UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Nsyilxcen modality : semantic analysis of epistemic modality Menzies, Stacey 2013

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

Item Metadata

Download

Media
24-ubc_2013_spring_menzies_stacey.pdf [ 409.03kB ]
Metadata
JSON: 24-1.0073500.json
JSON-LD: 24-1.0073500-ld.json
RDF/XML (Pretty): 24-1.0073500-rdf.xml
RDF/JSON: 24-1.0073500-rdf.json
Turtle: 24-1.0073500-turtle.txt
N-Triples: 24-1.0073500-rdf-ntriples.txt
Original Record: 24-1.0073500-source.json
Full Text
24-1.0073500-fulltext.txt
Citation
24-1.0073500.ris

Full Text

Nsyilxcen Modality Semantic analysis of epistemic modality by Stacey Menzies  B.A., The University of British Columbia, 2010  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in The Faculty of Graduate Studies (Linguistics)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) January, 2013 c Stacey Menzies 2013  Abstract The aim of this thesis is to describe and analyze the modal system of Nsyilxcen, an Interior Salish language spoken in south central British Columbia and northern Washington State. In particular, it focuses on the epistemic modals mat and cmay, which express necessity and possibility with respect to certain bodies of knowledge. Similar to modals in St’´at’imcets (Rullmann et al. 2008) and Gitksan (Peterson 2010) these modals lexically encode an epistemic modal base and an indirect inferential evidential restriction. I propose that these two modals can be distinguished based on their modal force distinction, where mat has variable modal force and cmay a strictly encoded existential modal force. Based on these generalizations, I propose a formal semantic analysis for the epistemic modals drawing from Kratzer (1977, 1981, 1991, 2012), Rullmann et al. (2008), Peterson (2010), and Deal (2011). The analysis defines each modal in a way that accounts for the strictly encoded modal base and evidential restriction, as well as the variable modal force for mat and the strictly encoded existential modal force for cmay. In addition to the epistemic modals mat and cmay this thesis documents the reportative modal kw ukw as well as how Nsyilxcen encodes non-epistemic modality. It looks at the bouletic modal cakw and how Nsyilxcen encodes a deontic, circumstantial, ability, and teleological modal base. This thesis will discuss how the Nsyilxcen system fits into a preliminary modal typology based on the semantics of these modals.  ii  Table of Contents Abstract  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ii  Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  iii  List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  v  List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  vi  List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . .  vii  Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  viii  1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 A guide to the chapters . . . . . . . . . . . . . . . . . . . . . . . 1.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Nsyilxcen language background . . . . . . . . . . . . . . . . . . 1.4 Nsyilxcen syntax . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.1 Syntax of epistemic modals in the pre-predicative domain 1.4.2 Syntax of pre-predicative particles and epistemic modals .  1 4 4 5 7 9 13  2  Nsyilxcen Modality . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Colville-Okanagan modality: A. Mattina 1973 . . . . . . . 2.2 Nsyilxcen epistemic modality: The Douglas Lake dialect . . 2.2.1 Source of evidence and Nsyilxcen epistemic modals 2.3 Nsyilxcen non-epistemic modality . . . . . . . . . . . . . . 2.3.1 cakw bouletic modal . . . . . . . . . . . . . . . . . 2.3.2 -ks irrealis . . . . . . . . . . . . . . . . . . . . . . 2.3.3 Further ways to express non-epistemic modality . . 2.3.4 Epistemic modals in non-epistemic contexts . . . . 2.4 Chapter summary . . . . . . . . . . . . . . . . . . . . . . .  17 18 19 26 29 30 32 36 38 39  . . . . . . . . . .  . . . . . . . . . .  . . . . . . . . . .  iii  3  Theoretical Background . . . . . 3.1 Kratzerian theory of modality 3.2 Modality and evidentiality . . 3.3 Cross-linguistic variation . . .  . . . .  . . . .  . . . .  . . . .  . . . .  41 41 44 45  4  Epistemic Modals: Comparative Semantic Analyses . . . . 4.1 Comparative models . . . . . . . . . . . . . . . . . . . . 4.1.1 St’´at’imcets modal analysis: Rullmann et al. 2008 4.1.2 Gitksan modal analysis: Peterson 2010 . . . . . . 4.1.3 Nez Perce: modals without scales . . . . . . . . .  . . . . .  . . . . .  . . . . .  . . . . .  53 53 54 58 63  5  Formal Semantic Analysis for the Nsyilxcen Epistemic Modals 5.1 Option #1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Option #2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Option #3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 Implications and further research . . . . . . . . . . . . . .  . . . . .  . . . . .  . . . . .  68 68 72 76 79  6  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  83  7  References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  85  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  iv  List of Tables 1.1 1.2 1.3 1.4  Salish language family (Kroeber 1999: 4) . . . . . . . Independent clause particles (Mattina 1973: 126) . . . Order of Nsyilxcen modals . . . . . . . . . . . . . . . Order of epistemic modals and pre-predicative particles  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  6 10 13 16  2.1 2.2 2.3 2.4  Nsyilxcen modality . . . . . . . . . . . . . . Modal particles (A. Mattina 1973) . . . . . . Nsyilxcen epistemic modals . . . . . . . . . Nsyilxcen particles in non-epistemic contexts  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  17 18 19 29  3.1 3.2 3.3 3.4 3.5 3.6 3.7  Cross-linguistic variation: conversational background Cross-linguistic variation: evidence source . . . . . . Cross-linguistic variation: modal force . . . . . . . . Javanese modality . . . . . . . . . . . . . . . . . . . Blackfoot modality . . . . . . . . . . . . . . . . . . Cross-linguistic modal typology 1 . . . . . . . . . . Cross-linguistic modal typology 2 . . . . . . . . . .  . . . . . . .  . . . . . . .  . . . . . . .  . . . . . . .  . . . . . . .  . . . . . . .  . . . . . . .  47 49 50 50 51 51 52  4.1 4.2  Cross-linguistic modal typology 3 . . . . . . . . . . . . . . . . . Three types of modal systems (Deal 2010: 34) . . . . . . . . . . .  53 66  5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9  Kratzer gradable modality (1981) . . . . . . . . . . . . Nsyilxcen particles in circumstantial contexts . . . . . Necessity and weak necessity . . . . . . . . . . . . . . Possibility and strong possibility . . . . . . . . . . . . Cross-linguistic modal typology . . . . . . . . . . . . St’´at’imcets modal system (Matthewson in press [2]: 9) Gitksan modal system (Matthewson in press [2]: 9) . . Nsyilxcen modal system . . . . . . . . . . . . . . . . English modal system . . . . . . . . . . . . . . . . . .  70 70 73 75 80 81 81 81 82  . . . .  . . . .  . . . .  . . . . . . . . .  . . . . . . . . .  . . . . . . . . .  . . . . . . . . .  . . . . . . . . .  . . . . . . . . .  v  List of Figures 1.1  Salish language family map (Kroeber 1999: xxxi) . . . . . . . . .  5  3.1  Types of evidence (Willett 1988) . . . . . . . . . . . . . . . . . .  44  4.1 4.2 4.3 4.4  Strong universal quantification . Weak universal quantification . . Weak existential quantification . Strong existential quantification  56 56 57 57  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  . . . .  vi  List of Abbreviations abs = absolutive  incep = inceptive  aff = affirmative  irr = irrealis  asp = aspect  lex = lexical particle  boul.mod =bouletic modal  loc =locative  caus = causative  mid = middle  ctr = control  mod = modal  cont = continuous verbal prefix  neg = negation  con = continuative verbal prefix  nom =nominalizer  cust = customary  obl =oblique  comp = complementizer  part =particle  dem = demonstrative  pl =plural  det = determiner  poss = possessive  der = derivative  q =question particle  dir = directive transitivizer  red =reduplication  emph = emphatic  res = resultive  epis.mod = epistemic modal  sg =singular  erg = ergative  stat = stative  fut = future  trans =transitivizer  imp = imperative  ynq = Yes/No question particle vii  Acknowledgements There are so many people to thank for helping me complete this thesis. First off, my 2010 MA cohort in alphabetical order: Sihwei Chen, James Crippen, Ella FundReznicek, Tianhan Liu, Joash Johannes, Michael McAuliffe, Kevin McMullin, Lauren Quinn, and Inga Schniske. If it wasn’t for your positivity, smiles, and support everyday, I would not have made it through this program. Thank you as well to the Modality 530 Seminar class for the interesting and insightful discussions about modality. Thank you to Vera Hohaus, Patrick Littell, Meagan Louie, Claire Turner, and Amelia Reis Silva who took time outside of class to discuss my work with me. Next, I want to thank all of the professors at the University of British Columbia who taught me during my undergrad and graduate programs. I would especially like to thank my committee members; Hotze Rullmann who sat with me and repeatedly explained difficult semantic concepts until I understood them. Henry Davis whose dedication to the First Nations languages of the Pacific Northwest is incredibly inspiring. Finally, to my supervisor Lisa Matthewson whose endless patience and encouragement kept me going through an incredibly tough program, you are a huge inspiration. You all expected excellence and helped me achieve it, thank you. A big thank you also to fellow Okanaganists John Lyon and Joel Dunham. Thank you for sharing the long drives to Merritt, for helping me with elicitations, and for always being there to talk about the Okanagan language with me. A very special thank you to the Douglas Lake consultants Lottie Lindlay, Sarah McLeod, and Nancy Saddleman for sharing your language with me. Thank you for being patient with me as I tried to speak your beautiful language. Thank you also to Kristine Jack whose dedication to teaching the Okanagan language kept me inspired to keep working on Okanagan. Finally, I want to thank my family. To my mom and dad, thank you for always being supportive and encouraging, and for always believing in my ability to do this. To my siblings Jen and Brian who have led the way and have always had good advice. To my Grandma Lorna who knew I could do this even when I didn’t think I could. Finally to my best friends Salish Fenwick-Wilson and Tessa Hamilton thank you for the conversations that always lifted my spirits. viii  Chapter 1  Introduction This thesis is an investigation into the modal system of the Nsyilxcen Salish language. Modality deals with the expression of possibilities and necessities. Modalized sentences allow speakers to talk about possible alternatives to the way the world actually is at the utterance time relative to certain bodies of knowledge, or facts (Kratzer 1977, 1981, 1991, 2012; von Fintel and Heim 2011). The goals of this investigation are summarized in (1): (1) i. To provide a detailed description of the Nsyilxcen modal system in its entirety. ii. To provide a formal semantic analysis for the inferential epistemic modals of this system that adequately explains the meanings of epistemic modality in Nsyilxcen. iii. To contribute to the documentation of the Nsyilxcen language. This thesis takes Kratzer’s theory of modality as the theoretical basis for exploring the Nsyilxcen modal system. Opinions vary as to how the modal space is divided, but there is general agreement that a basic distinction is drawn between epistemic and non-epistemic modality. Epistemic modality expresses possibility and necessity with respect to the speaker’s knowledge and beliefs relative to the situation, and with respect to the evidence available to the speaker at the time of utterance. For example, consider the following English example provided by Kratzer (1991), where a police inspector is investigating a crime scene. Given the evidence available to him/her the detective can draw conclusions as to who the murderer is. The more evidence available to the detective the stronger the statement he/she can make (Kratzer 1991). (2)  a. Michl must be the murderer b. Michl is probably the murderer 1  c. There is a good possibility that Michl is the murderer d. Michl might be the murderer e. There is a slight possibility that Michl is the murderer The claim is that all languages are able to express epistemic possibility and necessity and the grades of strength in between. Consider the Nsyilxcen example (3) where the speaker is making a statement about Mary’s whereabouts given what he/she knows about Mary and her schedule. In this possibility context the speaker’s knowledge is not conclusive and they are uncertain whether Mary is running, or if she is shopping or at school. Nsyilxcen has two modals that can be used to express the speaker’s high level of uncertainty in this context. (3)  Context: You know that Mary loves to go running and often goes on runs randomly. However, she could also be at the store or at school. I ask you, where is Mary? You reply: a. Mary mat ac-q´ıc-lx Mary epis.mod cust-run-lex ‘Mary might be running’ b. Mary cmay ac-q´ıc-lx Mary epis.mod cust-run-lex ‘Mary might be running’  Example (4) provides a necessity context where there is more evidence available to the speaker, leading the speaker to be more certain that Mary is running. In this context, the modal cmay is no longer permitted. (4)  Context: Mary runs every day to train for a marathon. She usually runs at 6pm on Tuesdays. Today is Tuesday and it’s 6pm. I ask you, where is Mary? a. Mary mat ac-q´ıc-lx Mary epis.mod cust-run-lex ’Mary must be running’ b. # Mary cmay ac-q´ıc-lx Mary epis.mod cust-run-lex ’Mary might be running’ 2  In addition to necessity and possibility with respect to a speaker’s knowledge and beliefs, recent literature has argued that epistemic modals also specify the source of evidence upon which the modal statement has been based (Izvorski 1997, von Fintel and Gillies 2010, Matthewson 2011, Kratzer 2012). Nsyilxcen makes a distinction between epistemic modality based on indirect inferential evidence and evidence from reports. The reportative kw ukw is used to express that the source of evidence is from a report. (5)  Context: My friend tells me that our friend Jill got married over the weekend. I go to my parents’ house. My parents know Jill. I tell them kw ukw Jill c-mrim rep jill cont-get.married ‘[someone told me] Jill got married’  The conclusions based on these generalizations will show that Nsyilxcen is similar to other Pacific Northwest languages like St’´at’imcets and Gitksan in that each modal is strictly assigned one type of modality, epistemic or non-epistemic. Furthermore, like the epistemic modals in these languages mat shows variable modal strength. It is permitted in contexts that are compatible with both necessity and possibility interpretations. These modal systems are different from the English system where the modals must and might are strictly assigned a modal strength but are flexible with respect to the type of modality. Where the Nsyilxcen modal system differs from the St’´at’imcets and Gitksan systems is with respect to the modal strength of the second epistemic modal cmay. cmay is not only strictly an epistemic modal like mat, it is also strictly a possibility modal. Unlike mat, it does not have variable modal strength. This thesis will show that Nsyilxcen is not directly similar to any other modal system discussed in this thesis. This is due to how the two epistemic modals behave with respect to their modal strength. mat is similar to the St’´at’imcets and Gitksan epistemic modals as it is strictly epistemic and shows variable modal strength. However, cmay is restricted to contexts that are compatible with possibility readings only; a subset of contexts where mat is permitted. These findings call into question the idea that modal systems can be directly compared as a whole across languages based on a simple dichotomy between conversational background and modal force. Finally, in order to fully document the Nsyilxcen modal system, data for nonepistemic modality is presented in this thesis. Non-epistemic modality covers a  3  number of modal readings that express possibility and necessity relative to different kinds of facts. These include deontic modality, expressing possibility and necessity relative to certain laws; circumstantial modality, relative to certain circumstances; ability, relative to the abilities of people or entities; teleological, relative to achieving a particular goal; and bouletic, relative to the desires of an individual. Nsyilxcen has one specialized non-epistemic modal cakw that expresses bouletic modality. As there are no other specialized non-epistemic modals, the speakers paraphrase to express all types of non-epistemic modality by making use of the imperative, irrealis, and the simple bare predicate.  1.1  A guide to the chapters  Chapter 1- The remainder of chapter 1 provides background information about the Nsyilxcen language, including a section discussing the aspects of the syntax relevant to the Nsyilxcen epistemic modals. Chapter 2 - Chapter 2 focuses on providing a descriptive overview of the Nsyilxcen modal system. This provides the empirical base for a formal semantic analysis, but also focuses on providing a non-theoretical description that would be useful for non-linguists. Chapter 3 - Chapter 3 provides an overview of the theory of modality and evidentiality that provided the theoretical tools used to elicit the Nsyilxcen modals. This theory was also used to provide a formal analysis for the modals in Nsyilxcen. Chapter 4 - Chapter 4 provides the formal semantic analyses for the St’´at’imcets, Gitksan, and Nez Perce modal systems which provide a cross-linguistic comparison and base from which the Nsyilxcen system could potentially be analyzed. Chapter 5 - Chapter 5 proposes three potential formal analyses for the Nsyilxcen epistemic modals. Each option discusses how to account for the strictly encoded modal type and evidential restrictions of mat and cmay as well as the variable modal strength of mat and strictly encoded modal strength of cmay. This thesis explores both epistemic and non-epistemic modality in Nsyilxcen, with particlar focus on developing a formal semantic analysis for the epistemic modals mat and cmay.  1.2  Methodology  The methodology of elicitation involved the collection of primary data through one-on-one interviews with fluent Nsyilxcen consultants. This semantic research 4  crucially required eliciting felicity judgments from the consultants. Following Matthewson (2004) the task first required collecting complete Nsyilxcen sentences through a translation task from English to Nsyilxcen. These sentences were then paired with a context and the consultants were asked to judge the appropriateness of the sentence within those contexts. Multiple follow up elicitation sessions were conducted in order to discuss how the sentence can be interpreted, why it is or is not correct, how acceptable the utterance is within the presented context, and how to correct the context or the utterance so that it is felicitous. This was done through direct questioning, storyboards, and pictures. Semantic research of this type is of particular importance as these judgments cannot be extracted from texts and must, therefore, be done directly with the speakers.  1.3  Nsyilxcen language background  The Nsyilxcen language, also known as Okanagan, is spoken in south central British Columbia extending from Enderby BC southward through the Okanagan to Kelowna, Penticton and Osoyoos and into north central Washington State in Omak (Lyon 2009). The empirical data for this thesis is based on the Douglas Lake dialect spoken on the Upper Nicola reserve in Merritt, BC (Lyon 2009).  