provided with a third person singular antecedent, (as in 5) and (7). Only in constructions where a non-coreferent first, second person or plural DP intervenes between^/- and a third person antecedent as in 52 (3), (4) and (6) is there a problem. To reiterate this point, the following pattern is not possible: (iii). A' *3rdi •John' A *lst/2ndj pluralj/u/j 4.2.3. Solution: Disjoint Antecedent Because yi- is inherently specified for third person singular phi-features, its A -antecedent must agree with it in phi-features. We have seen that the A-antecedent for yi- must also be third person singular. But crucially, let us now propose thatj/- must also agree with its (disjoint) A-antecedent. This immediately explains the ungrammatically of the pattern in (iii). Yi- must agree with both antecedents: it must be contra-indexed with its A-antecedent, and it must be coindexed with its A'-antecedent. We will represent this relationship formally by adopting the indexing schema proposed by Huang & Tang (H&T) (1991). H&T propose that anaphoric relationships be (iv). A 3rdi A 3rdj -4 53 represented using two different sets of indices, one for phi-features and one for coreference. They represent phi-feature by (phi-(i,j,k, ...)) and referential indices by (R-(1,2,3,...)). Thus, in a simple relation of coreference, the antecedent will have the same phi-feature indices and referential indices as the anaphor. (v) . John (phi-(i),R(2)) hurt his(phi-(i), R(2)) foot. H&T further suggest that the assignment of referential indices is dependent on the matching of phi-feature indices. We can adopt this idea simply for the Navajo case under consideration. Since yi- is inherently specified for third person singular, any potential antecedent, either disjoint or coreferent, must agree with it in phi-features in order for a referential dependency (either coreferent or disjoint) to hold. Thus both the coreferent A'-antecedent and the disjoint A-antecedent for>7- must be specified as third singular, as shown below. (vi) . A A yi-(phi- (i) CO- CO) (R (2) (not 3) (2)) (13) illustrates how this works. In the following sentence >7- agrees with both 'Mela' and 'Dugi' in phi-features. Yi- corefers with 'Dugi' its A'-antecedent and is disjoint in reference from 'Mela' its A-antecedent (13) Mela Dugi yiyiiitsa . . .yilft' ya yizloh Mela<phi-(i),R(2)) Dugi^i)^)) yiyiiitsa . . .yifoMiyWw' ya y i z l o h 'Mela saw Dugi. . . she roped his horse for him' 54 The next data in (14) (repeated from 3) illustrate a case where yi- does not agree with its A'-antecedent. The A-antecedent 'Joe' agrees with it in phi-features, yet this is insufficient since both the A-antecedent (first person pro) and the A'-antecedent must agree withy/-. (14) * Hayit'eego Joe bika'eshwod, . . . yilft' seloh Hayit'eego Joe (phi-(i), R(D bika'esh^hi^^wod, . . .yi(phi-<i),R(0)'2/»il'4' seloh 'How did J_ help Joe, . . . I roped his horse' 4.3 Conclusion In this chapter, I have shown that the analysis ofy/'- as a disjoint anaphor together with the independently supported assumption thaty/- is inherently specified as third person singular, account straightforwardly for the fact thaty/'- is only licensed when both its A'-antecedent and its (disjoint) A-antecedent are specified as third person singular. I have employed the dual indexing mechanism proposed by Huang & Tang (1991) to give a formal account of the phi-feature restriction on binding ofy/'-. It is important to point out that the behavior ofy/'- with first, second person and plural A-antecedents provides important supporting evidence for the disjoint anaphor analysis ofy/'-, since without this analysis it is quite mysterious as to why a non-coreferent DP would have an effect on yi- binding. 