@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix skos: . vivo:departmentOrSchool "Arts, Faculty of"@en, "Linguistics, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Gambarage, Joash Johannes"@en ; dcterms:issued "2019-10-22T15:31:32Z"@en, "2019"@en ; vivo:relatedDegree "Doctor of Philosophy - PhD"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description """This thesis makes two inter-related claims about the augment (a.k.a pre-prefix or initial vowel) based on evidence from Nata (Eastern Bantu, E45). Syntactically, the Nata augment is the realization of the functional category D(eterminer). The view that the augment is D is consistent with the claim that argument expressions are DPs, while predicate nominals obligatorily lack the D shell (cf. Longobardi 1994; Matthewson 1998; Déchaine and Tremblay 2011 and others). Semantically, I argue that the D distinction in Nata is solely based on speaker’s belief of existence. Beyond Nata, I claim that the core notion of existence is pertinent to other Bantu languages as well. The thesis challenges the widely held view that the D position is necessarily related to specificity or definiteness. I demonstrate that, once definiteness and specificity are controlled for in a precise fashion, the true contribution of Nata Ds as belief-of-existence Ds can be discerned. Cross-linguistically, the Bantu belief-of-existence D intersects with Salish assertion-of-existence Ds. In Salish, existence is asserted based on the speaker’s personal knowledge (Matthewson 1998). In Nata, this requirement is lacking. The Nata belief of existence D thus behaves as “the weakest D”, as it does not require a speaker to have personal knowledge of the individual. The theoretical implications of this analysis are twofold. First, existence Ds come in (at least) two guises, belief-of-existence versus assertion-of-existence. Second, existence Ds—in both Bantu and Salish—differ from “common ground” Ds of the type found in English, with the latter (but not the former) coding definiteness/specificity."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/72023?expand=metadata"@en ; skos:note "BELIEF-OF-EXISTENCE DETERMINERS: EVIDENCEFROM THE SYNTAX AND SEMANTICS OF NATAAUGMENTSbyJoash Johannes GambarageB.A (Education), The University of Dar Es Salaam, 2004M.A (Linguistics), The University of Dar Es Salaam, 2007A THESIS SUBMITTED IN PARTIAL FULLFILMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinThe Faculty of Graduate and Postdoctoral Studies(Linguistics)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)October 2019© Joash Johannes Gambarage, 2019The following individuals certify that they have read, and recommend tothe Faculty of Graduate and Postdoctoral Studies for acceptance, the dis-sertation entitled:Belief-of-Determiners in Bantu: Evidence from the Syntax and Semanticsof Nata Augmentssubmitted by Joash Johannes Gambarage in partial fulfillmentof the requirements forthe degree of Doctor of Philosophyin LinguisticsExamining Committee:Lisa MatthewsonCo-supervisorRose-Marie DéchaineCo-supervisorMichael Rochemont (deceased)Supervisory Committee MemberHotze RullmannUniversity ExaminerGloria Onyeoziri-MillerUniversity ExamineriiAbstractThis thesis makes two inter-related claims about the augment (a.k.a pre-prefix or initial vowel) based on evidence from Nata (Eastern Bantu, E45).Syntactically, the Nata augment is the realization of the functional categoryD(eterminer). The view that the augment is D is consistent with the claimthat argument expressions are DPs, while predicate nominals obligatorilylack the D shell (cf. Longobardi 1994; Matthewson 1998; Déchaine andTremblay 2011 and others). Semantically, I argue that the D distinction inNata is solely based on speaker’s belief of existence.Beyond Nata, I claim that the core notion of existence is pertinent toother Bantu languages as well. The thesis challenges the widely held viewthat the D position is necessarily related to specificity or definiteness. Idemonstrate that, once definiteness and specificity are controlled for in aprecise fashion, the true contribution of Nata Ds as belief-of-existence Dscan be discerned.Cross-linguistically, the Bantu belief-of-existence D intersects