Figure 1.1: Salish language family map (Kroeber 1999: xxxi) 5  Nsyilxcen is part of the Southern Interior branch of the Salish language family which is comprised of four languages; Okanagan, Moses-Columbian, Coeur d’Alene and the Spokane-Kalispel-Flathead dialect continuum (Lyon 2009). Table 1.1 displays the historical division of the Interior Salish languages (Kroeber 1999). Table 1.1: Salish language family (Kroeber 1999: 4) Interior Salish Division Northern St’´at’imcets Thompson Shuswap Southern Moses-Columbian Colville-Okanagan Coeur d’Alene Spokane-Kalispel-Flathead  The Ethnologue reports that there are approximately 400 speakers of the Nsyilxcen language, of which there are currently 10 fluent speakers in the Upper Nicola dialect area (Lyon 2011, Lewis 2009). This thesis is based on data from three speakers living in Merritt, BC. Prior research on the Nsyilxcen language has come primarily from Anthony Mattina, who produced numerous research papers (19732004), a dictionary for the Colville-Okanagan language (Mattina 1973), and a grammatical sketch of Colville-Okanagan as spoken in eastern Washington (1987). Anthony Mattina also worked in cooperation with Peter Seymour (1974, 1985) documenting narratives, as well as with Clara Jack (1982, 1986, 1992), and Sarah Peterson (1997). Substantial work has also been conducted by Donald Watkins (1970), and Yvonne Hebert (1982), Nancy Mattina (1993-1999), and Maxine Baptiste (2002). Sarah Peterson (2005) is also responsible for producing Okanagan language learning materials. More recently work has been produced by John Lyon and Joel Dunham, PhD students at the University of British Columbia. A. Mattina (1973) recorded and glossed the Nsyilxcen modals in the Colville grammar; however this thesis is the first attempt to provide a semantic analysis for the Nsyilxcen modal system based on the standard semantic modal theory.  6  1.4  Nsyilxcen syntax  The following section will introduce the aspects of Nsyilxcen syntax that are relevant to the study of the modals discussed in this thesis. It will focus specifically on where the modal particles are found relative to the subject and predicate as well as where the modal particles are found relative to each other in the pre-predicative domain. To begin, Nsyilxcen allows for flexible word order. N. Mattina (1996) and Lyon (2010, 2012) discuss the permitted word orders for the Nsyilxcen language. Intransitive predicates with overt DP subjects show VS and SV word orders as seen in (6a) and (6b). (6)  a. xw -xw´ıs-t iP tkłm´ılxw red-walk-stat det woman ‘The woman started walking’ (Lyon 2010) b. iP tkłm´ılxw xw -xw´ıs-t det woman red-walk-stat ‘The woman started walking’ (Lyon 2010)  For transitive predicates with overt DP subjects and DP objects both SVO (7) and VSO (8) are accepted word orders. (7)  iP s´amaP wik-s iP x´ıxwt@m det white.person see-(dir)-3.sg.erg det little.girl ‘The white person saw the little girl. / * The little girl saw the white person’ (Lyon 2010)  (8)  wik-s iP x´ıxwt@m iP s´amaP see-(dir)-3.sg.erg det little.girl det white.person ‘The little girl saw the white person. / * The white person saw the little girl’ (Lyon 2010)  Lyon (2012) provides an in-depth analysis of non-verbal predicates in Okanagan. He argues that pre-predicative particles structurally must precede the predicate core. These include the modals that will be discussed in this thesis as well as question particles, future markers, and negation. (9) shows that the subject can precede, follow, or appear in between the particle and main predicate string, however the particle must precede the main predicate (Lyon 2012:3). 7  (9) a. cmay John [IJcqw aqw ] mod John cry ‘John might cry’ (Lyon 2012) b. cmay [IJcqw aqw ] John mod cry John ‘John might cry’ (Lyon 2012) c. John cmay [IJcqw aqw ] John mod cry ‘John might cry’ (Lyon 2012) d. * [IJcqw aqw ] cmay John cry mod John ‘John might cry’ (Lyon:2012) I assume that the pre-predicative particles in Nsyilxcen, including the modals, lie in functional projections. Functional projections are the syntactic projections of functional heads like determiners (DP), tense (TP), and aspect (aspP). Cinque (1999) proposes a universal hierarchy of 30 different functional projections for different mood, modal, and aspectual heads. Looking specifically at the modal projections, according to Cinque (1999) the functional projection for necessity modality is structurally higher than that for possibility modality. This appears to coincide with the Nsyilxcen data that will be presented in section (1.4.1), however, more research is necessary to determine whether the other functional projections in Nsyilxcen fit within this functional hierarchy. (10) [.... [Modnecessity Necessarily [Modpossibility Possibly ...]]] The Nsyilxcen modals listed in (11a-c) are found in the pre-predicative domain of the clause. However, the reportative modal kw ukw in (11d) shows different syntactic behaviour. Section 1.4.1 will discuss how these modals interact with each other in the pre-predicative domain. 8  (11) a. mat - epistemic modal b. cmay - epistemic modal c. cakw - bouletic modal d. kw ukw - reportative Afterwards, section 1.4.2 will discuss the word order of the modals relative to other pre-predicative particles listed in (12). (12) a. ha - question marker b. Puc - question marker c. lut - negation  1.4.1  Syntax of epistemic modals in the pre-predicative domain  A. Mattina (1973) describes the pre-predicative particles in table 3.1 as clause particles that modify the clause in which they are found. These are considered independent clause particles which introduce clauses that can stand alone as independent sentences (Mattina 1973: 125). Mattina charts the independent particles in a way that he states reflects the normal order of occurrence of each particle relative to each other. The following table gives Mattina’s ordered list of these particles as he presents and translates them (1973: 126). 1 1  Mattina (1973) does not state whether particles that share the same column of table 3.1 are mutually exclusive or freely ordered. Recent elicitations reveal that the modals mat (//mt//), and cmay ´ are not mutually exclusive and are ordered when they occur together in the same sentence. (//cm//) More research is necessary to determine the relative order of the other particles in each column.  9  Table 1.2: Independent clause particles (Mattina 1973: 126) 1 h´aP Simple Interrogative  2 naˇx@mł Constrastive  3 Pał´ı P Clausal  4 ´ cm Probable  P´uc Dubitative Interrogative  ´txw Evidential  cakw Conditional  mt Conjectural  Puł additive  mi Future nixw Additional  My elicitation on the Douglas Lake dialect of Nsyilxcen has revealed the following ordering of the modals. To begin, (13) shows the permitted word order and per´ mitted stacking of the epistemic modals mat (//mt//) and cmay (//cm//). (13c) and (13d) show that both epistemic modals can be stacked together in a specific order. (13e) shows that these particles cannot be placed after the predicate. 2 (13) IJ a. mat John sc-kr-mix mod John asp-swim-imp ‘John must be out swimming/ Maybe John is swimming’ IJ b. cmay John sc-kr-mix mod John asp-swim-imp ‘John might be swimming’ IJ c. mat cmay John sc-kr-mix mod mod John asp-swim-imp ‘Maybe John is swimming’ Mattina’s (1973) transcriptions write //cm// ´ as c@m and //mt// as m@t and he states that schwa epenthesis is predictable. After listening to my primary data, I have decided to transcribe these modals with full vowels. Also, cmay was never reduced to c@m ´ during my elicitations. A closer phonetic analysis will be necessary to determine which analysis is more accurate. 2  10  IJ d. * cmay mat John sc-kr-mix mod mod John asp-swim-imp ‘Maybe John is swimming’ IJ e. * John sc-kr-mix mat/cmay John asp-swim-imp mod/mod ‘Maybe John is swimming’ I do not have data where the predicate precedes the modal and subject , however I predict that it will be incorrect given the data presented in (9d). (14)  IJ * sc-kr-mix mat/cmay John asp-swim-imp mod/mod John  (Not elicited)  ‘Maybe John is swimming’ Next, the bouletic modal cakw can only occur in the pre-predicative sphere before the predicate core as seen in (15a) and (15b). Also, cakw cannot co-occur with the epistemic modals mat or cmay as seen in (15c-d). Sentences presented to the consultants where modals are stacked with cakw were often corrected to contain only one of the modals depending on the context. (15) a. cakw John kw u-łw´ın-(n)t-m boul.mod John 1.pl.abs-leave-dir-3.sg.erg ‘John should leave us’ b. * John kw u-łw´ın-(n)t-m cakw John 1.pl.abs-leave-dir-3.sg.erg boul.mod ‘John should leave us’ c. * cakw mat/cmay John kw u-łw´ın-(n)t-m boul.mod mod/mod John 1.pl.abs-leave-dir-3.sg.erg ‘John should leave us’ d. * mat/cmay cakw John kIJw u´ l’-m t pwmin mod/mod boul.mod John make-mid obl drum ‘John should be making drums’  11  Next is the reportative kw ukw . There is mixed data with respect to the ordering of the epistemic modals relative to the reportative. Data from one consultant in (16) shows that the reportative cannot co-occur with the epistemic modals. However, there is no data from this consultant where mat or cmay precedes kw ukw . (16) a. kw ukw ks-kIJw u´ l-aPx t pwmin rep irr-make-incep obl drum ‘I heard she is going to make a drum’ t pwmin b. * kw ukw cmay ks-kIJw u´ l-aPx rep mod irr-make-incep obl drum ‘I heard she might make a drum’ c. * kw ukw mat ks-kIJw u´ l-aPx t pwmin rep mod irr-make-incep obl drum ‘I heard she might make a drum’ Data from a second consultant in (17) reveals that the epistemic modals and the reportative kw ukw can be stacked together in the same sentence. (17) IJ England a. kw ukw cmay kw łwin-(n)t-m kl rep mod go-dir-3.sg.erg loc England ‘I heard he might go to England’ IJ England b. cmay kw ukw kw łwin-(n)t-em kl mod rep go-dir-3.sg.erg loc England ‘I heard he might go to England’ IJ England c. kw ukw mat kw łwin-(n)t-em kl rep mod go-dir-3.sg.erg loc England ‘I heard he might go to England’ IJ England d. mat kw ukw kw łwin-(n)t-em kl mod rep go-dir-3.sg.erg loc England ‘I heard he might go to England’ 12  (18) shows that the reportative has different syntactic behaviour from the epistemic and bouletic modals as it can occur after the predicate core. (18) a. John kw ukw c’aP-nt-´ıs iP k@kw´ap John rep hit-trans-3.sg det dog ‘I heard John hit the dog’ b. John c’aP-nt-´ıs kw ukw iP k@kw´ap John hit-trans-3.sg rep det dog ‘I heard John hit the dog’ c. * John c’aP-nt-´ıs mat iP k@kw´ap John hit-trans-3.sg mod det dog ‘John must have hit the dog’ As this data shows, mat and cmay have a fixed ordering where mat must always precede cmay. Recall from section (1.4) that this ordering fits with Cinque’s predicted ordering for necessity and possibility modals. Secondly, mat and cmay are mutually exclusive with respect to the bouletic modal cakw . Finally, according to one speaker, the reportative can be stacked with the epistemic modals and can preceded and follow both mat and cmay. Table 1.3 summarizes these generalizations. Items in the table that share a column are mutually exclusive.  kw ukw  1.4.2  Table 1.3: Order of Nsyilxcen modals mat cmay kw ukw cakw  Syntax of pre-predicative particles and epistemic modals  Next, I looked at how the epistemic modals stack relative to other pre-predicative particles listed in (19). (19) a. Puc - question marker b. ha - question marker 13  c. lut - negation To begin, the question particle ha has been labeled as the interrogative particle (Mattina 1973) and the yes/no question particle (Lyon 2012). (20) shows that the question particle belongs to the pre-predicative domain as it also cannot be preceded by the predicate (Lyon 2012:26). (20) a. ha [yamˇxw aP ynq basket  NP ]  ixiP dem  ‘Is that a basket?’(Lyon 2012: 26) b. ha ixiP [yamˇxw aP ynq dem basket  NP ]  ‘Is that a basket?’(Lyon 2012: 26) c. ixiP ha [yamˇxw aP dem ynq basket  NP ]  ‘Is that a basket?’(Lyon 2012: 26) d. * [yamˇxw aP basket  NP ]  ha ixiP ynq dem  ‘Is that a basket?’ (Lyon 2012: 26) Relative to the epistemic modals, (21) shows that the question particle ha can precede or follow mat and (22) shows that ha can only precede cmay. This provides further evidence that mat is located higher than cmay in the pre-predicative domain. (21) a. mat ha Brian kw u-ks-k-xiP-nt-m mod ynq Brian 2.pl.abs-irr-res-come-trans-3.sg.erg ‘Is Brian going to come with us?’ ‘I wonder if Brian is going to come with us?’ b. ha mat Brian kw u-ks-k-xiP-nt-m ynq mod Brian 2.pl.abs-irr-come-trans-3.sg.erg ‘Is Brian going to come with us?’ ‘I wonder if Brian is going to come with us?’ 14  (22) a. ha cmay ilxw in-k@kw´ap ynq mod hungry poss-dog ‘Is my dog hungry?’ IJ am´ IJ ayaPm b. * cmay ha John sxw m´ mod ynq John teacher ‘Is John a teacher?’ There is a second question particle Puc which has been labeled as the dubitative question particle (Mattina 1973). Puc, similar to interrogative ha, can only precede cmay as seen in (23a) and (23b) but can both precede and follow mat as seen in (23c) and (23d). The translations of the sentences and their felicity do not change relative to where the question particles are found. (23) a. Puc cmay Brian kw u-ks-k-xiP-nt-m q mod Brian 2.pl.abs-irr-res-come-trans-3.sg.erg ‘Is Brian going to come with us?’ b. *cmay Puc Brian kw u-ks-k-xiP-nt-m mod q Brian 2.pl.abs-irr-res-come-trans-3.sg.erg ‘Is Brian going to come with us?’ c. Puc mat Brian kw u-ks-k-xiP-nt-m q mod Brian 2.pl.abs-irr-res-come-trans-3.sg.erg ‘Is Brian going to come with us?’ d. mat Puc Brian kw u-ks-k-xiP-nt-m mod q Brian 2.pl-irr-res-come-trans-3.sg.erg ‘Is Brian going to come with us?’ Finally, the negation particle lut is found preceding and following the modals mat and cmay as shown in (24) and (25). There is no difference in the translation relative to where the negation particle is found.  15  (24) a. lut mat IJxw up iP Canucks neg mod win det Canucks ‘The Canucks must not have won’ b. mat lut IJxw up iP Canucks neg mod win det Canucks ‘The Canucks must not have won’  (25) a. cmay lut ks-q´ıc-@lx mod neg fut-run-lex ‘I might not run’ b. lut cmay ks-q´ıc-@lx neg mod fut-run-lex ‘I might not run’ Summarizing, my elicitations from the Douglas Lake dialect suggest that the order of the epistemic modals relative to the other pre-predicative particles appears to be flexible with a few fixed orderings. Table 1.4 puts these findings into order. Table 1.4: Order of epistemic modals and pre-predicative particles lut Puc mat Puc cmay lut ha ha  The data presented in this section was intended to give the syntactic ordering of the pre-predicative domain in Nsyilxcen with specific focus on the epistemic modals. More research is necessary to determine the syntactic ordering of the entire prepredicative domain in Nsyilxcen. Furthermore, this data is based on what surface word orders are permitted and not permitted by consultants. This data does not consider a more abstract syntactic structure that accounts for the different basegenerated positions plus movements.  16  Chapter 2  Nsyilxcen Modality There are four specialized modals in Nsyilxcen; three epistemic modals, mat, cmay, and kw ukw and one non-epistemic bouletic modal cakw . Deontic, teleological, ability, and pure circumstantial modality are expressed by using the imperative, irrealis as well as the bare predicate. Table 2.1 summarizes the Nsyilxcen modal system for the Douglas Lake dialect. Table 2.1: Nsyilxcen modality Modal mat cmay kw ukw  Modal cakw x and -´ık  ks∅  Epistemic Translation must/might/maybe might/maybe ‘I heard’/ ‘Someone told me’ Non-Epistemic Translation ‘should’, ‘would’, ‘wish’ Transitive imperative and intransitive imperative Irrealis Bare predicate  Gloss Epistemic Epistemic possibility Reportative  Gloss Bouletic Deontic, Teleological  Deontic, Teleological Deontic, Teleological, Ability, Circumstantial  The present chapter introduces the basic features of the Nsyilxcen modal system. Section 2.1 discusses A. Mattina’s (1973) data for the modals discussed in his thesis. Section 2.2 summarizes how Nsyilxcen expresses epistemic modality, providing data and generalizations for mat and cmay and for the reportative modal kw ukw . Finally, section 2.3 discusses how Nsyilxcen speakers express non17  epistemic modality. It defines each type of non-epistemic modality and provides a separate section for each particle used to express non-epistemic modality.  2.1  Colville-Okanagan modality: A. Mattina 1973  Anthony Mattina’s thesis on the grammatical structure of Colville-Okanagan lists the modal particles in their underlying form and provides a translation for each (A. Mattina 1973). Table 2.2 provides Mattina’s description of the pre-predicative modal particles in their underlying form. Table 2.2: Modal particles (A. Mattina 1973) Translation Modal Particle ´ Probable //cm// Conjectural //mt// Conditional //cakw // ´ as a particle that marks the proposition as probable. Mattina (1973) describes //cm// //mt// marks the proposition as possible but in doubt (Mattina 1973: 135). However, the examples provided in (26) and (27) from Mattina (1973) suggest that the inverse is true, where //cm// ´ expresses possibility and //mt// expresses probability. 3  (26)  IJ ct-@lx c@m lut mod say-3.pl.erg neg ‘They might say: no’ (Mattina:1973)  (27)  m@t conjectural IJyi ...definite  tw-m-´ı-s-t-(´ı)s myrtle sell-middle-secondary-causative-transitive-she Myrtle citxw -s house-hers  ‘Myrtle’s house must have been sold’ (Mattina:1973) ´ produced by My fieldwork on the Douglas Lake dialect has revealed that //cm//, the speakers as cmay, expresses solely notions of possibility and doubt, and //mt//, presented as mat here, expresses notions of possibility and doubt, but also covers stronger modal notions of probability and necessity. 3  cmay alternates with the form c@m.  18  A. Mattina (1973) labels the modal particle cakw as a conditional expressing notions of ‘should’, ‘would’, or ‘wish’. My fieldwork revealed similar translations which will be presented in section 2.3.1. 4 (28)  cakw suliPck´ıcx ‘Wish that Julie would get here’ (Mattina 1973: 134)  (29)  cakw Pl´aP kiP n@Qc´us@nt@m ‘We should set a trap for it right here.’ (Mattina 1973: 134)  2.2  Nsyilxcen epistemic modality: The Douglas Lake dialect  Epistemic modality is the expression of possibility and necessity given the speaker’s knowledge and beliefs. The three Nsyilxcen particles used to express epistemic modality are repeated in table 2.3.  Modal mat cmay kw ukw  Table 2.3: Nsyilxcen epistemic modals Translation Gloss Must, might, maybe Epistemic Necessity, Epistemic Possibility (EPIS) Might, maybe Epistemic Possibility (EPIS.POSS) I heard, Someone told me Reportative (REP)  To begin, mat and cmay express epistemic modality and are felicitous in contexts where the speaker is making an inference about the situation at hand based on their previous knowledge, beliefs, and experiences. Where these two modals differ is based on their modal strength, whether they are felicitous in necessity or possibility contexts. A necessity context is one where the speaker is more certain about the truth of a proposition based on strong evidence or reasoning. This can either mean that the speaker has convincing physical evidence available to them or the speaker has a lot of previous knowledge about the situation at hand. Often any type of physical evidence whether it is visual, auditory or olfactory evidence will 4  There are no morpheme glosses for (28) and (29) in Mattina (1973).  19  serve to strengthen the context resulting in a necessity interpretation. A strong necessity context is one where the English modal must is used. In (30), the bloody handprints, footprint, and murder weapon close to the window, and the fact that the window was open and the door was locked from the inside is strong evidence that the murderer escaped out the window. (30)  Context: You are a detective in a murder case. In the room where the murder took place, the only door was bolted on the inside. However you see bloody handprints on the window sill, footprints just outside the window, and a dropped knife outside the window which is probably the murder weapon. He must have escaped out the window ?? He may have escaped out the window  A possibility context is one where the speaker is less certain about the truth of the proposition. This could be due to the presence of poor physical evidence or the fact that the speaker cannot make a strong statement based on their intuition or reasoning because there are too many options that could be true. In (31), there is evidence of a footprint outside the window but the speaker cannot deduce that the murderer escaped that way because there were other ways out of the room and the murderer left no trace. (31)  Context: You are a detective in a murder case. In the room where the murder took place, there are several doors, all of which were unlocked when the police arrived. There are no fingerprints anywhere, and no sign of a murder weapon, but you see a faint footprint just outside the window. You say: ?? He must have escaped out the window He may have escaped out the window  During elicitation, pairs of contexts like (30) and (31) were presented to the consultant. These contexts, and variations of the contexts were paired with Nsyilxcen modal sentences to elicit felicity judgments and translations. The examples presented here are those that were consistently judged as good or bad within each context. 