5 5 Chapter 5 Disjoint Anaphora in Dogrib: A Brief Comparison with Navajo 5.0. Introduction Dogrib and Navajo each have a set of cognate morphemes that developed from proto-Athapaskan *ye- and *we- (Thompson 1991). These are .ye- and we- in Dogrib andyi- and bi-in Navajo. As with Navajo .y/-, Dogrib ye- attaches to verbs and post-positions as well as nouns. In the following however, we will limit our comparison Xoyi-lye- on nouns. For reasons of comparison, I also include the pronominals bi- and its cognate we-. In Dogrib ye- has been analyzed as a disjoint anaphor (DA) (Saxonl984a, 1986,1995). This means that .ye-, just likej/-, must be in an A-binding relation with an antecedent from which it is disjoint in reference. Saxon has analyzed Dogrib we-, like Navajo bi-, as a pronoun subject to condition B. In the following data we compare the Dogrib DA .ye- to the Navajo DAyi-. The Dogrib data is from Saxon (1995) 5.1 Disjoint Anaphors (DA) as Non-Subject Possessors In the following data set, I compare ye- and yi- in the possessor position of non-subject DPs: Dogrib Navajo 1) a. Joe yeta gha ela whehtsjb. b.*Joe yizhee'i tsina'eei ya ayiila DA.father for canoe 3.S.Pf.make DAfather boat for 3.S.Pf.make 'Joej built a canoe for hisvj father' 'Joej built a canoe for hisi/j father' In Dogrib, the example is grammatical, in contrast to Navajo. As we have seen in chapter 3, 56 without context possessory/'- in Navajo is bad. If context is provided, as in (2) then the sentence in (lb) becomes grammatical: 2) Frank Joe yizhee'i tsina'eei ya ayiila Frank JoeDA.father boat for 3.S.Pf.make 'Frankf built a canoe for Joe' Sj father' It thus appears that in contrast to Navajo, Dogribye- does not need an A'-antecedent. However, Saxon (p.c.) observes that the two languages may be closer than the data above seem to indicate. Further, investigation is clearly necessary. 5.2. Subject possessor In the following examples, possessory/'- and ye- are shown attached to subject DPs. As expected, the result is ungrammatical in both languages. This follows from the analysis of bothy/'- and ye- as disjoint anaphors, since there will be no A-antecedent available for a possessor in subject position. 3) a. *Yetsee edaizhe b. Yichei 'adaadzodli 3.grandfather 3.S.Imp.be.clever 3.grandfather 3.S.Imp.be confident 'Her grandfather is clever' v 'Her grandfather is boastful' 5.3. Summary In (4), I summarize the comparison between Dogribye-, and Navajo y/'-. 4) Dogrib Navajo *ye-NP *y/'-NP (subj possessor) NP ye-NP *NP y/'-NP (non-subj poss) As (4) shows the two DAs behave alike in subject position, but differ in non-subject positions. 57 This will follow if we assume thaty/- and ye- both need an A-antecedent from which they must be disjoint in reference, but only Navajo yi- also acts as an A'-anaphor which must have a coreferent antecedent in an A'-position, possible supplied by prior disourse. This is summarized in the table in (5): 5) Dogrib Navajo Disjoint anaphor (Contra-indexed A-antecedent) yes yes A '-anaphor (obligatorily coindexed A'-antecedent) no yes 58 Chapter 6 Conclusion 6.0. Summary In chapter one I have introduced^/'- as a possessor prefix that attaches to noun bases to refer to the possessor. In chapter 21 have provided a brief summary of prior analyses of the yi-/bi-altenation. Most of this work deals with verbal^/'- andyi- on post-positions. My main proposal that_v/"- is a disjoint anaphor is contained in chapter 3. I have shown that yi- must have an A'-antecedent with which it is coreferent and an A-antecedent from which it is disjoint. I have accounted for these generalizations by claiming that yi- is subject to both conditions A and C of Generalized Binding Theory (Aoun 1985). In Chapter 4,1 have shown that both the A'-antecedent with which it is coindexed, and the A-antecedent with which it is contra-indexed must agree with yi- in phi-features. Chapter 5 contains a brief comparison of yi- with Dogrib ye-: both Navajo yi- and Dogrib ye- are disjoint anaphors (DA), and both require an antecedent from which they must be disjoint. Importantly, Navajo >v- also requires an A'-antecedent; further cross-linguistic investigation is necessary to establish whether this is also the case for Dogrib ye-. 6.1. Implications for future research The obvious question which results from this research is whether the analysis proposed here for possessive J 7 - can be extended to yi- on verbs and/or post-positions. 