with Sal-ish assertion-of-existence Ds. In Salish, existence is asserted based on thespeaker’s personal knowledge (Matthewson 1998). In Nata, this require-ment is lacking. The Nata belief of existence D thus behaves as “the weak-est D”, as it does not require a speaker to have personal knowledge ofthe individual. The theoretical implications of this analysis are twofold.First, existence Ds come in (at least) two guises, belief-of-existence versusassertion-of-existence. Second, existence Ds—in both Bantu and Salish—differ from “common ground” Ds of the type found in English, with thelatter (but not the former) coding definiteness/specificity.iiiLay SummaryIn this thesis I studied the determiner systems of the Nata (Bantu) language.I concluded that common semantic features of definiteness and specificityfound in other well-studied languages are missing in Nata. My work opensup the notion of existence as it relates to the article system of Lillooet (Sal-ish) following Matthewson’s (1998) study. While the similarities betweenBantu and Salish – two unrelated families – suggest that the notion of ex-istence is robustly available as a determiner distinction, I proposed thatexistence determiners come in (at least) two guises, one is a system likeNata in which a speaker’s personal knowledge of the referent is not re-quired (belief of existence), the second is a system like Lillooet in which aspeaker’s personal knowledge of the referent is required (assertion of ex-istence). I have argued that existence determiners are also found in otherBantu languages.ivPrefaceThis dissertation consists of original and independent work by the author,Joash Johannes Gambarage, and is mainly based on fieldwork data fromnative speakers of Nata spoken in Tanzania and from the introspective judg-ments of the researcher. This fieldwork is covered by UBC Ethics Certificatenumber H16-01163 under the title “Weak Determiners in Bantu: Evidencefrom the Syntax and Semantics of Pre-prefixes in Nata”.A version of Chapter 2 of this dissertation, The Nata augment: now yousee it, now you don’t! appeared earlier as a qualifying paper under the UBCWorking Papers in Linguistics volume 34, pages 45–59. Some sections ofthe same chapter also appeared earlier as an article entitled The Pre-prefix inNata: An Interface Account in the Selected Proceedings of the 43rd AnnualConference on African Linguistics, University of Kansas volume 30, pages194–266.The picture presented as Figure 1.1 was taken with permission from theScope Fieldwork Project directed by Dr. Benjamin Bruenning (University ofDelaware). Map 1, The Ikoma-Nata-Isenye Cline, is from Higgins (2011) whocites the Survey Department of SIL’s Uganda-Tanzania Branch for providingher with such maps.vTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviiList of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . .xviiiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . xxiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xxiv1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 The goals of the thesis . . . . . . . . . . . . . . . . . . . . . . 41.2 Theoretical assumptions . . . . . . . . . . . . . . . . . . . . . 41.2.1 The DP hypothesis . . . . . . . . . . . . . . . . . . . . 41.2.2 The predicate/argument hypothesis . . . . . . . . . . 51.2.3 The assertion-of-existence hypothesis . . . . . . . . . 51.2.4 Choice function accounts . . . . . . . . . . . . . . . . 61.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . 7vi1.4 The language and the people . . . . . . . . . . . . . . . . . . 91.4.1 Language classification . . . . . . . . . . . . . . . . . 101.4.2 Dialect continuum and Nata neighbours . . . . . . . . 111.4.3 Language endangerment . . . . . . . . . . . . . . . . . 131.4.4 Previous literature on Nata . . . . . . . . . . . . . . . 131.4.5 Orthography and transcriptions . . . . . . . . . . . . . 141.5 Why study Nata? . . . . . . . . . . . . . . . . . . . . . . . . . 151.5.1 Phonology of the augment . . . . . . . . . . . . . . . . 161.5.1.1 The V–type augments . . . . . . . . . . . . . 171.5.1.2 The VV–type augment . . . . . . . . . . . . . 171.5.1.3 The CV–type augment . . . . . . . . . . . . . 181.5.1.4 The CVV–type augment . . . . . . . . . . . . 