5 Any speaker comments are provided below the examples and are often 5  Due to the complexity of the contexts presented to the speakers, some variability in judgments occurred. The majority of variability was between the different speakers. Any variability between the answers of individual speakers were reconciled by clarifying the contexts and checking the data and contexts over a number of elicitation sessions.  20  insightful as to whether the speaker believed the context was compatible with a necessity interpretation or a possibility interpretation. Take into consideration the following Nsyilxcen data. In (32), the speaker is making an inference about Mary’s whereabouts. In this case there are multiple places that Mary could be at the present moment; running, shopping or at school. In this context both mat and cmay are felicitous and are translated as might. The speaker is stating that, based on their knowledge about Mary’s schedule and interests, it is a possibility that Mary is running at the present moment. (32)  Context: You know that Mary loves to go running and often goes on runs randomly. She could also be at the store or at school. I ask you, where is Mary?  a. Mary mat ac-q´ıc-@lx Mary epis.mod cust-run-lex ’Mary might be running’ b. Mary cmay ac-q´ıc-@lx Mary epis.mod cust-run-lex ’Mary might be running’ However, the possibility context in (32) stands in opposition to the necessity context in (33) where the speaker has more information about Mary. In this case, the speaker knows that Mary has a strict training schedule and runs every Tuesday at 6pm. Because of the strict nature of her exercise regime this context is much stronger than if she runs randomly during the week and, as a result, the speaker makes a stronger modal statement. In this context mat is again felicitous but with a necessity must translation. cmay on the other hand is now infelicitous and the speaker comments that cmay is inappropriate in this context because it is used to make a guess. (33)  Context: Mary runs every day to train for a marathon. She usually runs at 6pm on Tuesdays. Today is Tuesday and it’s 6pm. I ask you, where is Mary?  a. Mary mat ac-q´ıc-@lx Mary epis.mod cont-run-lex ’Mary must have went running’  21  b. # Mary cmay ac-q´ıc-@lx Mary epis.mod cont-run-lex ’Mary might be running’ Consultant comment: ‘cmay. We are just guessing. Mary might be running.’ The generalization based on this data is that the epistemic modals mat and cmay are distinguished by their modal strength. mat is variable and is permitted in both necessity and possibility contexts and cmay is restricted to possibility contexts. To elicit this difference in strength, different contexts were presented to the speakers which were designed to specifically elicit a stronger necessity modal claim or weaker possibility modal claim. (34) shows that mat and cmay are both felicitous in a context where there is a degree of uncertainty. This context was set in a larger discussion about how unusually cold the summer had been and how the berries may ripen later than usual. As a result the speaker is unsure whether the berries are ripe or not. (34)  Context: You know that the berries in the Okanagan are usually ripe in July. You don’t know if the berries are ripe right now but, since it is July then based on your knowledge the berries might be ripe.  a. mat p’y’aq iP sia Qapn´aP epis.mod ripe det berries now ’The berries might be ripe now’ b. cmay p’y’aq iP sia Qapn´aP epis.mod ripe det berries now ’The berries might be ripe now’ (35) provides an example with visual evidence where the speaker can reason with certainty that John is going hunting based on the fact that he is wearing his hunting gear and is holding a gun. As a result mat is permitted with a necessity translation and cmay is infelicitous. (35)  Context: Our friend John has been talking about going hunting for weeks. One day we see John with camouflage clothing on and a gun in his hand.  22  a. mat ks-c-piˇx-aPx epis.mod irr-cust-hunt-incep ‘He must be going hunting’ b. # cmay ks-c-piˇx-aPx epis.mod irr-cust-hunt-incep ‘He must be going hunting’ The following context (36) has been translated as both a possibility and necessity context. When mat was translated as must the speaker’s comment indicated that she was interpreting it as a necessity context. On a separate occasion, the fact that mat was translated as might and that cmay is felicitous indicates that (36) can also be a possibility context. (36)  Context: Our friend John has been talking about going hunting for weeks. One day John isn’t around. Where is John?  a. mat ks-c-piˇx-aPx epis.mod irr-cust-hunt-incep ’He might be going hunting’ b. mat ks-c-piˇx-aPx epis.mod irr-cust-hunt-incep ’He must be going hunting’ Consultant (LL): He must have got up and left before anyone else got up. c. cmay ks-c-piˇx-aPx epis.mod irr-cust-hunt-incep ’He might be going hunting’ (37) provides a context with poor physical evidence. The speaker is directly looking at John down the beach but because she has poor eye-sight and cannot see clearly, she must infer it is John based on her knowledge about John’s appearance. In this case both mat and cmay are felicitous with possibility might translations as the speaker cannot be sure that the blurry vision of the man down the beach is in fact John.  23  (37)  Context: Sarah is walking down the beach and sees someone walking towards her far away. She forgot her glasses at home and can’t make out who it is, but it looks like it might be John.  a. mat ix´ıP John epis.mod dem John ’That might be John’ b. cmay ix´ıP John epis.mod dem John ’That might be John’ Evidence can also be auditory and can serve to strengthen the context. In (38) a growling stomach is a strong indication that I am hungry. When epistemic contexts contain strong sensory evidence cmay is infelicitous, and mat elicits a necessity translation. (38)  Context: We are working away and my stomach starts to growl. You think, she must be hungry.  a. mat tal´ı Pilˇxw t epis.mod very hungry ‘She must be hungry’ b. # cmay tal´ı Pilˇxw t epis.mod very hungry ‘She might be hungry’ Consultant: With cmay it means maybe she is hungry. It’s like you are asking or curious. You can’t say that if you hear her stomach growl. (39) shows that cmay is only permitted if context (38) is changed to a weaker epistemic context where the speaker has no evidence and is uncertain about whether I am hungry. In this case both mat and cmay are felicitous with possibility translations. (39)  Context: You don’t hear my stomach growling and I’m not in the room. It is supper time and we have been working for a while.  24  a. mat tal´ı Pilˇxw t epis.mod very hungry ‘She might be hungry’ b. cmay tal´ı Pilˇxw t epis.mod very hungry ‘She might be hungry’ In contexts (40-41) the same visual evidence is presented to a father and son who have different background knowledge and experience through which they are evaluating the situation at hand. However, because the father has more background knowledge about what the evidence tells him he can be more certain that the proposition ‘a bear came by here’ is true. The father can use mat which is translated as must, and cmay is infelicitous. (40)  Context: A father is taking his young son out to teach him how to hunt. The son doesn’t know much about hunting or tracking animals. His dad is very good at hunting and tracking and he is teaching his son everything he knows about tracking animals. They come across a broken branch on the trail and the leaves of the nearby bush have been flattened. These are signs that a large animal like a bear has been by this way. The father says:  a. mat at´aP xiPw´ılx iP sk@m´xist epis.mod dem walk.by det bear ‘A bear must have come by here’ b. # cmay at´aP xiPw´ılx iP sk@m´xist epis.mod dem walk.by det bear ‘A bear might have come by here’ The son on the other hand, has less background knowledge about hunting and tracking animals so the son is less certain that a broken branch means a bear came by this way. This context was presented to the speaker in a way that focused on the fact that the son does not know what type of animal created the damage that he sees. As a result both mat and cmay are permitted and can be translated as might.  25  (41)  Context: A father is taking his young son out to teach him how to hunt. The son doesn’t know much about hunting or tracking animals. His dad is very good at hunting and tracking and he is teaching his son everything he knows about tracking animals. They come across a broken branch on the trail and the leaves of the nearby bush have been flattened. The little boy is not sure what kind of animal came by this way. The son says:  a. mat ataP xiPw´ılx iP sk@m´xist epis.mod dem walk.by det bear ‘A bear might have come by here’ b. cmay ataP xiPw´ılx iP sk@m´xist epis.mod dem walk.by det bear ‘A bear might have come by here’ The Nsyilxcen data in this section provides support for the claim that mat and cmay are both epistemic modals. The data also shows that mat is permitted in contexts that are compatible with both necessity and possibility translations. This is opposed to cmay which is only permitted in contexts that are compatible with possibility translations. The next section will discuss that mat and cmay also have an evidential restriction expressing that the speaker is basing the modal statement on indirect inferential evidence. Furthermore, section 2.3.4 will show that mat and cmay are only permitted in epistemic contexts and are infelicitous in non-epistemic contexts.  2.2.1  Source of evidence and Nsyilxcen epistemic modals  As discussed by von Fintel and Gillies (2010) and Matthewson (2011) among others, epistemic modals can specify the source of evidence. Willett (1988) defines direct evidence as claims about situations that were directly perceived by the speaker and indirect evidence as claims about situations that are not directly perceived by the speaker. Modal statements using mat and cmay specifically encode an indirect evidential restriction and are not used if the situation was witnessed directly. mat and cmay are infelicitous in contexts where the speaker has direct evidence of the action or event as seen in (42b) and (42c). (42)  Context: I look outside and see that it is raining.  26  a. way s-c-IJ qitx aff nom-cust-rain ‘Yes, It is raining’ b. # mat s-c-IJ qitx epis.mod nom-cust-rain ‘It must/might be raining’ c. # cmay s-c-IJ qitx epis.mod nom-cust-rain ‘It might be raining’ The Nsyilxcen epistemic modals also make a division between indirect evidence based on inference and indirect evidence based on a report. This fits with Willett’s (1988) cross-linguistic evidential typology which will be discussed in section 3.2. The Nsyilxcen has a third specialized reportative modal kw ukw used to express that the source of evidence comes through a report from another person. In contexts where the speaker did not directly witness the event being reported, the reportative kw ukw is felicitous as seen in (43). However, kw ukw is infelicitous when the speaker has directly witnessed the event as shown in (44). (43)  Context: My friend Tessa saw Sarah kiss Rodger and she tells me about it. I then tell you what I heard from Tessa. kw ukw Sarah IJtmPas Roger rep Sarah kiss Roger ‘[someone said] Sarah kissed Roger’  (44)  Context: My friend Tessa saw Sarah kiss Rodger and she tells me about it. I also saw Sarah kiss Rodger. I then tell you what I heard from Tessa. # kw ukw Sarah IJtmPas Roger uł wik-n rep Sarah kiss Roger and see-1.sg.erg ‘Someone said that Sarah kissed Roger and I saw it’  The following examples (45) and (46) show the same result. (46) is infelicitous because the speaker was present at the wedding and witnessed Jill get married. 27  (45)  Context: My friend tells me that our friend Jill got married over the weekend. I go to my parents’ house. My parents also know Jill. I tell them. kw ukw c-mrim Jill rep cont-married Jill ‘I heard Jill got married’  (46)  Context: I was at Jill’s wedding, I saw them get married and helped them celebrate. John didn’t see me there so he tells me ”Hey! Jill got married”. I tell my parents. # kw ukw Jill c-mrim rep Jill cont-married ‘I heard Jill got married’ Consultant: You can’t say that if you were there  (47a) and (47b) show that mat and cmay are infelicitous in such contexts and cannot be translated as a report. (47)  Context: My friend tells me that our friend Jill got married over the weekend. I go to my parents’ house. My parents know Jill. I tell them  a. # Jill mat c-mrim Jill epis.mod cust-get.married ‘Jill might have gotten married’ b. # Jill cmay c-mrim Jill epis.mod cust-get.married ‘Jill might have gotten married’ As the Nsyilxcen data shows, the epistemic modals mat and cmay express necessity and possibility as well as an evidential restriction specifying the speaker was using indirect inference from previous knowledge, beliefs, and experience. This is opposed to kw ukw which specifies a reportative evidential restriction and is used to express that the source of evidence comes indirectly through a report. I propose that mat and cmay are distinguished from each other with respect to their differences in modal strength. mat is permitted in contexts that allow for both necessity and possibility interpretations. cmay is restricted to contexts that allow for possibility interpretations only. Whether it is a necessity or possibility context is connected to the strength of the evidence available to the speaker. 28  2.3  Nsyilxcen non-epistemic modality  This section aims to document how Nsyilxcen speakers express non-epistemic modal readings. Non-epistemic readings express necessity and possibility over worlds that are compatible with a deontic, circumstantial, bouletic, ability, or teleological conversational background. Each type of non-epistemic modality is defined in (48) (von Fintel 2006). (48) a. Ability - Expresses possibility given the physical limits of a particular person or thing. b. Bouletic - Expresses possibility and necessity given a person’s desires. c. Circumstantial - Expresses possibility and necessity given particular circumstances. d. Deontic - Expresses possibility and necessity given what is permitted or obligatory given a set of laws, regulations or moral principles. e. Teleological- Expresses what is possible and necessary for achieving a goal. Table 2.4 lists Nsyilxcen particles that have been elicited in non-epistemic contexts. The first is a modal cakw , the second two are imperative markers, the third is the irrealis marker, and finally the speakers can simply use the basic predicate. Table 2.4: Nsyilxcen particles in non-epistemic contexts Nsyilxcen cakw x and -´ık -ks ∅  Gloss ’should’, ’would’, ’wish’. Transitive imperative and intransitive imperative Irrealis Bare predicate  Context Bouletic Deontic, Teleological Deontic, Teleological Deontic, Teleological, Ability, Circumstantial  My fieldwork on the Douglas Lake dialect shows that there is only one non-epistemic modal cakw , which is used in bouletic contexts. Because Nsyilxcen speakers do not have deontic, teleological, ability, or circumstantial modals they have other 29  ways of paraphrasing. The irrealis ks- is felicitous in deontic, teleological, circumstantial, and ability contexts. The irrealis is a grammatical category indicating that a situation has not yet occurred at the time of utterance. The imperative, a grammatical form often used in commands, is permitted in permission contexts. Finally, Nsyilxcen speakers can use the basic predicate in deontic and ability modality contexts. The following sections provide data for each of the particles and discusses how each can be used to express non-epistemic modality in Nsyilxcen.  2.3.1  cakw bouletic modal  cakw is a bouletic modal and is used to express the speaker’s wishes and desires. Examples (49-51) provide contexts eliciting a bouletic conversational background. In (49), based on the Olympian’s desire to remain unhurt, he should drop the weights. (49)  Context: We are watching the weight lifting category during the Olympics. We see an athlete holding a very heavy weight above his head. He is shaking very hard, if he doesn’t drop it soon his arms are going to give out. I say to you: cakw yxw mi-s boul.mod drop-3.sg.erg ‘He should drop it’  In (71), based on the speaker’s desire for more meat, it is necessary that John goes hunting. (50)  Context: We running low on meat supplies and need to get more so we can make it through the winter. You say that John should go hunting. cakw ac-piˇx-m boul.mod cust-hunt-mid ‘He should go hunting’  Finally in (51), based on the general traits and our desires for a good teacher, it is necessary that a teacher is smart. (51)  Context: We are discussing the traits that a teacher should have.  30  IJ m´ IJ ay´aPm cakw sysy´us iP scmaP boul.mod smart det teacher ‘A teacher must/should be smart’ Context (52) presents as weak necessity context where the speaker’s goal is to reach Penticton from Merritt in the fastest way possible. The speakers use the bouletic modal cakw to express that if the speaker’s desire is to achieve their goal of getting to Penticton quickly then they should take the car. (52)  Context: I am in Merritt and am trying to get home to Penticton quickly. You know that the train takes a very long scenic route, and the bus stops a lot and takes a very long time. You tell me, “ If you want to get home quickly, you should take your car’ kw -s-xw s-lx xw uy. cakw lan automobile ki kł-xw uy 2.sg.abs-nom-rush-lex go boul.mod loc car comp irr-go ‘You are in a rush. You should take your car when you go’  When elicited in epistemic contexts, cakw is infelicitous or corrected to a bouletic translation as seen in (53). (53)  Context: You and a friend are at the store and you see a dollar coin on the ground. It could be yours, your friend’s, or the person in front of you in line may have dropped it. # cakw yxw mi-nt-s boul.mod drop-trans-3.sg.erg ‘He might/must have dropped it’  When mat or cmay are elicited under a bouletic context, the translation changes to one where the speaker is making an inference. The consultant’s comment that it sounds like the speaker is guessing indicates that it is no longer a bouletic statement. (54)  Context: My sister hurt her leg. The doctor says that she should not walk on it for a week.  a. lut mat t’-ks-xw isc neg epis.mod emph-irr-walk ‘She might not walk’ 31  b. cmay t’-ks-xw isc epis.mod emph-irr-walk ‘She must walk’ Consultant comment: ‘You are guessing’ c. cakw t’-ks-xw isc boul.mod emph-irr-walk ‘She should not walk’ As discussed by Mattina (1973), cakw is used to express notions of ‘should’, ‘would’, or ‘wish’. These notions can be classified under the bouletic non-epistemic modality label.  2.3.2  -ks irrealis  The irrealis marker -ks can be used to express deontic, teleological, circumstantial, and ability modality. First, deontic modality expresses possibility and necessity given what is permitted or obligatory given a set of laws, regulations, or moral principles (von Fintel 2006). Examples (55) and (56) present deontic necessity contexts where a mother has authority over her daughter and is enforcing the rules of her household. In a context where the mother is speaking directly to the daughter the irrealis can be used. (55)  Context: Mary’s mother has told Mary that she must finish three chores before she can play with her friends. (TFS Working Group 2011) 6 c´u(n)-nt-@m “lut kaPì´ıs a-kì-s-k’w ’ul’@m sic kw -c-iP´ıck@n” say-trans-mid “neg, three 2.sg.poss-irr-nom-work then 2.sg.abs-cust-play” ‘She said “No, you will do three chores before you play” ’  In example (56) the little girl is telling her friends that she is not permitted to go play because she must finish her three chores. 6 This data was elicited using the story Chore Girl developed by Totem Fields Storyboards team: TFS Working Group (2011). Chore Girl. Retrieved from http://totemfieldstoryboards.org/stories/chore girl/ on Oct 25, 2012.  