59 Though this question is well beyond the scope of this thesis, I will sketch out here the outlines of such an analysis and point out some of the ways nominal yi- differs in its behavior from verbal and post-position yi-. Let us begin by making the null hypothesis that the analysis given here can be extended without modification to verbal and post-position yi-. This will imply thaty;'- must always have a disjoint A-antecedent and a coreferent A'-antecedent. Is this true for verbs and post-postions? It is the case thaty/'- on the verb must have 2 antecedents. (1) 'asdzaan 'awee' yjdeelchid woman baby 3-3.touch(with hands) 'The woman touched the baby' (2) 'awee' yjdeelchid baby 3-3.touch(with hands) 'S/he touched the baby' Verbal yi- refers to the object. Is there an A-antecedent from which it is disjoint? Yes, ify/'-refers to the object, then it will be disjoint from the subject; otherwise, it would be a reflexive morpheme. Does it have ah antecedent with which it is coindexed? Yes, the object itself, i.e. 'baby'. However note that the object DP is in A-position. It thus appears that the A ' -antecedent that we need for possessive yi- is replaced by an A-antecedent for verbal yi-. This tentative analysis is summarize in the table below. 60 Verb vi- Possessor yi-Disjoint antecedent yes yes DA in A-postion yes yes Coindexed yes yes Coreferent antecedent NO yes Coreferent antecedent yes NO 61 References Aoun, J. (1985). A Grammar of Anaphora. MIT, Cambridge MA. Frishberg, N . (1972). Navajo and the Great Chain of Being, in Syntax and Semantics, Vol.1, New York Academic Press. Haile, Fr.B. (1941) Learning Navajo, Vols. I&II, St. Michaels, Arizona: St. Michaels Press. Hale, K. (1968). Problems in Navajo Grammar, manuscript, Cambridge: MIT Hale, K. (1973). A Note on Subject Object Inversion in Navajo, In B. Kachru et al., Issues in Linguistics, Urbana, 111. University of Illinois Press. Huang, C-T.J. (1984). On the Distribution and Reference of Empty Pronouns. Linguistic Inquiry. V15:N4. 531-575, MIT. and C.-C.Jane Tang (1991) The local nature of the long-distance reflexive in Chinese, in J. Koster, & E. Reuland, eds. Heim, I., Lasnik, H., Robert May. (1991). Reciprocity and Plurality. Linguistic Inquiry. V22:N1, MIT. Jelinek, Eloise. (1984) Empty Categories, Case, and Configurationality. Natural Language and Linguistic Theory 2:1, 39-76. Perkins, Ellavina Tsosie. (1978) The Role of Word Order and Scope in the Interpretation of Navajo Sentences. Ph.D. Dissertation. University of Arizona. Platero,P. (1974) The Navajo Relative Clause, UAL 40:202-246. (1982). Missing noun phrases and grammatical relations in Navajo. International Journal of American Linguistics 48(3):286-305. Reichard, G. (1951) Navajo Grammar, Publication of the American Ethnological Society, 21. Salego, 1. (1977). The Heirarchical Classification of Nouns in Navjo, manuscript, San Diegeo . University of California. Saxon, L. (1984a). Disjoint anaphora and the binding theory. In Proceedings of the West Coast Conference on Formal Linguistics, Volume 3, 242-251. Saxon, L. (1986). The Syntax of Pronouns in Dogrib. PhD. Dissertation. University of California San Diego. Saxon, L. (1995). Complex pronominal, disjoint anaphora, and indexing. LSA, New Orleans. Speas, M . (1990). Phrase Structure in Natural Language. University of Massachusetts, Amherst: Kluwer Academic Publishers. and E.Parsons Yazzie. (1996). Quantifiers and the Position of Noun Phrases in Navajo. Jelinek, Midgette, Rice and Saxon, eds, Athabaskan Language Studies, University of New Mexico. 62 (1997). From Rules to Principles in the Study of Navajo Syntax. Navajo Language Academy Workshop on Syntax and Semantics. Wetherspoon, G. (1977). Language and Art in the Navajo Universe, Ann Arbor. University of Michigan Press. Willie, M.A. 1989. Why there is nothing missing in Navajo relative clauses. In Cook and Rice (Eds.), Athapaskan Linguistics: Current Perspectives on a Language Family, 407-437. Mouton DeGruyter. (1991). Pronouns and Obviation in Navajo. Tucson: University of Arizona Dissertation. Young, R , and W. Morgan. 1987. The Navajo Language: A Grammar and Colloquial Dictionary. Albuquerque, New Mexico: University of New Mexico Press. , 1992. Analytical Lexicon of Navajo. Albuquerque, New Mexico: University of New Mexico Press. 63 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items