191.5.2 Morphology of the augment . . . . . . . . . . . . . . . 201.5.3 Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . 221.5.3.1 Syntactic categories . . . . . . . . . . . . . . 221.5.3.2 Agreement . . . . . . . . . . . . . . . . . . . 231.5.3.3 The verb complex . . . . . . . . . . . . . . . 251.6 Thesis outline . . . . . . . . . . . . . . . . . . . . . . . . . . . 272 The Nata augment: now you see it, now you don’t! . . . . . . 282.1 The puzzling behaviour of the Nata augment . . . . . . . . . 302.1.1 Sometimes the Nata augment is there . . . . . . . . . 302.1.2 Sometimes the Nata augment isn’t there . . . . . . . . 322.2 Possible accounts and why they don’t work . . . . . . . . . . 342.2.1 The mass-count contrast does not condition the aug-ment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342.2.2 Case does not condition the augment . . . . . . . . . . 402.2.2.1 Verb extensions do not license the augment . 452.2.2.2 No evidence for morphological Case in Nata 472.2.2.3 The augment is not semantically vacuous . . 492.2.3 Deixis does not condition the augment . . . . . . . . . 512.2.3.1 Defining deixis . . . . . . . . . . . . . . . . . 512.2.3.2 The Nata augment does not encode deixis . . 51vii2.2.4 Definiteness does not condition the augment . . . . . 552.2.4.1 Defining definiteness . . . . . . . . . . . . . 562.2.4.2 The Nata augment does not encode novelty-familiarity . . . . . . . . . . . . . . . . . . . 562.2.4.3 The Nata augment does not presuppose ex-istence . . . . . . . . . . . . . . . . . . . . . 582.2.4.4 The Nata augment does not presupposeuniqueness . . . . . . . . . . . . . . . . . . . 602.2.4.5 The Nata augment does not assert uniqueness 622.2.4.6 The Nata augment does not presup-pose/assert maximality . . . . . . . . . . . . 632.2.4.7 Augments are not weak/strong German def-inite Ds . . . . . . . . . . . . . . . . . . . . . 652.2.5 Specificity does not condition the augment . . . . . . 702.2.5.1 Defining specificity . . . . . . . . . . . . . . 702.2.5.2 The Nata augment does not encode specificity 742.2.5.3 The augment is not the English this-specificindefinite . . . . . . . . . . . . . . . . . . . . 772.2.5.4 The augment is not the English indefinite a . 802.2.6 The Nata augment is not a ‘domain restrictor’ . . . . . 822.2.6.1 Similarities between the Nata Ds and do-main restriction Ds . . . . . . . . . . . . . . 842.2.6.2 Differences between the Nata augment anddomain restriction Ds . . . . . . . . . . . . . 902.3 Solving the Nata puzzle: the two ingredients . . . . . . . . . 942.3.1 Ingredient 1: argument vs predicate nominals . . . . . 952.3.2 Ingredient 2: overt versus covert augment . . . . . . . 962.4 Summary and conclusion . . . . . . . . . . . . . . . . . . . . 973 The Syntax of Nata D . . . . . . . . . . . . . . . . . . . . . . . 993.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 993.2 The internal syntax of the Nata DP . . . . . . . . . . . . . . . 1003.2.1 The decomposition of the Nata noun . . . . . . . . . . 100viii3.2.2 The augment as a proclitic D . . . . . . . . . . . . . . 1033.2.2.1 The augment does not co-occur with theDEM proclitic . . . . . . . . . . . . . . . . . 1043.2.2.2 The augment does not co-occur with thehonorific proclitic . . . . . . . . . . . . . . . 1053.2.3 Predictions for the proposal that augments are Ds . . 1063.3 Predicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1083.3.1 Nata nominal predicates are φ-N . . . . . . . . . . . . 1093.3.1.1 Simple nominal predicates lack a D . . . . . 1093.3.1.2 D-linked wh-phrases as complex nominalpredicates . . . . . . . . . . . . . . . . . . . 1123.3.2 Nata adnominal predicates are φ-A . . . . . . . . . . 1143.3.2.1 Post-copula adjectives lack a D . . . . . . . . 1143.3.2.2 Adjectival modifiers lack a D . . . . . . . . . 1153.3.3 Nata secondary predicates are φ-X . . . . . . . . . . . 1163.3.3.1 Nata secondary nominal predicates lack a D 1173.3.3.2 Nata secondary adjectival predicates lack a D1183.3.4 Nata adverbials lack a D . . . . . . . . . . . . . . . . . 1203.3.5 Nata infinitives lack a D-layer . . . . . . . . . . . . . . 1223.4 Argument nominals are D-φ-N . . . . . . . . . . . . . . . . . 1233.4.1 D is required in all argument positions . . . . . . . . . 