32  (56)  Context: The little girl’s friends come by and ask her to go out and play with them. She informs them that her mother says she can’t go outside because she has to/must finish her chores (TFS Working Group 2011). lut ks-ac-iP´ıck@n mi wiP-st-in-s-kw lmi sic neg irr-cust-play fut finish-trans-1.sg.poss-nom-chores then kn-ac-Pickn 1.sg.abs-cust-play ‘No, I will not play, I will finish my chores, then play’  In context (57) the doctor is speaking to my mother about my little sister, giving an order that my little sister must stay off of her broken leg. (57)  Context: My little sister broker her leg. The doctor said to my mom that she is not permitted to walk on it so that it has a chance to heal. lut aP-ks-xw i-stc neg asp-irr-walk-caus ‘She will not walk’  Example (58) is a negated deontic possibility context, where Mary has finished her chores and is given permission to go outside. In this context the mother is speaking directly to the daughter. (58)  Context: Mary has finished all of her chores so her mom says, ”you can go outside and play” (TFS Working Group 2011). IJ cu(n)-nt-@m iP t tums “way ks-xuy-Pax. Pickn-x” say-ctr-mid det obl woman’s.mother “yes, irr-go-incep. play-imp” ‘Her mother said “yes, you will go play”  From the first person point of view the little girl is expressing that she is permitted to go out and play. (59)  Context: Mary says to her friends that she is allowed to go outside and play (TFS Working Group 2011).  33  way kn ks-P´ıckn-aPx Qapan´aP aff 1.sg.abs irr-play-incep now ‘Yes, I can go play now’ Next, teleological modality expresses the necessary or possible means for achieving a goal (von Fintel 2006). Context (60) is a necessity context where there is no other option to achieve the goal of getting to Penticton. (60)  Context: I am in Merritt and am trying to get home to Penticton. I find out that the Greyhound is not running from Penticton to Merritt anymore. You say to me “In order to get home, you have to take a car” (There is no bus option, there is no train option) ł(a)-an-ˇxm´ınk a-ks-xw uy lan automobile mi kw -ł-xw uy comp-2.sg.poss-want 2.sg.poss-irr-go loc car fut 2.sg.abs-dir-go ‘If you want to go home, you will take your car’  Context (61) provides a possibility context, where there are many options for getting to Penticton; by train, bus, or car. The speaker suggests to the addressee that they can take the bus. (61)  Context: I am in Merritt and want to get to Penticton. There are many options for getting there; train, bus, or car. You tell me that I can take the bus. łan-xamink aP-ks-xw uy n-Pmut iP l bus kw u-xw uy 2.sg.poss-want asp-irr-go part-sit det loc bus 2.sg-go ’You want to go. Take the bus to go’  The irrealis can also be used to express circumstantial modality, which expresses necessity and possibility given a certain set of circumstances. The following examples provide a circumstantial context based on Kratzer’s (1991) hydrangeas example. The context that Kratzer presents is used to distinguish between circumstantial and epistemic modality. Kratzer’s example is provided in (62). (62)  Context: Suppose I acquire a piece of land in a far away country and discover that soil and climate are very much like at home, where hydrangeas prosper everywhere. Since hydrangeas are my favourite plants, I wonder whether they would grow in this place and inquire about it. 34  a. Hydrangeas can grow here. b. There might be hydrangeas growing here. In this context, (62a) can be true whether or not there are hydrangeas already in existence. It does not make a statement about whether there are already hydrangeas growing there. All that matters is the type of soil, climate, and the properties of the hydrangeas. (62b) on the other hand is only true if the evidence in the actual world indicates that hydrangeas are growing there. It would be false if the evidence is not compatible with the growth of hydrangeas in the actual world. A similar context was presented to the Nsyilxcen consultants. Examples (64) and (65) provide other circumstantial examples. (63)  Context (Circumstantial): My mother is buying plants in Vancouver to grow in Penticton. Penticton has a very dry climate, which is very different from Vancouver which has a very wet climate. My mom asked if these plants will grow in Pentiction, given the temperature, weather, and soil in Penticton. The sales person says: waIJy, ks-pl-m´ıxaPx iP snpintkn aff, irr-grow-incep det penticton ‘Yes, the flowers will grow in Penticton’  Example (64) provides another circumstantial context where the irrealis particle is felicitous. Given the fact that I am strong enough to climb the cliff then my sister, being stronger than I am, is also strong enough to climb the cliff. (64)  Context: My sister and I are rock climbing together. My sister is much stronger and better at climbing than I am. If I am able to climb to the top of the cliff then my sister is also strong enough to climb the cliff. I say to my sister: way t’-ks-kw n-nun-(n)t-m aff emph-irr-climb-manage.to-dir-mid ‘You are going to make it’  Finally, example (65) also provides a circumstantial example, where given the state of my nose, I have no choice but to sneeze. 35  (65)  Context: I have to sneeze (given the state of my nose) (Kratzer 1977). kn ks-Qat’´ısx-ax 1.sg.abs irr-sneeze-con ‘I am going to sneeze’  Example (66) shows how the irrealis marker is also felicitous in ability contexts. (66)  Context: I am physically strong enough to lift the big rock. way ks-snwislx-st-n iP ut aff irr-lift-intr-? det rock ‘Yes, I am going to lift the rock’  2.3.3  Further ways to express non-epistemic modality  There is limited data for the bare predicate and imperative markers. The bare predicate is felicitous in a deontic context as seen in example (67a) where the little girl is telling her friends that she is not permitted to go play because she must finish her three chores. Example (67b) provides an example where the little girl has finished her chores and is permitted to go outside. (67)  Context: The little girl’s friends come by and ask her to go out and play with them. She informs them that her mother says she can’t go outside because she has to/must finish her chores (TFS Working Group 2011).  a. cut lut, xIJy ti s-ckw ult say neg, finish part nom-work ‘She said, “ No, I am finishing work”’ b. waIJy kn-ac-Pickn PapanaP aff 1.sg.abs-cust-play now ‘Yes, I will play now’ The bare predicate can also be used in ability contexts as seen in (68). (68)  Context: I am physically strong enough to lift the big rock.  36  tali kn-k’w @ck’w a´ ct n-wis-lx-st-n iP x ut very 1.sg.abs-strong lift-intr der-lift-lex-trans rock ‘I am very strong, I lift the rock’ Example (69) provides another ability context where Mary’s broken leg has healed and she is physically able to walk on it again. The bare predicate can be used in this case as well. (69)  Context: When Mary’s leg heels the doctor says that she has the strength to walk on it without getting hurt. way x@st-w´ılx way xw ist aff good-dev aff walk ‘She is better now, she can walk on it’  Example (70) provides an example where the mother is speaking directly to the daughter. This is a deontic possibility context, where Mary has finished her chores and is given permission to go outside. In this case the imperative or the irrealis can be used. (70)  Context: Mary has finished all of her chores so her mom says, ”you can go outside and play” IJ cun-nt-@m iP t tums ”way xuy-x Pickn-x” say-ctr-mid det obl woman’s.mother ”aff, go-imp play-imp” ‘Her mother said “yes, go play” (TFS Working Group 2011)  Nsyilxcen appears to have only one specialized modal cakw that expresses bouletic modality. There are no specialized deontic, circumstantial, teleological, or ability modals. As the data shows, Nsyilxcen speakers paraphrase to fill the absence of these non-epistemic modals. The irrealis particle ks- is also frequently used to express many different types of non-epistemic modality; deontic, teleological, circumstantial, and ability modality. The imperative particles are used when the speaker is directly giving an order to the addressee. Finally, the bare predicate is used to express deontic and ability modality.  37  2.3.4  Epistemic modals in non-epistemic contexts  The following data shows that mat and cmay are strictly epistemic and are never volunteered in non-epistemic contexts. (71b) and (71c) show that mat and cmay are rejected in a bouletic context where cakw is found. In this context, the speaker is expressing that based on their desires to survive the winter, John should go hunting. (71)  Context (Bouletic): We are running low on meat supplies and need to get more so we can make it through the winter. You say John should go hunting.  a. cakw ac-p´ıxˇ -@m deon cust-hunt-middle ‘He should go hunting’ b. # mat ac-p´ıxˇ -@m epis.mod cust-hunt-gen ‘He must have gone hunting’ c. # cmay ac-p´ıxˇ -@m epis.mod cust-hunt-middle ‘He might have gone hunting’ (72) presents an ability context where the speaker talking about how lifting the large rock is within the restrictions of the subject’s physical ability. In (72a) the bare predicate is used to paraphrase. (72b) and (72c) show that mat and cmay are not permitted in ability contexts such as this. (72)  Context (Ability): He is physically strong enough to lift a large rock.  a. taliP kIJw @ckw a´ ct n-wis-lx-st iP x ut very strong der-lift-lex-trans det rock ‘He is very strong. He lifts this rock’ b. # mat n-wis-lx-st iP x IJut epis.mod der-lift-lex-trans det rock ‘He must have lifted this rock’ c. # cmay n-wis-lx-st iP x IJut epis.mod der-lift-lex-trans det rock ‘He might have lifted this rock’ 38  In Kratzer’s hydrangeas context, the modal propositions with mat and cmay are not felicitous as seen in (73b-c). (73)  Context (Circumstantial): My mother is buying plants in Vancouver to grow in Penticton. Penticton has a very dry climate, which is very different from Vancouver which has a very wet climate. My mom asked if these plants will grow in Pentiction, given the temperature, weather, and soil in Penticton. The sales person says:  a. waIJy, ks-pl-mixax iP snpintktn aff, irr-grow-incep det Penticton ‘Yes, the flowers will grow in Penticton’ IJ snpintktn b. # mat plal kl epis.mod grow loc Penticton ‘The flowers might grow in Penticton’ IJ snpintktn c. # cmay plal kl epis.mod grow loc Penticton ‘The flowers might grow in Penticton’ mat and cmay are infelicitous because, in this context, the sales person is speaking about whether the flowers have the potential to grow in Penticton given the soil and the climate. The Nsyilxcen consultants translate the sentences with mat and cmay as if the speaker is guessing about whether, in our world, the flowers would actually grow in Penticton.  2.4  Chapter summary  This chapter has provided Nsyilxcen data for epistemic and non-epistemic modality. It showed that mat and cmay are only permitted in epistemic contexts and are infelicitous in non-epistemic contexts. mat and cmay express that indirect inferential evidence was used to make the modal statement and neither modal can be used to express something that is being directly perceived. Where mat and cmay differ is with respect to their modal strength. mat is a variable modal that is felicitous in contexts that are compatible with both necessity and possibility readings. cmay on the other hand is restricted to contexts that are compatible with a possibility reading only. The third epistemic modal kw ukw is used to express that the source of 39  evidence is from a report. For non-epistemic modality, there is a specialized modal cakw that expresses bouletic modality and the Nsyilxcen speakers make use of different particles in contexts which support other types of non-epistemic modality. The following chapters focus on developing a semantic analysis for the epistemic modals mat and cmay. Chapter 3 discusses the semantic theory underlying this research. Chapter 4 discusses different approaches to similar modal systems. Finally, Chapter 5 discusses three potential analyses for these modals.  40  Chapter 3  Theoretical Background Modality is the expression of necessity and possibility with respect to certain bodies of knowledge (epistemic), or certain facts (non-epistemic). This chapter discusses how the information expressed by a modalized sentence is semantically encoded. Under a standard theory of modality as developed by Kratzer (1977, 1981, 1991) the evaluation of a modal expression consists of three integral components; possible worlds, modal force, and conversational background. How these components are encoded across languages provides a basis for how modal systems can be compared. Section 3.1 will discuss how the modal force and conversational background interact to result in different modal readings. Section 3.2 discusses the evidential restrictions and epistemic modality and section 3.3 will provide a summary of cross-linguistic differences that have emerged from this modal framework.  3.1  Kratzerian theory of modality  A modalized sentence evaluates the basic proposition (prejacent) relative to a set of possibilities (von Fintel 2006). These possibilities are alternatives to the actual world and allow speakers to discuss situations that are not rooted in the here and now. This is represented in modal theory as a set of possible worlds W; a potentially infinite set of alternate worlds that can be very similar to or vastly different from the actual world. Different modal readings emerge as a result of how the conversational background and modal force interact to select a specific subset of the set of possible worlds W. The strength of the modal claim is represented by the modal force. Like quantifiers taking universal or existential quantification over sets of individuals, modals quantify over possible worlds. Modal necessity is formalized as universal (∀) quantification over worlds, and modal possibility is formalized as existential (∃) quantification over worlds. Finally, modality expresses necessity and possibility with respect to certain knowledge or facts. As previously stated there is a basic distinction drawn between epistemic modality expressing the speaker’s knowledge and beliefs, and non-epistemic modality covering a number of readings including deontic, pure circumstantial, and ability. 41  The different types of modality are provided by the conversational background. English modals are not specified with respect to the type of modality and therefore the conversational background is supplied by the context. For instance, (74) shows that the modal must can be used for both epistemic and deontic modality. Other languages, including some Salish languages, may specify the type of conversational background for each modal (Matthewson et al. 2007, Rullmann et al. 2008). (74) John must be at home (In view of what I know) John must be at home (In view of what the rules require)  (Epistemic) (Deontic)  To formally distinguish between different modal readings there are two conversational backgrounds, the modal base and the ordering source (Kratzer 1977, 1981, 1991). The modal base is expressed as a function f that takes a world as an argument and returns a set of propositions. With an epistemic modal base the set of propositions, f (w), represents what the speaker knows in w. A deontic modal base will consist of a set of propositions f (w) representing the relevant laws, moral obligations or social conventions in w (Kratzer 1977, 1981, 1991). The intersection of the set of propositions, f (w), yields the set of worlds in which the propositions in f (w) are all true. These worlds are considered to be accessible from w. A world v is epistemically accessible from w if everything the speaker knows in w is also true in v, or it is deontically accessible if all of the rules that are present in w are obeyed in v. In the epistemic example John must be home, the modal base will consist of propositions that express what the speaker knows about John. For instance, this can include propositions expressing the evidence available to the speaker, the speaker’s knowledge or previous experiences. f (w) = {The lights are on at John’s house, John’s car is in the driveway, John goes home after work, ...} Given these propositions, the set of worlds that are accessible from the actual world are those where these propositions in the modal base are all true. Next, the ordering source g is also a function that takes a world w as an argument and yields a set of propositions. This set of propositions g(w) places a partial ordering on the set of worlds in f (w). For an epistemic modal base there can be a stereotypical or doxastic ordering source. A stereotypical ordering source will consist of propositions that express the normal course of events in w and a doxastic ordering source will consist of propositions that express the speaker’s beliefs. 42  g(w) : {Lights on mean that someone is home, John drives everywhere, etc} The worlds in f(w) where the propositions in the ordering source g(w) are all true will be ranked above all other worlds. The ordering source is formally defined in von Fintel & Heim (2011: 61) as follows: (75)  Given a set of worlds X and a set of propositions P, define the strict partial order <P as follows: ∀w1 , w2 ∈ X: w1 <P w2 iff {p ∈ P : p(w2 ) =1} ⊂ {p ∈ P : p(w1 ) =1}  For all worlds w1 and w2 that are members of the set of modal base worlds X. w1 is ranked above w2 if and only if the set of propositions from the ordering source propositions P that are true in w2 is a proper subset of the set of propositions that are true in w1 (von Fintel & Heim 2011). The best worlds are then selected from the ordered worlds. This is formalized using a max operator as in (76) (von Fintel & Heim 2011:61). (76)  For a given partial order <P on worlds, define the selection function maxP that selects the set of < p best worlds from any set X of worlds: ∀ X ⊆ W: maxP (X) = {w ∈ X: ¬ ∃w’ ∈ X: w’ <P w}  For all sets of worlds X that are a subset of worlds W, the operator maxP selects a world w that is a member of X such that it is not the case that there is some world w’ that is a member of X where w’ is ranked above w. In the given example John must be at home the speaker uses the necessity modal must, which universally quantifies over the best worlds selected by the modal base and ordering source. In all the best worlds compatible with the speaker’s knowledge and beliefs, John is home. The English modal must is defined in (77) given a contextually given modal base f, and ordering source g. (77)  [[must]] w,g = λ f <s,st> . λ g <s,st> . λ q <s,t> . ∀ w’ ∈ maxg(w) (∩f(w)): q(w’) =1. (von Fintel & Heim 2011: 61)  A modal claim using must states that in the best worlds among the set of worlds where John’s car is in the driveway, the proposition John must be at home is true. All accessible worlds w’ that are selected by the modal base and maxP operator are worlds where the proposition q is true. 43  3.2  Modality and evidentiality  In addition to the quantification over worlds and contextually given conversational backgrounds, it has also been claimed that epistemic modals can specify the evidence source (Izvorski 1997, von Fintel & Gillies 2010, Kratzer 2010, and Matthewson 2011). Willett’s (1988) cross-linguistic survey of evidentiality discusses what types of evidence source are encoded across a number of languages. Based on his survey, figure 3.1 displays the types of evidentiality that languages encode. Figure 3.1: Types of evidence (Willett 1988) Direct Indirect Attested Reported Visual  Auditory  Inferred  Other Sensory Second-hand  Reasoning Third-hand  Folklore  Empirical evidence from Willett’s survey shows that languages make a basic distinction between direct and indirect evidence. According to Willett’s definitions, direct evidence refers to claims about situations that are directly perceived by the speaker and indirect evidence refers to claims about situations that are not directly perceived by the speaker. Indirect evidence covers both reportative and inferential evidentials. 7 Inferential evidence is further split into two separate sources of evidence, inference based on results and inference based on reasoning. Willet defines these two concepts as follows (Willett 1988): (78) a. Inference from results : The speaker infers the situation described from the observable evidence (i.e. from perception of the results of the causing event or action) b. Inference from reasoning : The speaker infers the situation described on the basis of intuition, logic, a dream, previous experience, or some other mental construct 7  Willett refers to inferential evidentials as inferred evidentials in figure 3.1  44  Results  Many Pacific Northwest languages lexically specify the type of evidence that the modal statement is based on. Epistemic modals in languages like St’´at’imcets and Gitksan specifically encode the source of indirect evidence for each modal (Matthewson et al. 2007, Rullmann et al. 2008, Peterson 2010). von Fintel and Gillies (2010) similarly argue that the English epistemic modal must encodes indirect evidential semantics. Epistemic modals cannot be used if there is direct evidence available to the speaker, as shown in (79) where the modal must is infelicitous if the speaker sees that it is raining.8 (79)  [Seeing pouring rain]  a It’s raining b ?? It must be raining  von Fintel and Gillies show that, although languages like English do not differentiate modals based on evidence source, they take into consideration the evidence source provided by the context. This provides support for the statement that many if not all epistemic modals also have an evidential restriction (Matthewson 2011).  3.3  Cross-linguistic variation  Given the theory of modality discussed in the previous section, modal systems can differ based on three separate dimensions: conversational background, evidential restriction, and modal force. This section provides data from English and two Pacific Northwest languages, St’´at’imcets and Gitksan, and focuses on how the three dimensions interact to create differences between the epistemic modals of the 8  The argument that the English modal must indicates evidence source emerges from a well known argument that given standard modal logic where modals quantify over possible worlds, a modalized sentence like mustφ makes a stronger statement than the basic proposition φ (Kratzer 1991, von Fintel and Gillies 2010). Consider example (i), where the non-modalized statement in (ia) would be considered weaker than the modalized statement in (ib) given standard modal logic. [i] (Kratzer 1991: 645) a She climbed Mount Toby. b She must have climbed Mount Toby. However, this is not the case. von Fintel and Gillies’s analysis claims that must φ does not entail φ (Kratzer 1991, von Fintel and Gillies 2010), rather, must remains a strong necessity modal but crucially signals an evidential restriction that states the speaker is basing their modal statement on indirect evidence. The apparent weak reading for must that Kratzer describes (1991) actually emerges from an evidential restriction rather than a force difference.  45  three different modal systems. The question to be answered in later chapters is how Nsyilxcen encodes the modal base, conversational background, and evidential restriction and where the Nsyilxcen modal system fits relative to other languages in the cross-linguistic modal typology. Details about the formal semantic analysis for these languages are presented and discussed in chapter 4. To begin, there is a basic distinction between an epistemic and non-epistemic conversational background. As shown in (80) English must and may are found in both epistemic and non-epistemic contexts. The distinction between the two conversational backgrounds is provided by the context. (80) John must be at home (In view of what I know) John may be at home (In view of what I know) John must be at home (In view of what the rules require) John may be at home (In view of what the rules require)  (Epistemic) (Epistemic) (Deontic) (Deontic)  This is opposed to St’´at’imcets, a Northern Interior Salish language, and Gitksan, a Tsimshianic language, which lexically encode the difference between an epistemic and non-epistemic conversational background. The epistemic modal k’a in St’´atimcets is permitted in epistemic contexts as in (81), and is infelicitous in nonepistemic contexts. (81)  w´a7=k’a s-t’al l=ti=tsitcw-s=a s=Philomena be=epis stat-stop in=det=house-3.sg.poss=exis nom=Philomena ‘Philomena must/ might be in her house’ (Matthewson 2010)  The non-epistemic modal ka is permitted in non-epistemic contexts as in (82), and is infelicitous in epistemic contexts (Matthewson et al. 2007, Rullmann et al. 2008, Davis et al. 2009). (82)  l´an=lhkacw-ka a´ ts’x-en ti=kwt´amts-sw=a already=2.sg.subj-deon see-dir det=husband-2.sg.poss=exis ‘You must/ can/ may see your husband tomorrow’ (Rullmann et al. 2008: 238)  Table 3.1 summarizes the lexical encoding of conversational background in St’´at’imcets and Gitksan, as discussed by Matthewson et al. (2007), Rullmann et al. (2008), Davis et al. (2009), Peterson (2010), and Matthewson (in press [1]). 46  Table 3.1: Cross-linguistic variation: conversational background Epistemic Non-epistemic St’´at’imcets ku7, k’a, -an’ ka Gitksan =ima, =kat da’ak(hl)xw, anook, sgi English must, might  Focusing on the epistemic modal systems of St’´at’imcets, Gitksan, and English shows that modals across languages can also differ with respect to evidence source. St’´at’imcets lexically encodes a difference between inference made from reports, and any previous knowledge the speaker has that is relevant to the situation. Matthewson et al. (2007) analysed three St’´at’imcets epistemic modals that lexically encode the source of evidence. The first is ku7, an epistemic reportative shown in (83). (83)  wa7 ku7 ku sts’´ets’qwaz’ l-ta stsw´aw’cw-a be report det trout in-det creek-det ‘[I heard] There are trout in the creek’ (Rullmann et al. 2008: 346)  The second is k’a, an indirect inferring evidential with no specific source. The speaker is basing their utterance on indirect inference based on both reasoning (84) as well as observed evidence (85) (Rullmann et al. 2008). (84)  Context: You have a headache that won’t go away, so you go to the doctor. All the tests show negative. There is nothing wrong so it must just be tension. nilh k’a lh(el)-(t)-en-s-w´a(7)-(a) ptinus-em-s´ut foc infer from-det-1.sg.poss-nom-impf-det think-mid-ooc ‘It must be from my worrying’ (Rullmann et al. 2008: 321)  (85)  Context: His car isn’t there plan k’a qwats´ats already infer leave ‘Maybe he’s already gone’ (Rullmann et al 2008: 321) 47  Finally, -an’ is an epistemic modal that is permitted in a subset of contexts where k’a is permitted. It is restricted to contexts where the speaker is making an inference based on the results of an action or event. In (86) the speaker is making an inference based on their knowledge about John. In this context - an’ in (86b) is semantically odd as there is no perceived evidence available to the speaker. (86)  Context: You had five pieces of ts’wan (wind-dried salmon) left when you checked yesterday. Today, you go to get some ts’wan to make soup and you notice they are all gone. You are not sure who took them, but you know that John is the person in your household who really loves ts’wan and usually eats lots whenever he gets the chance.  a. ts’aqw-an’´as k’a i ts’w´an-a eat-dir-3.erg infer det.pl wind-dried.salmon-det ‘John must have eaten the ts’wan’ (Rullmann et al. 2008: 347) b. ?? ts’aqw-an’-´as-an’ i ts’w´an-a kw s-John eat-dir-3.erg-perc.evid det.pl wind-dr.salmon-det det nom-john ‘John apparently ate the ts’wan’ (Rullmann et al. 2008: 347) However, when the context has observable evidence both k’a and an’ are permitted as seen in examples (87a) and (87b). (87)  Context: Same as in (86), except that it’s not just that you think he’s the one who likes ts’wan. This time, you see the ts’wan skins in his room.  a. ts’aqw-an’´as k’a i ts’w´an-a eat-dir-3.erg infer det.pl wind-dried.salmon-det ‘John must have eaten the ts’wan’ (Rullmann et al. 2008: 347) b. ts’aqw-an’-´as-an’ i ts’w´an-a kw s-John eat-dir-3.erg-perc.evid det.pl wind-dr.salmon-det det nom-john ‘John apparently ate the ts’wan’ (Rullmann et al. 2008: 347) Gitksan, a Tsimshianic language, also lexically encodes the source of evidence for the modal evidentials =ima, an indirect inferring evidential as shown in (88), and =kat, a reportative as shown in (89).9 ¯ 9  ´ Peterson also presents a non-modal evidential nakw, expressing inference based on physical evidence. Peterson gives the evidential n´ akw a non-modal analysis due to its syntactic and semantic differences compared to the modal evidential =ima. (Peterson 2010)  48  (88)  Context: You’re wondering what your friend is doing. You notice his rod and tackle box are not in their usual place. yugwimahl dim iixwt yukw=ima=hl tim iixw-t prog=mod=cnd fut fish-3 ‘He might/must be going fishing’ ‘He’s probably going fishing’ ‘He’s likely going fishing’ ‘He could be going fishing’ ‘Maybe/perhaps he’s going fishing’ (Peterson 2010: 27)  (89)  si-hon=kat=t John caus-fish=kat=pnd john ‘[I heard] John is doing up fish’ (Peterson 2010: 12)  These modal systems differ from the English system which does not usually encode different types of evidence source. In contrast to St’´a’timcets, when must or might are used epistemically, they allow for different kinds of indirect evidence with or without a specific source.10 This shows that English, St’´at’imcets and Gitksan differ with respect to how the conversational backgrounds are encoded in the language. Table 3.2 shows that the St’´at’imcets and Gitksan epistemic modals are lexically encoded for the type of modal base whereas the modal base for English epistemic must and might are contextually determined and do not explicitly distinguish different sources of evidence. Table 3.2: Cross-linguistic variation: evidence source Reportative St’at’imcets Gitksan English  ku7 =kat ¯  Indirect inferring evidential k’a  Inferring evidential from results -an’ =ima  must, might  10  This a simplified description of the English modal system, which is intended to show the basic distinction between different modal systems.  49  Languages also show variation with regards to modal force. In English, must is always a necessity modal taking universal quantification over possible worlds and might is a possibility modal taking existential quantification over possible worlds. This differs from the St’´at’imcets and Gitksan modals which are permitted in contexts that permit necessity and possibility interpretations (Matthewson et al. 2007, Rullmann et al. 2008, Davis et al. 2009, Peterson 2010). This difference is shown in table 3.3 where epistemic modals in St’´at’imcets and Gitksan are permitted in contexts that permit necessity and possibility interpretations. 11  St’at’imcets Gitksan English  Table 3.3: Cross-linguistic variation: modal force Universal Force (Necessity) Existential Force (Possibility) ku7, k’a, -an’ =kat,=ima ¯ must might  Comparative evidence from St’´at’imcets, Gitksan, and English shows that crosslinguistic differences between modal systems are due to how the modal space is organized based on how the modal force and conversational background are encoded (Rullmann et al. 2008). The English modal system shows variable conversational background and lexically encodes modal force. The St’´at’imcets and Gitksan modal systems lexically encode conversational background and have variable modal force interpretations. There is a third system examined by Vander Klok (2008); the Javanese modal system lexically encodes both modal force and conversational background. There is a specialized universal deontic modal kudu, two existential deontic modals ento and ol´eh, a universal epistemic modal mesthi, and an existential epistemic modal mungkin.  Deontic Epistemic  Table 3.4: Javanese modality Necessity Possibility kudu ento, ol´eh mesthi mungkin  Maria Am´elia Reis Silva’s (2010) examination of the Blackfoot modal system shows that Blackfoot modals are lexically encoded for both modal force and conversational background. Blackfoot makes a lexical distinction between necessity 11 Although St’´at’imcets and Gitksan modals can be used in both necessity and possibility contexts and crucially do not encode differences in force, the analyses by Matthewson et al. (2007), Rullmann et al. (2008), and Peterson (2010) lexically encode the modal force and have mechanisms in place to allow for the variable readings.  50  and possibility like Javanese, however there is also a distinction between weak and strong necessity. Reis Silva examines six modals. There is an epistemic strong necessity modal a´ a´ hk, an epistemic possibility modal a´ a´ hkam´a’p, as well as a third epistemic weak necessity modal a´ a´ hkam. This three way distinction in force is also present in the circumstantial modals. There is a circumstantial strong necessity modal sstsina’, a circumstantial possibility modal ohkott, as well as a circumstantial weak necessity modal a´ a´ kohk.  Circumstantial Epistemic  Table 3.5: Blackfoot modality Strong Necessity Weak Necessity a´ a´ kohk sstsina’ a´ a´ hk a´ a´ hkam  Possibility ohkott a´ a´ hkam´a’p  Lastly, Amy Rose Deal (2011) analyzes the Nez Perce circumstantial modal o’qa. Similar to St’´at’imcets and Gitksan modals, o’qa is restricted to a single modal base and is felicitous in contexts compatible with necessity or possibility readings. Deal argues for an encoded possibility modal force and analyses o’qa’s variable strength in terms of what she calls quantifiers without a scale. Deal’s analysis will be discussed in detail in section 4.1.3. To summarize the cross-linguistic variation just discussed, research has yielded a preliminary modal typology as shown in table 3.6. Table 3.6: Cross-linguistic modal typology 1 Variable Conversational Encoded Conversational Background Background Variable Force St’´at’imcets, Gitksan, Nez Perce- o’qa Encoded Force English Javanese Based on the empirical data presented in chapter 2, I argue that mat lexically encodes an epistemic conversational background and has variable modal force like the epistemic modals in St’´at’imcets and Gitksan. However, cmay also lexically encodes an epistemic conversational background but is restricted to existential modal force contexts, similar to the possibility epistemic modal ento in Javanese. This presents an interesting puzzle for the theoretical typology presented in table 3.6. Within the typology presented below, Nsyilxcen cross-cuts multiple systems within the same language and within the same (epistemic) domain.  51  Table 3.7: Cross-linguistic modal typology 2 Variable Conversational Encoded Conversational Background Background Variable Force St’t’imcets, Gitksan, Nez Perce- o’qa, Nsyilxcen- mat Encoded Force English Javanese, Nsyilxcen- cmay Section 4.1 will summarize in more detail the formal modal analyses presented by previous work on St’´at’imcets, Gitksan, and Nez Perce, and will discuss how they contribute to an analysis for the partially variable system seen in Nsyilxcen. Finally, Chapter 5 will discuss three potential formal analyses for the Nsyilxcen epistemic modals.  52  Chapter 4  Epistemic Modals: Comparative Semantic Analyses As discussed in section 3.3 epistemic modal systems across languages differ with respect to how they encode conversational background, evidential restrictions, and modal force. Based on the data presented in 3.3 and the Nsyilxcen data presented in chapter 2, table 5.5, shows where Nsyilxcen fits into the preliminary modal typology. Table 4.1: Cross-linguistic modal typology 3 Variable Conversational Encoded Conversational Background Background Variable Force St’t’imcets, Gitksan, Nez Perce- o’qa, Nsyilxcen- mat Encoded Force English Javanese, Nsyilxcen- cmay This chapter will discuss how previous semantic analyses for languages like St’´at’imcets, Gitksan, and Nez Perce, account for modal force variability. These analyses as well as the modal theory discussed in chapter 3 provide a basis for which the Nsyilxcen modals can be analysed.  4.1  Comparative models  This section will discuss semantic analyses for St’´at’imcets (Matthewson et al. 2007, Rullmann et al. 2008), Gitksan (Peterson 2010), and Nez Perce (Deal 2011) to better understand modals that are compatible with necessity and possibility interpretations. This is to determine what type of an analysis will best account for the Nsyilxcen data. There are three possible analyses that can account for this data, each of which make different empirical predictions.  53  As discussed in section 2, the Nsyilxcen epistemic modals lexically encode the modal base and specify the type of evidence used to make the modal statement. mat is compatible with both necessity and possibility translations and cmay is compatible with a possibility interpretation only. Working under the assumption that mat and cmay are modals operating at a propositional level, an analysis similar to the S’´at’imcets (Matthewson et al 2007, Rullmann et al 2008) or Gitksan (Peterson 2010) evidential modals can be used to account for the flexibility of mat. However, like English and Javanese modals, the analysis of cmay will need to need to account for its restricted possibility interpretation. Section 4.1.3 introduces a view of flexible modal force interpretations from Deal’s (2011) analysis for Nez Perce which looks at modals as quantifiers without a scalar implicature. It will discuss how Deal approaches a modal with flexible force interpretations and how this approach may not apply to Nsyilxcen.  4.1.1  St’´at’imcets modal analysis: Rullmann et al. 2008  Matthewson et al. (2007) and Rullmann et al. (2008) analysed three St’´at’imcets evidential modals, ku7, k’a and -an’. These modals lexically encode an epistemic conversational background and are flexible between necessity and possibility interpretations. This section will focus on presenting the indirect inferring evidential k’a in St’´at’imcets and the analysis accounting for a default universal modal force with weakening. The following contexts show that k’a encodes that the speaker is making an epistemic claim based on indirect inference (Matthewson et al. 2007, Rullmann et al. 2008). These examples also show the variability of modal force translations for k’a. (90)  Context: You have a headache that won’t go away, so you go to the doctor. All the tests show negative. There is nothing wrong so it must just be tension (Rullmann et al 2008: 321) nilh k’a lh(el)-(t)-en-s-w´a(7)-(a) ptinus-em-s´ut foc infer from-det-1.sg.poss-nom-impf-det think-mid-ooc ‘It must be from my worrying’  (91)  Context: His car isn’t there (Rullmann et al 2008: 321) plan k’a qwats´ats already infer leave ‘Maybe he’s already gone’ 54  The Rullmann et al. (2008) analysis proposes a fixed universal quantification over possible worlds for the St’´at’imcets modals. Although the modals are shown to allow for both necessity and possibility interpretations there is a preference for a universal interpretation. To account for the evidential restriction, lexically encoded conversational background, and universal modal force with weakening, Rullmann et al. (2008) adopt an analysis based on Klinedinst’s (2006, 2007) for English possibility modals, in combination with Reinhart’s choice function analysis for specific plural indefinites. The choice function f in a modal analysis will pick out a non-empty subset of possible worlds that are accessible from the actual world. (92)  Modal Choice Function (Rullmann et al. 2008: 337) A function f of type <st,st> is a modal choice function iff for any set of worlds W, f(W) ⊆ W and f(W) ∅  The evaluation of the St’´at’imcets evidential modals like k’a requires two parameters. B(w) is equivalent to f(w) discussed in chapter 3. The intersection of the set of propositions, f (w), yields the set of worlds in which the propositions in f (w) are all true. These worlds are considered to be accessible from w. A world v is epistemically accessible from w if everything the speaker knows in w is also true in v. The second parameter is the choice function f which picks out a non-empty subset of the worlds selected by the modal base, B(w). The modal then universally quantifies over the worlds in f (B(w)) (Rullmann et al 2008: 327). Variability in modal strength emerges from the size of the domain of quantification. This is similar to domain restrictions for quantification over individuals. Universal quantification over a set of individuals can be weakened if the domain of quantification is restricted. For instance, ‘everyone in the world likes hockey’. The universal quantifier selects the entire set of individuals. In this case it selects the very large set of individuals in the world.  55  Figure 4.1: Strong universal quantification  ‘Everyone in the world likes hockey’  Individuals in this room is a much smaller set of individuals. In this case, the universal quantifier selects a subset of the larger set of individuals in the world. Therefore, ‘Everyone in this room likes hockey’ is weaker than ‘Everyone in the world likes hockey’. Figure 4.2: Weak universal quantification  ‘Everyone in this room likes hockey’  The opposite occurs with existential quantification over individuals; the larger the domain of quantification the weaker the reading. Existential quantifiers select an individual from a set of individuals. In this case, the quantifier selects an individuals from the very large set of individuals in the world. This yield a weak reading.  