1253.4.1.1 D is required in subject position . . . . . . . 1253.4.1.2 D is required in direct object positions . . . . 1293.4.1.3 D is required in indirect object position . . . 1323.4.1.4 D is required with gerunds . . . . . . . . . . 1343.4.1.5 D is required with temporal arguments . . . 1353.4.2 D is required with complex nouns . . . . . . . . . . . 1363.4.2.1 D is required with N-N compounds . . . . . 1363.4.2.2 D is required with determiner spread . . . . 1383.5 Polarity-sensitive argument DPs are D∅-φ-N . . . . . . . . . 1413.5.1 Negation . . . . . . . . . . . . . . . . . . . . . . . . . 1463.5.1.1 Sentential negation licenses D∅ . . . . . . . 1483.5.1.2 Adversative predicates license D∅ . . . . . . 155ix3.5.1.3 Covert negation/reduced pitch licenses D∅ . 1573.5.1.4 Negative light verbs license D∅ . . . . . . . 1603.5.2 Questions . . . . . . . . . . . . . . . . . . . . . . . . . 1623.5.2.1 Polar questions license D∅ . . . . . . . . . . 1623.5.2.2 Wh-questions do not license D∅ . . . . . . . 1643.5.3 Conditionals . . . . . . . . . . . . . . . . . . . . . . . 1653.5.4 Modals . . . . . . . . . . . . . . . . . . . . . . . . . . 1673.5.4.1 Epistemic modals license D∅ . . . . . . . . . 1673.5.4.2 Subjunctive mood licenses D∅ . . . . . . . . 1703.5.4.3 Evidentials license D∅ . . . . . . . . . . . . . 1713.6 Summary and conclusion . . . . . . . . . . . . . . . . . . . . 1754 The semantics of Nata D . . . . . . . . . . . . . . . . . . . . . . 1774.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1774.2 Defining Existence Ds . . . . . . . . . . . . . . . . . . . . . . 1784.3 The proposal: Nata Ds encode ‘existence’ . . . . . . . . . . . 1814.4 Properties of speaker-oriented existence Ds . . . . . . . . . . 1864.4.1 Speaker-oriented existence Ds . . . . . . . . . . . . . . 1874.4.2 Ds encode a core notion of existence . . . . . . . . . . 1874.4.2.1 Existence with actual referents . . . . . . . . 1874.4.2.2 Existence with non-actual referents . . . . . 1904.5 Correlated properties of speaker-oriented existence Ds . . . . 1914.5.1 Speaker-oriented Ds do not encode definiteness . . . . 1924.5.2 Speaker-oriented Ds do not encode specificity . . . . . 1934.5.3 Polarity Ds must be licensed . . . . . . . . . . . . . . 1954.5.3.1 Licensing in St’át’imcets . . . . . . . . . . . . 1964.5.3.2 Licensing in Nata . . . . . . . . . . . . . . . 1984.6 The Locus of Parametric Variation . . . . . . . . . . . . . . . 2024.6.1 Requirement for speaker’s personal knowledge . . . . 2034.6.2 Surmising contexts . . . . . . . . . . . . . . . . . . . . 2054.6.3 Non-materialized referents . . . . . . . . . . . . . . . 2104.6.4 Possible worlds: attitude verbs . . . . . . . . . . . . . 2124.6.5 Deictic features in D . . . . . . . . . . . . . . . . . . . 215x4.7 Choice function analysis . . . . . . . . . . . . . . . . . . . . . 2204.7.1 Defining choice functions . . . . . . . . . . . . . . . . 2204.7.2 Interpretation of Ds in assertion-of-existence contexts 2224.7.2.1 Interpretation of overt Ds in declaratives . . 2244.7.2.2 Interpretation of overt Ds with/under nega-tion . . . . . . . . . . . . . . . . . . . . . . . 2244.7.2.3 Interpretation of overt Ds in interrogatives . 2254.7.2.4 Interpretation of overt Ds in modals . . . . . 2264.7.2.5 Interpretation of overt Ds in conditionals . . 2274.7.3 Interpretation of overt Ds in belief-of-existence con-texts: surmising, possible worlds... . . . . . . . . . . . 2274.7.4 Interpretation of overt Ds in quantifier phrases . . . . 2304.7.4.1 Universal quantifiers . . . . . . . . . . . . . 2304.7.4.2 The problem with the SG-ọsẹ quantifier . . . 2324.7.4.3 Simple generics . . . . . . . . . . . . . . . . 2354.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2365 Existence in other Bantu languages . . . . . . . . . . . . . . . 2385.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2385.2 Belief of existence D systems . . . . . . . . . . . . . . . . . . 2425.2.1 Existence Ds in Runyankore-Rukiga . . . . . . . . . . 2425.2.1.1 R/Rukiga Ds do not encode definiteness . . . 2425.2.1.2 R/Rukiga Ds do not encode specificity . . . . 2435.2.1.3 D distinction in R/Rukiga . . . . . . . . . . . 2445.2.1.4 Overt Ds as belief of existence Ds . . . . . . 