56  Figure 4.3: Weak existential quantification  ‘Someone in the world likes hockey’  For a smaller subset of individuals in the world, like the people in this room, the existential quantifier selects an individual from that set. Existential quantification over a smaller set of individuals, like people in this room, will yield a stronger reading. As a result, ‘Someone in this room likes hockey’ is stronger than ‘Someone in the world likes hockey’. Figure 4.4: Strong existential quantification  ‘Someone in this room likes hockey’ The same weakening and strengthening effects of domain restriction arise when the quantification is over worlds rather than individuals. With an encoded universal modal force the flexibility between necessity and possibility force interpretations in St’´at’imcets results from the set of worlds selected by the choice function f. If the choice function is the identity function it picks out the entire set of worlds selected by the modal base. This results in universal quantification over the biggest possible subset of worlds yielding a necessity reading. On the other hand, if f selects a proper subset of worlds in B(w), universal quantification over a smaller set of worlds will result in a weaker possibility reading. In both cases the modal 57  will universally quantify over the set of worlds selected by the modal base and choice function. In addition to the modal base and choice function, there is also a presupposition specifying that the speaker was basing the modal statement on inferential evidence. The following is the formal semantic denotation for the k’a modal from Rullmann et al. (2008: 350). (93)  Semantics of k’a (inferential) [[ k’a7 ]]c,w is only defined if c provides an epistemic modal base B such that for all worlds w’, w’ ∈ B(w) iff the inferential evidence in w holds in w’. [[ k’a ]]c,w = λ f <st,st > . λp <s,t> . ∀w’[w’ ∈ f(B(w)) → p(w’)]]  Rullmann et al. (2008) mention in a footnote that using an ordering source provided by the standard modal framework is an alternative to a choice function analysis. Kratzer (2012), Peterson (2010), and Portner (2009) also discuss how an ordering source analysis would work for the St’´at’imcets modal system. This type of analysis will be discussed further in section 4.1.2.  4.1.2  Gitksan modal analysis: Peterson 2010  The Gitksan modal =ima is strictly an epistemic modal and is compatible with contexts that allow for necessity or possibility interpretation. It also encodes an evidential restriction that the speaker was basing the modal statement on indirect inferential evidence. Example (94) shows the variability of modal force translations for =ima (Peterson 2010). (94)  Context: You’re wondering what your friend is doing. You notice his rod and tackle box are not in their usual place. yugwimahl dim iixwt yukw=ima=hl tim iixw-t prog=mod=cnd fut fish-3 ‘He might/must be going fishing’ ‘He’s probably going fishing’ ‘He’s likely going fishing’ ‘He could be going fishing’ ‘Maybe/perhaps he’s going fishing’ (Peterson 2010: 27)  58  In this context, Peterson argues that there is no reason to assume that it is necessarily the case that your friend is fishing or that it is a mere possibility that your friend is fishing. However, the context can force either a necessity or possibility reading. Example (95) shows a context where a necessity translation is preferred over a possibility translation (Necessity > Possibility). Example (96) presents a context where a possibility translation is preferred over a necessity translation (Possibility > Necessity). (95)  Context: The speaker’s father was away frequently when she was a child. naa’ayima ’an yookxwinIJy ¯ naa’=ima ’an yookxw-in-IJy ¯ mother(informal)=mod s.rel eat-claus-1sg ”It must have been mother who fed/cooked for me.” >”It might have been mother who fed/cooked for me.” (Peterson 2010: 162)  (96)  Context: You see your uncle stopped at the intersection talking to some people through the window of his pickup. You and your friends don’t recognize the people. wilaayimas nibibIJy (IJ nidiit) wilaa-i=ima=s nibib-IJy (IJnidiit) know-tr-3=mod=pnd mother’s.brother-1sg (3pl) ‘My uncle might know them’ >‘My uncle must know them’ (Peterson 2010: 163)  Peterson also argues that =ima has a preferred possibility modal force reading. This is based on Gitksan speakers’ judgments for =ima when asked specifically about the strength as well as their preference for might-like out-of-the-blue translations (Peterson 2010: 171). Examples (97a) and (97b) also provide support for this claim as they show that Gitksan has a way of disambiguating =ima as an existential modal. There is a non-modal evidential in Gitksan, nIJakw, which is only permitted in a subset of the contexts where =ima is felicitous. Both are found in contexts with indirect observable evidence from the results of an action or event. In these contexts nIJakw receives a stronger necessity reading and =ima is restricted to the default possibility interpretation. Both n´ akw and =ima are felicitous in (97). Peterson claims that in contexts like (97) if the speaker chooses to use n´ akw over =ima 59  the speaker is making a stronger claim from the physical evidence available to him/her. When =ima and nakw are felicitous in the same context, because nakw is unambiguously a necessity modal claim the speakers will no longer translate =ima as a necessity modal (Peterson 2010: 168). (97)  Context: You and a friend are at the store and you see a dollar coin on the ground. It could be yours, your friend’s, or the person in front of you in line may’ve dropped it.  a. kwoodindimahl ´ daalat ´ kwoo-T-in-t=ima=hl daala-t drop-t-caus-3.sg=mod=cnd money-3.sg ‘S/he might’ve dropped his/her money’ (Peterson 2010: 168) b. n´ akwt kwoodindimahl ´ daalat ´ ´ nakw=t kwoo-T-in-t=ima=hl daala-t evid=3.sg drop-t-caus-3.sg=mod=cnd money-3.sg ‘S/he must’ve dropped his/her money’ (Peterson 2010: 168) To summarize the generalizations made by Peterson (2010), =ima is a general inferential evidential modal with variable modal force readings. When =ima is found in contexts where the non-modal evidential nIJakw is also permitted, it is restricted to its default possibility reading. As discussed in section 3.3, the Gitksan epistemic modal system, like the St’´at’imcets modal system, is the inverse of the English modal system. Peterson’s analysis assumes a fixed existential quantificational force for =ima which accounts for the preferred possibility modal force reading. Variability arises as a result of the ordering source, which Peterson assumes is contextually determined. The semantic denotation for =ima that Peterson proposes is shown below (Peterson 2010:180). (98)  [[=ima]]c,w is only defined if c provides a modal base B such that for all worlds w’ ∈ B(w), the inferential evidence in w holds in w’. If defined [[=ima]]c,w = λp. ∃w’ [w’ ∈Og(w) (B(w)) ∧ p(w’) =1] (Peterson 2010: 179)12  12  Peterson’s formal semantics introduces the functions B and O. The B is equivalent to f (w). Again, this is the intersection of the set of propositions that yields a set of worlds accessible from w in which the propositions in f(w) are all true. O is equivalent to the max operator that takes the partial ordering of worlds by the ordering source function g(w) and selects the best of those worlds.  60  Given that the analyses for St’´at’imcets and Gitksan argue for a fixed modal force, universal for St’´at’imcets and existential for Gitksan, the ordering source analysis will work in a similar way but will have the opposite result for the Gitksan modals. The following example shows how the modal base and ordering source interact to allow for the default possibility interpretation for =ima (Possibility >Necessity). (99)  yugwima=hl iixwis John yukw=ima=hl iixw-(t)=s John prog=mod=cnd fish-3=cnd John ”John might be fishing” >John must be fishing.” (Peterson 2010: 176)  B(w) Epistemic: { John’s rubber boots are missing; his truck is not in the driveway, it’s fishing season } g(w) Empty: ∅ The modal base B is lexically encoded to be epistemic and is a function from worlds to a set of propositions. The intersection of these propositions yields a set of possible worlds that are compatible with the speaker’s knowledge (Peterson 2010:181). The possibility interpretation of (99) arises because the ordering source is empty. An empty ordering source will have no propositions and will yield a larger set of worlds.13 Existential quantification over a larger set of worlds selected by the modal base yields a possibility interpretation. In this case the speaker does not believe that rubber boots are only used for fishing. Therefore Og(w) will also include worlds where John is hunting or berry picking as well as worlds where John is fishing. This results in there being fewer worlds where it is true that, if John’s boots are missing it means that John is fishing. This yields a weak modal force interpretation where it is only a slight possibility that ‘John is fishing’ in w (Peterson 2010: 177). The following example (100) gives Peterson’s modal base and ordering source for a necessity interpretation of =ima (Necessity >Possibility) (Peterson 2010). 13  Kratzer (1991) points out that the ordering source cannot be strictly empty otherwise we would get a strong reading where must φ would always entail φ. In actuality, as argued by Kratzer and assumed by many others, must φ is actually weaker than φ. In order to weaken the modal proposition the worlds quantified over are restricted by the ordering source propositions. However, it is crucial that those propositions cannot all be true in the actual world as this would still necessarily include the actual world resulting in the incorrect strong reading. In order to get the weaker reading, we must be able to quantify over a set of worlds that potentially does not include the actual world. See von Fintel and Gillies for an alternative view on the relative strengths of must φ and φ (2010).  61  (100)  yugwima=hl iixwis John yukw=ima=hl iixw-(t)=s John prog=mod=cnd fish-3=cnd John ”John must be fishing” >John might be fishing.” (Peterson 2010: 176) B(w) Epistemic: { John’s rubber boots are missing, his truck is not in the driveway, it’s fishing season } g(w) Stereotypical: { Rubber boots are used for fishing, rubber boots are not ideal for hunting, or berry picking}  The necessity interpretation for (100) arises if there is a non-empty ordering source containing propositions that place restrictions on the set of worlds to be existentially quantified over. If the ordering source is stereotypical, like in example (100), it will include propositions that express the normal course of events. For (100), it will include propositions such that rubber boots are used for fishing only and not for other activities. Worlds where John uses his boots solely for fishing will be ordered higher than worlds where John uses his rubber boots for berry picking or hunting. Furthermore, the stereotypical ordering source contains propositions that represent the normal course of events in w and are therefore assumed to be true. Therefore the actual world is probably one of the smaller set of worlds selected by the ordering source which implies a necessity reading; it is probably the case that John is fishing in the actual world. For (100), worlds where John uses his boots for fishing, are worlds in which John is fishing is true and the speaker believes w to be such a world. This yields a strong modal force interpretation, translated as must. As mentioned in section 4.1.1, an ordering source analysis can also account for the St’´at’imcets data. However, because the modal k’a is a universal quantifier the ordering source will have the opposite effect on the modal force readings from the Gitksan existential modals. Universal quantification over a smaller domain of possible worlds will result in a weaker reading. A non-empty ordering source will restrict the set of possible worlds that are universally quantified over. Crucially, the extra propositions that serve to restrict the set of possible worlds may be true or false in the actual world. To illustrate this, consider the epistemic example provided by Rullmann et al. (2008). (101)  Context: John played bingo last night and is spending lots of money today. t’cum k’a kw s-John win(mid) infer det nom-john ‘John must/may have won’ (Rullmann et al 2008: 339) 62  The modal base yields a set of worlds that are epistemically accessible from the evaluation world w based on the speaker’s knowledge. In this case, worlds that are epistemically accessible are those where the propositions that John played bingo last night and John is spending lots of money today are true (Rullmann et al. 2008:339). Next the ordering source will determine the size of the set of worlds to be universally quantified over. An empty ordering source will yield the largest possible set of worlds which will result in the stronger modal force reading. Conversely, consider a stereotypical ordering source that contains propositions where the normal course of events would be where responsible people only spend money when they have extra funds. In this case, only those worlds that are highly ranked by the stereotypical ordering source will be universally quantified over. Given that John played bingo last night and is spending money today, John must have won. This results in a weaker reading for k’a.14 The set of accessible worlds will be restricted even further to those worlds where 1) the evidence holds and 2) the evidence indicated that John won playing bingo last night. A non-empty ordering source will result in a weaker modal force reading, as universal quantification over a more restricted set of worlds will yield a weaker reading. (102) is Peterson’s lexical entry for the St’´at’imcets modal k’a (Peterson 2010: 184/185). (102)  [[k’a]]c,w is only defined if c provides a modal base B such that for all worlds w’ ∈ B(w), the inferential evidence in w holds in w’. If defined, [[k’a]]c,w = λg. λp. ∀w’ [w’ ∈ Og(w) (B(w)) → p(w’) = 1] B(w) = {The lottery draw was last night, John is spending money today, etc.} Doxastic: g(w) = {Responsible people only spend money when they have extra funds, etc.}  This shows that a unified ordering source analysis can also account for a system like St’´at’imcets. Deal (2011) presents an alternative approach for understanding modal systems that show flexibility between necessity and possibility interpretations. Based on the Nez Perce circumstantial modal o’qa, Deal develops an analysis inspired by predictions for quantifiers that lack scalar implicatures.  4.1.3  Nez Perce: modals without scales  In Nez Perce the modal o’qa is strictly encoded for a circumstantial modal base and has variable modal force translations. To account for o’qa, Deal draws a parallel between quantifiers over individuals and modals which quantify over possible 14  See discussion in Peterson (2010: 185).  63  worlds. When two elements form a scale and a speaker uses a quantifier like English some or might it gives rise to an implicature that there is an alternative, like English all or must (Deal 2010), that was not used and therefore believed by the speaker to be false. Quantifiers that lack a scale will not create a scalar implicature and certain predictions can be made about how they should behave in upward and non-upward entailing contexts. Nez Perce consultants provide both necessity and possibility translations for o’qa and also volunteer o’qa in contexts that allow for necessity or possibility interpretations in upward entailing contexts. (103)  Context: A friend is preparing for a camping trip. I am taking this person around my camping supplies and suggesting appropriate things. I hand them two blankets and say (Deal 2010: 4) ’in´ehne-no’qa ’ee kii lep´ıt c´ıickan take-modal you dem two blanket a) ‘You can take these two blankets’ b) ‘You should take these two blankets’  The modal force variability persists only in upward entailing contexts. In nonupward entailing contexts o’qa loses its variability and is only permitted with possibility interpretations. Deal gives data for o’qa in non-upward entailing contexts like scoping under negation, in the restriction of universals or in the antecedent of a conditional sentence. In all of these non-upward entailing environments o’qa only receives a possibility translation. In non-upward entailing contexts, an existential quantifier is stronger than a universal quantifier. For instance, the sentence ‘It is not the case that some students passed’ entails that it is not the case that every student passed. Since the negated existential quantifier is stronger this predicts that there will be no scalar implicature which will not result in flexible readings for o’qa in non-upward entailing contexts. Deal shows that these predictions hold for o’qa. The prediction is that o’qa will be rejected in possibly-not and not-necessary contexts. To begin, Deal shows that negation always scopes over the modal in NP constructions and therefore cannot generate a possibly-not reading. As a result, the only reading for o’qa in non-upward entailing is a not-possible reading and scopes under negation. Example (104) shows that o’qa is felicitous in a context that calls for not-possibly interpretation. (104)  Context: The referee is talking to an injured player  64  tam’´aalwit-wecet rule-reason ’´etke because  w´eet’u neg k’om´ay’c hurt  ’ee you ’ee you  xˆ eeleewi-y´o’qa play-modal wee-s-∅ ’´aatim be-p-pres arm  ‘According to the rules, you can’t play, because your arm in injured’ (Deal 2011:22) Next, example 105 shows that o’qa is rejected in contexts that call for not-necessary translations (Deal 2011: 22). The speaker’s comments for example (105) indicate that the correct interpretation for o’qa is not-possible. The desired reading in this example is not-necessary, but is infelicitous. Therefore o’qa is analyzed as a possibility modal. (105)  Context: You are explaining to someone who thinks they have to leave that they are not in fact required to do so. It’s not necessary for them to leave. #w´eet’u ’ee kiy-´o’qa not you go-modal Consultant: That’s a different conversation, not this one. You’re just saying ... ‘you can’t go’ (Deal 2011: 22)  For Nez Perce, Deal analyses the circumstantial modal o’qa as a modal without a scale and shows that it does behave as predicted in these environments. Since o’qa lacks a scale it has variable modal force translations. Nez Perce consultants provide both necessity and possibility translations for o’qa and also volunteer o’qa in contexts that allow for necessity or possibility interpretations in upward entailing contexts.15 Context: A friend is preparing for a camping trip. I am taking this person around my camping supplies and suggesting appropriate things. I hand them two blankets and say (Deal 2010: 4) (106) ’in´ehne-no’qa ’ee kii lep´ıt c´ıickan take-modal you dem two blanket a) ‘You can take these two blankets’ b) ‘You should take these two blankets’ 15  Modals with different conversational backgrounds cannot form a scale with one another. Deal shows that o’qa is strictly circumstantial and is not permitted in epistemic contexts. As a result, even though there exists a necessity epistemic modal in Nez Perce, it cannot act as a dual of o’qa.  65  There is however, a general preference for possibility translations for o’qa, similar to =ima in Gitksan. Furthermore, the modal force variability persists only in upward entailing contexts. In non-upward entailing contexts o’qa loses its variability and is only permitted with possibility interpretations. Deal compares the behaviour of o’qa to modal systems like English which do have scalar implicatures and modal systems like S’´at’imcets and Gitksan which also have modals without scales like Nez Perce. Deal provides a typology for the cross-linguistic variability in modal systems based on this semantic framework and makes predictions for their behaviour in non-downward-entailing and downward entailing contexts. This is provided in table 4.2 16 . Table 4.2: Three types of modal systems (Deal 2010: 34) ∀, ∃ ∃ ∀ Modals with scales Modals without scales Example English Nez Perce St’´at’imcets Gitksan Profile Inflexible modals Modals flexible in UE Modals flexible in DE contexts SI for ∃ in UE context; Barring other factors; Barring other factors: SI for ∀ in DE contexts modals inflexible in modals inflexible in DE contexts UE contexts. Inventory  This typology makes a basic distinction between modals with scales and modals without scales. Modals without scales can further distinguish between systems like Gitksan and Nez Perce with a preferred possibility modal reading and St’´at’imcets with a preferred necessity modal force reading. This analysis yields three way typology which makes certain predictions. The English modal system creates scalar implicatures with modals like must and might. Both modals, as predicted do not have variable modal force readings. might is strictly existential and must strictly universal. A scalar implicature can exist because both modals are of the same type, they can both be circumstantial or both be epistemic depending on the context. Scalar implicatures exist in upward entailing environments for possibility modals, for instance, ’He might like hockey’ implies ’it is not the case that he must like hockey’. Also a scalar implicature exists in non-upward entailing environments for universal modals, for instance ‘He doesn’t have to leave’ implies ’it is not the case that he can’t leave’. 16  SI = Scalar implicature, UE = non-downward-entailing, DE = Downward entailing  66  In St’´at’imcets, all modals prefer a necessity modal force interpretation but permit possiblity interpretations. For this type of system with a fixed universal modal force, Deal predicts that the inverse of the Nez Perce system should emerge. Specifically, in upward entailing contexts the modals should be inflexible and in nonupward entailing contexts the flexibility of the S’t´at’imcets modals should persist. However, contrary to what is predicted modal force variability persists in upward entailing contexts for St’´at’imcets modals. Deal explains that the choice function analysis set forth by Rullmann et al (2008) is a specialized mechanism that is necessary to account for the existence of variablity in upward entailing contexts. The only way to be certain that the predictions from a scalar implicature approach hold comes from non-upward entailing contexts. However, Rullmann et al (2008) report syntactic restrictions in St’´at’imcets that prevent properly testing the modals in non-upward entailing contexts. Also, if an ordering source analysis were to replace the choice function analysis this is no longer an additional mechanism used to account for the variability in an upward entailing context. Then the question arises of what would prevent o’qa from getting strengthened by the ordering source. The epistemic modal =ima in Gitksan is most similar in behaviour to o’qa. Deal’s scalar implicature approach could be applied to Gitksan as =ima is strictly epistemic, has variable modal force interpretations in upward entailing contexts, has a preference for an existential modal force, and does not have a dual. However, crucial data from non-upward entailing contexts is required to determine whether =ima will behave as Deal predicts. That is =ima is predicted to lack modal force variability and be restricted to a possibility force reading in non-upward entailing contexts. However, Matthewson (in press [2]) states that is also difficult to collect the relevant data in Gitksan. The Nsyilxcen epistemic modals present interesting empirical data for this approach. Because there is one variable modal and one non-variable modal, Nsyilxcen does not fit neatly into one single system. mat is similar to St’´at’imcets with variable force and cmay fits best in the English or Javanese systems with a strictly encoded possibility reading. This raises the question of whether Deal’s analysis will hold for Nsyilxcen. The following chapter will discuss three possible analytical approaches for the epistemic modals mat and cmay. The first will discuss encoding a universal modal force for mat and an existential modal force for cmay and the second two will discuss encoding an existential modal force for both mat and cmay with different approaches to deal with the partially variable system.  67  Chapter 5  Formal Semantic Analysis for the Nsyilxcen Epistemic Modals The empirical data shows that the Nsyilxcen epistemic modals strictly encode the modal base and have an indirect inferential evidential restriction. (107) provides the semantic formalism for these generalizations. (107)  Presupposition for Nsyilxcen modals: (Adapted from Rullmann et al. (2008) and Peterson (2010). [[Epistemic Modal]]c,w is only defined if c provides a modal base f such that for all worlds w’ ∈ f(w), the inferential evidence in w holds in w’.  Sections 5.1, 5.2, and 5.3 present three possible options to account for the modal force variability of mat and the strict existential modal force for cmay.  5.1  Option #1  The first option is to adopt a gradable modality approach. Gradable modals represent readings that extend beyond a simple distinction between logical possibility and necessity. As the Nsyilxcen data shows, mat is permitted in a wide range of contexts covering both possibility and necessity interpretations and cmay has a strict possibility interpretation. Kratzer (1981) discusses the notion of gradable modality using German as a case study and provides the following example to illustrate. 17 (108)  17  Context: Much-Girgl has been murdered on his way home. The police start investigations. Certain conclusions may be drawn from what is known about the circumstances of the crime. Utterances of the following sentences are likely to have occurred in such a situation (Kratzer 2010: 9):  For a more recent discussion of gradable modality see Lassiter (2010)  68  a. Der Kastenjakl kann der m¨order sein the kastenjakl can the murderer be ‘Kastenjakl may be the murderer’ b. Der Gauzner-Michl muB der M¨order sein the gauzner-michl must the murderer be ‘Gauzner-Michl must be the murderer’ (108a) and (108b) are two possible answers expressing the police inspector’s certainty given the evidence available to him. Given that the inspector does not know every detail about the real world he must draw conclusions from the available evidence. The evidence available to the inspector is compatible with a set of worlds selected by an epistemic modal base. The stereotypical ordering source then ranks these epistemically accessible worlds relative to what the normal course of events would be. Given an epistemic modal base and stereotypical conversational background it is possibly the case that Kastenjakl is the murderer or necessarily the case that Gauzner-Michl is the murderer in the world w given the modal base f and the ordering source g provided by the context. However, in languages like German and English, there are many more grades of modality that create a scale with possibility on one end and necessity on the other. In the murder example, as new evidence becomes available during the investigations the inspector can reach different conclusions of gradable strength based on what he knows. Examples (109)(112) show that German has many different ways of expressing grades of modal strength other than necessity and possibility. (109)  Es kann gut sein, daB der Gauzner-Michl der M¨order war it can well be that the gauzner-michl the murderer was ‘There is a good possibility that Gauzner-Michl was the murderer’ (Kratzer 2010: 16)  (110)  Es there der the  besteht aber immer noch eine geringe M¨oglichkeit daB is however still a slight possibility that Kastenjakl der M¨order war Kastenjakl the murderer was  ‘There is, however, still a slight possibility that Kastenjakl was the murderer’ (Kratzer 2010: 16)  69  (111)  Der Gauzner-Michl kann eher der M¨order sein als der Kastenjakl the gauzner-michl can rather the murderer be than the kastenjakl ‘Gauzner-Michl is more likely the murderer than Kastenjakl’ (Kratzer 2010: 16)  (112)  Es ist wahrscheinlich, daB der Gauzner-Michl der M¨order war it is proabable that the gauzner-michl the murderer was ‘It is probable that Gauzner-Michl was the murderer’ (Kratzer 2010: 16)  English equally has ways of expressing these four grades of modal expressions (Kratzer 1981: 299). Table 5.1: Kratzer gradable modality (1981) English There is a good possibility that... There is a slight possibility that ... It is more likely that ... than that... It is probable that...  German Es kann gut sein daB...  Modal expression Human possibility  Es besteht eine geringe M¨oglichkeit daB... Es kann eher sein daB... als daB... Es ist wahrscheinlich daB...  Slight possibility Comparative possibility Human necessity  The two conversational backgrounds provide the necessary tools to achieve these different grades of modality. Based on the generalizations presented for Nsyilxcen, given that cmay is restricted to possibility contexts it would cover the weakest possibility. mat on the other hand covers a large range of modal strength readings that would be a reading stronger than possibility up to necessity.  Table 5.2: Nsyilxcen particles in circumstantial contexts Modal Expression Nsyilxcen Slight possibility cmay Human/Good possibility mat Human necessity  70  For mat, Kratzer’s definition for good possibility could be adopted. (113) Good Possibility (Kratzer 1991: 644) A proposition p is a good possibility in a world w with respect to a modal base f and an ordering source g iff there is a world u ∈ f(w) such that for all v ∈ f(w): if v ≥g(w) u, then v ∈ p. If mat and cmay occupied different positions on the possibility/ necessity scale, a scalar implicature emerges. If the speaker chooses to use cmay to make the modal statement, this implies that there is a reason they did not choose the stronger modal mat. This approach does makes certain predictions with respect to what types of contexts mat and cmay are compatible with. It would predict that mat would not be permitted in the weakest contexts where cmay is permitted. To determine whether this is the case the data would need to show a weak context where mat is infelicitous. This is something which has not yet emerged during elicitations as mat appears to always be permitted in the same contexts where cmay is permitted. For instance, during elicitations the running example was presented to the consultants with extra emphasis that the speaker is completely uncertain about Mary’s whereabouts. It was emphasized that she doesn’t run at the same time everyday, she doesn’t run the same days each week, she is in school everyday and likes to go shopping at the store when it is open. The intention was to create the weakest context possible. In this case both mat and cmay were permitted. (114)  Context: You know that Mary loves to go running and often goes on runs randomly. She could also be at the store or at school. I ask you, where is Mary?  a. Mary mat ac-q´ıc-@lx Mary epis.mod cust-run-lex ‘Mary might be running’ b. Mary cmay ac-q´ıc-@lx Mary epis.mod cust-run-lex ‘Mary might be running’ Data such as (33) cast doubt on the idea that mat might be a good possibility modal. 71  5.2  Option #2  For each Douglas Lake consultant there was no consistent default reading for mat in out-of-the-blue contexts. As a result an alternative analysis is one where mat and cmay both have a strictly encoded existential force. To account for the modal force variability for mat and the strict modal force interpretations for cmay, von Fintel and Iatridou’s (2006) analysis for must/ have to and should/ought to can be adapted. In this analysis mat would be sensitive to an ordering source allowing for flexible modal force interpretations and cmay would not be sensitive to the ordering source. To begin, von Fintel and Iatridou (2006) explore the intuition that should/ ought to are weaker than must/ have to. This can be seen in (115) which shows that there is no contradiction when (115a) is uttered. Therefore, ought to is weaker than have to (von Fintel and Iatridou 2006). 18 (115) a.  You ought to do the dishes, but you don’t have to  b. # You have to do the dishes, but you don’t have to (contradiction) (116) also shows that ought to is weaker than have to as a stronger modal will entail the weaker reading in (116a) and not the other way around in (116b) (von Fintel and Iatridou 2006: 2). (116) a.  You ought to wash your hands- in fact you have to  b. # You have to wash your hands - in fact you ought to von Fintel and Iatridou use Kratzer’s standard theory of modality as a framework to achieve the distinction between a weak necessity reading for should/ ought to and the necessity reading for must/ have to. (117) a. You ought to do the dishes b. You have to do the dishes 18 These sentences were attempted in Nsyilxcen, however as mat is permitted in the same contexts as cmay it was difficult to test for these judgments.  72  The modal base worlds are worlds where the goal is achieved, and von Fintel and Iatridou argue that should/ought to and must/ have to make a distinction between good and the very best worlds amongst the modal base worlds. As these modals are universal quantifiers, the larger the set of accessible worlds the stronger the modal force reading. For a strong necessity modal like must/have to the statement in (117b) says that the prejacent is true in all worlds selected by the modal base and primary ordering source. For a weak necessity modal like should/ ought to, a secondary ordering source adds an additional restriction to the modal base worlds. This means that the prejacent is asserted to be true in an even smaller set of worlds. In this case all of the very best worlds among the accessible worlds are worlds where the prejacent is true. The ordering source propositions in the secondary ordering source have the potential to weaken the force of should/ ought to down to a possibility reading as shown in figure 5.3. However, von Fintel and Iatridou only consider these modals to cover weak necessity interpretations. Table 5.3: Necessity and weak necessity Necessity Weak necessity  must should ought to  Possibility Similar to should/ ought to, mat is permitted in both necessity and possibility contexts it and will be sensitive to the ordering source to allow for its modal force flexibility. Consider the necessity example repeated below in (118). (118)  Context: Mary runs every day to train for a marathon. She usually runs at 6pm on Tuesdays. Today is Tuesday and it’s 6pm. I ask you, where is Mary? Mary mat ac-q´ıc-@lx Mary epis.mod cust-run-lex ‘Mary must have gone running’ B(w) Epistemic: { Mary is not at home, Mary trains every Tuesday at 6pm, It is Tuesday at 6pm, Mary has a strict training schedule, etc. } O(w) Doxastic = { The most important race of Mary’s life is on the weekend, etc. } 73  In this case, the doxastic ordering source would contain propositions that the speaker believes are true. As a result, the modal is quantifying over all the worlds where the evidence holds and where the propositions in the ordering source also hold. As previously discussed, existential quantification over a more restricted set of worlds will yield a stronger reading. In order to allow for the possibility interpretations for mat an empty ordering source will not place a secondary restriction on the possible worlds yielding a larger domain of quantification. As a result existential quantification over a non-restricted set of worlds will yield a weaker reading. My proposal for Nsyilxcen is the inverse of the von Fintel and Iatridou (2006) analysis. While they propose a weakening via a secondary ordering source, I propose an analysis involving strengthening of a possibility modal via an empty ordering source. Next, as discussed throughout this thesis, cmay is restricted to possibility contexts. Similar to must/ have to, the present option will require an additional mechanism that does not allow cmay to be sensitive to the ordering source. Therefore, assuming that cmay has an existential modal force it will only quantify over the nonrestricted set of modal base worlds. (119)  Context: You know that Mary loves to go running and often goes on runs randomly. She could also be at the store or at school. I ask you, where is Mary? Mary cmay ac-qic-lx Mary epis.mod cust-run-lex ’Mary might be running’ B(w) Epistemic: { Mary is not at home, It is Tuesday at 6pm, Mary doesn’t have a strict training schedule, etc. } O(w): EMPTY  This analysis accounts for the possibility readings for both mat and cmay and also allows mat to covers a wide range of readings from pure possibility up to the strongest necessity. This is illustrated in figure 5.4.  74  Table 5.4: Possibility and strong possibility Necessity  mat  Strong Possibility Possibility  mat cmay  The lexical entry with an encoded existential modal force for mat is provided in (120) based on Peterson’s analysis for =ima (Peterson 2010: 179). (120)  [[mat]]c,w is only defined if c provides a modal base f such that for all worlds w’ ∈ f(w), the inferential evidence in w holds in w’. If defined [[mat]]c,w = λp. ∃w’ [w’ ∈g(w) ( f (w)) ∧ p(w’) =1]  The same lexical entry can be applied to cmay and is provided in (121). (121)  [[cmay]]c,w is only defined if c provides a modal base B such that for all worlds w’ ∈ B(w), the inferential evidence in w holds in w’. If defined [[cmay]]c,w = λp. ∃w’ [w’ ∈O(w) (B(w)) ∧ p(w’) =1]  For both Nsyilxcen and English, the ordering source affects the size of the domain of quantification depending on how many propositions it contains. Assuming that mat has a default existential modal force the more propositions in the ordering source, the smaller the domain of quantification and the stronger the modal force reading. Given that mat allows for such a wide range of modal force readings it is sensitive to the ordering source. This is similar to should/ ought to except that the English modals have a universal modal force and therefore the more restricted the domain of quantification the weaker the reading. cmay on the other hand is similar to English must/ have to as it is not sensitive to a non-empty ordering source. In this case cmay will take existential quantification over a non-restricted set of modal base worlds yielding a weaker reading. As von Fintel and Iatridou state for must/ have to, it is necessary to lexically encode that cmay is not sensitive to a non-empty ordering source. 75  However the Nsyilxcen epistemic modals are not identical to English must/ have to and should/ought to. According to von Fintel and Iatridou, should/ ought to are always weaker than must/have to and are always subject to a non-empty secondary ordering source (2008). Whereas in Nsyilxcen, mat is felicitous in the same pure possibility contexts where cmay is felicitous. In fact, there is no data available to show that mat is not permitted in possibility contexts where cmay is felicitous. Unlike English must/ have to and should/ ought to, the Nsyilxcen data does not differentiate between a possibility and a strong possibility interpretation. Rather an analysis that makes use of the ordering source can account for the variable force interpretations for mat and the strict force interpretation for cmay.  5.3  Option #3  The third approach is to strictly encode a universal modal force for mat - with an additional mechanism to weaken it to an existential modal force - and to strictly encode an existential modal force for cmay with an added stipulation that it cannot be strengthened to a universal modal force reading. Assuming mat is a universal modal over possible worlds will require an analysis like the ordering source analysis for the St’´at’imcets system presented in section 4.1.1. Consider the following necessity context (122). (122)  Context: Mary runs every day to train for a marathon. She usually runs at 6pm on Tuesdays. Today is Tuesday and it’s 6pm. I ask you, where is Mary? Mary mat ac-q´ıc-@lx Mary epis.mod cust-run-lex ‘Mary must be running’ f (w) Epistemic: { Mary is not at home, Mary trains every Tuesday at 6pm, It is Tuesday at 6pm, Mary has a strict training schedule, etc. }  For (122) the epistemic modal base consists of propositions supplying what is known from the evidence available to the speaker. In this case, universal quantification over a non-restricted set of worlds will yield the stronger necessity reading. As previously stated, the ordering source will account for the variability of the necessity and possibility readings for mat. To get the necessity reading, an empty ordering source will not contain propositions that further restrict the modal base worlds.On the other hand, a non-empty ordering source will pick out a smaller set 76  of worlds and universal quantification over the smaller set will result in a weaker reading. Consider example (123). (123)  Context: You know that Mary loves to go running and often goes on runs randomly. She could also be at the store or at school. I ask you, where is Mary? Mary mat ac-q´ıc-@lx Mary epis.mod cust-run-lex ’Mary might be running’ f (w) Epistemic: { Mary is not at home, etc. } g(w) Doxastic = { Mary attends classes during the week, Mary likes to go shopping }  In this case the doxastic ordering source will consist of propositions that may or may not be true in the actual world. Universal quantification over a restricted set of accessible worlds will yield the weaker possibility reading. The lexical entry for mat under this analysis is provided in (124). (124)  Lexical Entry for mat (Adapted from Rullmann et al. 2008) || mat ||c,w is only defined if c provides a modal base f such that for all worlds w’ ∈ f(w), the inferential evidence in w holds in w’. If defined || mat ||c,w = λp. ∀w’ [w’ ∈ g(w) ( f (w)) → p(w’) =1]  (124) can be paraphrased as “For all worlds w’ if w’ is a member of the set of worlds selected by the modal base and ordering source then p is true in w’.” Next, as presented in chapter 2 cmay is restricted to possibility contexts and is infelicitous in necessity contexts as seen in (125) and (126). (125)  Context: You know that Mary loves to go running and often goes on runs randomly. She could also be at the store or at school. I ask you, where is Mary? Mary cmay ac-q´ıc-@lx Mary epis.mod cust-run-lex ’Mary might have gone running’ 77  (126)  Context: Mary runs every day to train for a marathon. She usually runs at 6pm on Tuesdays. Today is Tuesday and it’s 6pm. I ask you, where is Mary? # Mary cmay ac-q´ıc-@lx Mary epis.mod cust-run-lex ’Mary must have gone running’  Like mat, the epistemic modal base for cmay consists of propositions supplying what is known based on evidence provided by the context. f (w) Epistemic: { Mary is not at home, Mary is training for a race, Mary does not have a strict training schedule, etc. } I propose that cmay lexically encodes an existential modal force. The following is the lexical entry for cmay. (127)  Lexical Entry for cmay (Adapted from Rullmann et al. 2008 and Peterson 2010) || cmay ||c,w is only defined if c provides a modal base f such that for all worlds w’ ∈ f(w), the inferential evidence in w holds in w’. If defined || cmay ||c,w = λp. ∃w’ [w’ ∈ g(w) (f (w)) ∧ p(w’) =1]  (127) can be paraphrased as “There is a world w’ such that w’ is accessible from the set of worlds selected by the modal base and ordering source and the proposition p is true in w’ ”. The speaker’s doxastic ordering source may be non-empty, in which case cmay would take existential quantification over a smaller subset of modal base worlds. g(w) Doxastic = { Mary likes to go shopping, Mary is taking classes during the week, Mary’s most important race of her life is this weekend, etc. } However, this presents a problem. If cmay is an existential quantifier over a smaller set of worlds it will yield a stronger reading. As we cannot prevent the speaker from believing certain propositions we cannot prevent cmay from being strengthened. Therefore, the lexical entry needs to lexically encode that cmay cannot be strengthened. 19 19  It should be noted that most modal systems will need this type of lexical restriction. For instance, English should or ought as discussed by von Fintel and Iatridou (2006) needs a secondary ordering source to allow for the weak necessity reading. The stronger modal must cannot be weakened via a secondary ordering source.  78  This analysis argues for a default universal force for mat and can be backed up with examples showing that there is an overt way of disambiguating in favor of an existential interpretation. Rullmann et al. (2008) provide examples showing that St’´at’imcets also has a way of disambiguating towards an existential interpretation for the inferential modal k’a. k’a allows for both necessity and possibility readings but has a preferred necessity interpretation. Rullmann et al. (2008) show that when k’a is found along side the existential modal sxek ‘maybe’ it yields an existential interpretation as seen in (128). (128)  nilh ku foc det sxek maybe  cw7aoy-s kw s-k’´uIJl-em m´uta7 ku p´ala7 k’a neg-3poss det nom-make-mid again det one infer xetsp´asq’et week  ‘So she wouldn’t have to make more for about a week maybe’ (Matthewson 2005: 147-148) Nsyilxcen also has a way of overtly disambiguating between a necessity and possibility interpretation for mat. As this section has argued, cmay has a strictly encoded existential modal force and when mat and cmay are in the same utterance, there is only a possibility interpretation. (129) mat cmay sl’ t spiPsc’´ılt epis.mod epis.mod lost obl yesterday ‘Maybe he must have gotten lost yesterday’ There are other options to account for the variable modal force for mat and the strictly encoded existential modal force for cmay.  5.4  Implications and further research  To account for the Nsyilxcen data, I propose option # 3 is the best analysis. The ordering source analysis is able to account for the modal force variability of mat and the strict modal force of cmay. However, whether mat has an existential modal force (option # 2) or a universal modal force (option # 3) requires more research. From the empirical data it is not yet clear as to whether mat has a default necessity or possibility interpretation. Even though mat does not have a preference for a possibility or necessity reading in out-of-the-blue contexts, there is data indicating that it has a universal modal force interpretation (see section 5.3). Further research and data is required to clearly determine the consultants’ judgments for mat. For 79  instance, data involving non-upward entailing contexts would aid in distinguishing between option # 2 and option # 3, however attempts at eliciting this data proved difficult and did not yield clear results. Regardless, I propose that option # 3 is the best option as it accounts for the variable modal force of cmay and the strictly encoded possibility modal force for cmay. The residual problem with using an ordering source is that there needs to be an extra mechanism that prevents cmay from being sensitive to the effects of the ordering source. For instance if mat and cmay have a lexically encoded existential modal force there will need to be an explanation for why mat is strengthened and cmay is not. This is also a problem for option # 3 where the ordering source can weaken the universal force for mat but does not strengthen the existential modal force for cmay. This is a detail that needs to be accounted for in English as well. von Fintel and Iatridou’s (2006) analysis for must/ have to and should/ ought to must also specify that must/ have to is not sensitive to a secondary ordering source. There are also many interesting questions that were beyond the scope of this thesis. First, an in depth syntactic analysis for the modal system. Secondly, developing a semantic analysis for the reportative kw ukw and the bouletic modal cakw . The theory of modality as developed by Kratzer (1977, 1981, 1997, 2012) makes predictions about what modal systems can look like. Modal theory along with the methodology of collecting translation and judgment tasks, has revealed both similarities and differences between the modal systems of languages like Nsyilxcen, St’´at’imcets, Gitksan, Blackfoot, Javanese, Nez Perce, and English. The partial variability of the Nsyilxcen modal system reveals interesting implications for a cross-linguistic modal typology. Table 5.5 shows that the Nsyilxcen epistemic modals do not fit in with one particular system.  Variable Force  Encoded Force  Table 5.5: Cross-linguistic modal typology Variable Conversational Encoded Conversational Background Background St’´at’imcets, Gitksan, Nez Perce- o’qa, Nsyilxcen- mat English Javanese, Nsyilxcen- cmay  80  In fact, when entire modal systems are considered, none of these languages can be classified based on a simple dichotomy between modal force and conversational background. This fact has also been discussed by Rullmann et al. (2008), Peterson (2010), and Deal (2011). For instance, the empirical data shows that circumstantial and epistemic modals behave in vastly different ways across systems. Matthewson (in press [2]) showed that other languages like St’´at’imcets and Gitksan do not fit neatly into a simplified modal typology. Research has revealed that each language covers the modal space in different ways. This is shown in tables 5.6 and 5.7. Table 5.6: St’´at’imcets modal system (Matthewson in press [2]: 9) Possibility Necessity Circumstantial Deontic ka Reportative ku7 Epistemic Indirect Inferential k’a Inference from results -an’  Table 5.7: Gitksan modal system (Matthewson in press [2]: 9) Possibility (Weak) Necessity Plain da’ak(hl)xw sgi Circumstantial Deontic anook Plain ima(’a) Epistemic Reportative gat The Nsyilxcen system also expresses possibility and necessity modal force and a circumstantial and epistemic modal base but in a different way from St’´at’imcets and Gitksan.  Circumstantial Epistemic  Table 5.8: Nsyilxcen modal system Possibility Necessity w Bouletic cak Plain mat cmay Reportative kw ukw  Finally, the English modals cover a different semantic space from Pacific Northwest languages. 81  Circumstantial Epistemic  Table 5.9: English modal system Possibility Weak Necessity Plain can Deontic may should can Plain might  Necessity must  Cross-linguistic comparison of modal systems reveals the extent to which languages choose to divide up the semantic modal space. Gaps in the lexicon not only reveal that modal systems should not be compared in their entirety, they also provide interesting insight into what formal notions languages can do without. Deal’s (2011) discussion of Nez Perce modality calls into question the Effability Hypothesis, which in its strongest form states all languages will have a way to express any proposition (Keenan 1978, Katz 1976). In reality, as Deal showed with Nez Perce, languages are not required to express every type of notional or formal category (2011:36). Matthewson (in press [2]) provides further support for this argument with data from multiple modal systems like S’´at’imcets, Gitksan, Blackfoot, and Javanese. The Nsyilxcen non-epistemic modal system also provides empirical evidence supporting this conclusion. Similar to how Nez Perce speakers paraphrase to express notions of circumstantial necessity, Nsyilxcen speakers also paraphrase to express certain types of circumstantial modality. Matthewson (in press [2]) discusses that using theory as a basis for elicitations will provide a fuller and more accurate empirical picture of the extent and nature of diversity. What theoretically informed semantic fieldwork has revealed is that languages can divide the modal space in many different ways and the more languages that are involved in typological research the more diversity that is uncovered. Due to this diversity most modal systems cannot be classified in their entirety based on a simple dichotomy between conversational background and modal force. This thesis provides further support for this claim and the Nsyilxcen data also indicates that even parts of modal systems, like the epistemic modality, may not be directly compared as a whole across languages.  82  Chapter 6  Conclusion The goal of this thesis was to document both epistemic and non-epistemic modality in the Nsyilxcen language with a special focus on providing a semantic analysis for the epistemic modals mat and cmay. My fieldwork on Nsyilxcen has looked at four modal particles. Two epistemic modals mat and cmay, the reportative kw ukw , and one circumstantial modal cakw . The empirical data shows that mat and cmay are both strictly epistemic modals and are not felicitous in non-epistemic contexts. Furthermore these modals specify that indirect inferential evidence was used to make the modal statement. Where these two modals differ is with respect to their modal force, where cmay is felicitous in a subset of the contexts where mat is felicitous. Next, kw ukw is an epistemic modal and expresses that the source of evidence comes from a report. Finally there is one non-epistemic modal cakw that is felicitous in bouletic contexts. To express notions of circumstantial, deontic, teleological, and ability non-epistemic modality the Nsyilxcen speakers paraphrase using a variety of different particles. This thesis also proposes a formal semantic analysis for mat and cmay which makes use of an ordering source. The formal analysis discussed in this thesis is based on Kratzer (1977, 1981, 1991, 2010), Matthewson et al. (2007), Rullmann et al. (2008), and Peterson (2010). The variability of the modal force for mat arises due to a contextually determined ordering source that either contains propositions which restrict the domain of quantification or is empty and does not affect the domain of quantification. I propose in section 5.3 that mat lexically encodes a universal modal force. To account for its variable modal force interpretations, I also propose that mat is sensitive to the ordering source. As a result, if there is a non-empty ordering source then the domain of quantification will be restricted. Universal quantification over a smaller set of worlds yields a weaker reading. If the ordering source is empty, universal quantification over a non-restricted set of worlds will yield a stronger reading. As cmay does not have variable modal force interpretations, it will not be sensitive to the ordering source and will always take existential quantification over the largest set of possible worlds selected by the modal base. Existential quantification over a non-restricted set of worlds will yield a weaker possibility-like interpretation. 83  The overall implications of this thesis provides further support for the discussion that all languages will have a way to express all types of modality making crosslinguistic comparison impossible. Empirical data from multiple languages point towards more diversity than similarities. Languages cannot be lumped into simple dichotomy between strength and force.  84  Chapter 7  References Baptiste, Maxine. 2002. WH-Questions in Okanagan Salish. M.A. thesis, University of British Columbia. Cinque, Guglielmo.1999. Adverbs and Functional Heads. Oxford: OUP. Deal, Amy Rose. 2011. Modals without scales. Language. 87:3, 559-585. von Fintel, Kai. 2006. Modality and Language. In Encyclopedia of PhilosophySecond Edition, Ed. by Donald M. Borchert. Detroit: MacMillan Reference USA. http://mit.edu/fintel/www/modality.pdf. von Fintel, Kai, and Anthony Gillies. 2010. Must...Stay...Strong!. Natural Language Semantics.18:4. 351-383. von Fintel, Kai and Irene Heim. 2011. Intensional Semantics. MIT lecture notes. http://stellar.mit.edu/S/course/24/sp11/24.973 von Fintel, Kai, and Sabine Iatridou. 2006. How to Say Ought in Foreign: The Composition of Weak Necessity Modals. Hebert, Yvonne M. 1978. Sandhi in Salishan Language: Okanagan. ICSNL 13:26-56. Victoria, B.C. —– 1979. A note on Aspect in (Nicola Lake) Okanagan. ICSNL 14: 173209. Bellingham, Washington. —– 1982. Aspect and Transitivity in (Nicola Lake) Okanagan. Syntax and Semantics. 15: 295-215. Izvorski, Roumyana. 1997. The present perfect as an epistemic modal. In Proceedings of SALT VII. 222-239. Klinedinst, N. 2006. Plurality and possibility. Ph.D. dissertation, UCLA. Kilnedinst, N. 2007. Plurals, possibilities, and conjunctive disjunction. In UCL Working Papers in Linguistics 19, ed. R. Breheny and N. Velegrakis, 85  261-284. Department of Phonetics and Linguistics, University College London. Kratzer, Angelika. 1977. What “must” and “can” must and can mean”. Linguistics and Philosophy. 1: 337-355. —–1981. The notioal category of modality. words, worlds, and contexts ed. by H.-J. Eikemeyer & H. Rieser, 38-74. Berlin: de Gruyter. —– 1991. Modality. Semantics: An international Handbook of Contemporary Research ed. by Dieter Wunderlich & Amim von Stechow, 649-650. Berlin: de Gruyter. —– 2012. Modals and Conditionals: New and Revised Perspectives. Oxford: OUP. Kroeber, 1999. The Salish Language Family: Reconstructing Syntax. Lincoln: The University of Nebraska Press. Lewis, M. Paul (ed.), 2009. Ethnologue: Languages of the World, Sixteenth edition. Dallas, Tex.: SIL International. Online version: http://www.ethnologue.com/. Lyon, John. 2009. Okanagan demonstratives and their uses in story: Some notes on their frequency, distribution, and function. ICSNL http://www.linguistics.ubc.ca/sites/default/files/2009 ICSNL.pdf. —– 2010. Nominal modification in Upper Nicola Okanagan: A working paper. ICSNL http://www.linguistics.ubc.ca/sites/default/files/2010 ICSNL.pdf. —– 2011.The Semantics of determiner phrases in Okanagan. ICSNL. http://www.linguistics.ubc.ca/sites/default/files/2011 ICSNL.pdf. —– 2012. Handout. Salish Reading Group. UBC. Matthewson, Lisa, Hotze Rullmann and Henry Davis. 2007. Evidentials as Epistemic Modals: Evidence from St’´at’imcets. The Linguistic Variation Yearbook.7. http://faculty.arts.ubc.ca/lmatthewson/pdf/evidentiality.pdf. —– 2004. On the Methodology of Semantic Fieldwork. International Journal of American Linguistics. 70: 369-415. —– in press [1]. When fieldwork meets theory: Evidence About Evidentials. To appear in Proceedings of Linguistic Evidence, ed. Britta Stolterfoht and Sam Featherston.  86  —– in press [2]. On How (Not) to Uncover Cross-Linguistic Variation. To appear in Proceedings of NELS 42. Mattina, Anthony. 1973. Colville Grammatical Structure. Honolulu. Ph.D. dissertation, University of Hawaii, Honolulu. DAI34A:3375. [Published at UHWPL 5.] —– 1974. The Narrative of Peter J. Seymour: Bluejay and his Brother-inlaw Wolf. ICSL. 9, Vancouver, B.C. —– 1978a. Parallels between the Colville Transitives and Pseudo-Intransitives. Of the Proceedings of the 8th Annual Meeting of WECOL. 104-109. —– 1980. Imperative Formations in Colville-Okanagan and in the Other Interior Languages. ICSL 15: 208-228. —– 1982. The Colville-Okanagan Transitive Systerm. IJAL 48: 421-435. —– 1985. The Golden Woman: The Colville Narrative of Peter J. Seymour. Tuscon: The University of Arizona Press. —– 1987a. Colville-Okanagan Dictionary. UMOPL 5. Missoula. —– 1993a. Okanagan Aspect: A Working Paper. ICSNL 28: 233-264, Seattle, Washington. —– 1994a. tułt and More on Okanagan Transitive Forms: A Working Paper. ICSNL 29, Pablo, Montana. —–1996. Interior Salish to-be and intention Forms: A Working Paper. ICSNL 31. 239-248, Vancouver, B.C. —– 1999. Parallels between Three CV-OK Aspectual Forms and their SpFl-Ka Cognates. ICSNL 33: 347- 365, Seattle, Washington. Mattina, Anthony, and Clara Jack. 1982. Okanagan Communication and Language. ICSNL 17: 269-294, Portland Oregon. —– 1986. Okanagan- Colville Kinship Terms. ICSNL 21:339-246. Seattle. —– 1992. Okanagan-Colville Kinship Terms. AL 34: 117-137. Mattina, Nancy. 1993. Some Lexical Properties of Colville-Okanagan Ditransitives. ICSNL 28: 265-284.Seattle, Washington.  87  —– 1993. Some Lexical Properties of Colville-Okanagan Ditransitives. ICSNL 28: 265-284. Seattle, Washington. —– 1994. Roots, Bases, and Stems in Colville-Okanagan. ICSNL 29. Pablo, Montana. —– 1996. Aspect and Category in Okanagan Word Formation. Ph.D. dissertation, Simon Fraser University. —– 1999. Future in Colville-Okanagan Salish.’ ICSNL 34: 215-230. Kamloops, B.C. Peterson, Sarah. 1995-06. Ns@lxc´ın 1: A Beginning Course in Okanogan Salish. The Center for Interior Salish, the Paul Creek Language Association, and the Lower Similkameen Indian Band. Peterson, Tyler. 2010. Epistemic Modality and Evidentiality in Gitksan at the Semantics-Pragmatics Interface. Doctoral Dissertation, University of British Columbia Reis Silva, Maria Am´elia. 2010. The expression of modality in Blackfoot. Paper presented at the VIII Workshop of Formal Linguistics. Universidade de So Paulo, So Paulo, Brazil. Rullmann, Hotze, Lisa Matthewson and Henry Davis. 2008. Modals as Distributive Indefinites. Natural Language Semantics. 16. 317-357 Vander Klok, J. (2008). Modality in Javanese: In between auxiliaries and verbs. In Annual conference of the Canadian Linguistics Association, UBC, Vancouver. Watkins, Donald. 1972. A Description of the Phonemes and Position Classes in the Morphology of Head of the Lake Okanagan. Ph.D. dissertation, University of Alberta, Edmonton. Willett, T. (1988). A cross-linguistic survey of evidentiality. Studies in Language, 12:51-97  88  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0073500/manifest

Comment

Related Items