2465.2.2 Existence Ds in Haya . . . . . . . . . . . . . . . . . . . 2485.2.2.1 Haya Ds do not contrast for definiteness . . 2495.2.2.2 Haya Ds do not encode specificity . . . . . . 2505.2.2.3 D distinction in Haya . . . . . . . . . . . . . 2515.2.2.4 Overt Ds as belief-of-existence Ds . . . . . . 2535.2.2.5 Accounting for residual cases in Haya . . . . 2545.2.3 Existence Ds in Luganda . . . . . . . . . . . . . . . . . 2585.2.3.1 Luganda Ds do not encode definiteness . . . 259xi5.2.3.2 Luganda Ds do not encode specificity . . . . 2605.2.3.3 D distinctions in Luganda . . . . . . . . . . . 2615.2.3.4 Overt Ds as belief-of-existence Ds in Luganda2625.2.3.5 Accounting for residual issues in Luganda . . 2645.2.4 Existence Ds in Kinande . . . . . . . . . . . . . . . . . 2655.2.4.1 Kinande Ds do not encode definiteness . . . 2665.2.4.2 Kinande Ds do not encode specificity . . . . 2675.2.4.3 D distinction in Kinande . . . . . . . . . . . 2675.2.4.4 Overt Ds behave as belief-of-existence Ds . . 2695.2.4.5 Accounting for residual issues in Kinande . . 2715.2.5 Existence Ds in Xhosa . . . . . . . . . . . . . . . . . . 2745.2.5.1 Xhosa Ds do not encode definiteness . . . . . 2755.2.5.2 Xhosa Ds do not encode specificity . . . . . . 2765.2.5.3 D distinction in Xhosa . . . . . . . . . . . . . 2775.2.5.4 Overt Ds as belief-of-existence Ds . . . . . . 2795.2.6 Existence Ds in Zulu . . . . . . . . . . . . . . . . . . . 2815.2.6.1 Zulu Ds do not encode definiteness . . . . . 2825.2.6.2 Zulu Ds do not encode specificity . . . . . . 2825.2.6.3 D distinction in Zulu . . . . . . . . . . . . . . 2835.2.6.4 Overt Ds as belief-of-existence Ds . . . . . . 2865.2.6.5 Accounting for residual cases in Zulu . . . . 2875.3 Assertion-of-existence D systems . . . . . . . . . . . . . . . . 2915.3.1 Existence Ds in Bemba . . . . . . . . . . . . . . . . . . 2915.3.1.1 Bemba Ds do not encode definiteness . . . . 2925.3.1.2 Bemba Ds do not encode specificity . . . . . 2935.3.1.3 D distinction in Bemba . . . . . . . . . . . . 2945.3.1.4 Overt Ds as assertion-of-existence Ds . . . . 2965.4 Ds that do not encode existence . . . . . . . . . . . . . . . . . 2995.4.1 Lack of Existence Ds in Dzamba . . . . . . . . . . . . 2995.4.2 Dzamba Ds contrast for novelty-familiarity . . . . . . 3005.4.3 Dzamba Ds do not contrast for specificity . . . . . . . 3025.5 Summary, remarks and conclusion . . . . . . . . . . . . . . . 305xii6 Locus of Variation in Bantu . . . . . . . . . . . . . . . . . . . . 3086.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3086.2 DP-internal constraints . . . . . . . . . . . . . . . . . . . . . . 3096.2.1 D and demonstratives . . . . . . . . . . . . . . . . . . 3096.2.2 D and modifiers . . . . . . . . . . . . . . . . . . . . . 3116.2.3 D and personal pronouns . . . . . . . . . . . . . . . . 3146.2.4 D and the universal quantifier ‘every’ . . . . . . . . . 3166.3 DP-external constraints . . . . . . . . . . . . . . . . . . . . . 3186.3.1 Sentence fragment answers . . . . . . . . . . . . . . . 3196.3.2 Surface c-command . . . . . . . . . . . . . . . . . . . 3206.3.3 Locality . . . . . . . . . . . . . . . . . . . . . . . . . . 3236.3.4 Clause-mate restrictions . . . . . . . . . . . . . . . . . 3266.3.5 Topic marking . . . . . . . . . . . . . . . . . . . . . . 3286.3.6 Focus licensing . . . . . . . . . . . . . . . . . . . . . . 3316.3.6.1 Focussed DPs in negative environments . . . 3326.3.6.2 Focussed DPs in relative clauses . . . . . . . 3336.3.6.3 Focused DPs with clefts . . . . . . . . . . . . 3356.3.6.4 Focussed DPs in declarative sentences . . . . 3376.3.6.5 Focussed stressed DPs in Xhosa/Zulu . . . . 3386.4 Areas of further research . . . . . . . . . . . . . . . . . . . . . 3426.4.1 D requirement on proper names . . . . . . . . . . . . 3426.4.1.1 N is not Pred in the ‘an Einstein’ test . . . . 3436.4.1.2 N is not strictly Pred in complement clauses 3456.4.1.3 The N-to-D movement test . . . . . . . . . . 3466.4.2 D requirement on prepositional objects . . . . . . . . 3476.4.2.1 Objects of comitative P permit overt orcovert D . . . . . . . . . . . . . . . . . . . . 3476.4.2.2 Objects of associative P permit overt orcovert D . . . . . . . . . . . . . . . . . . . . 3486.4.2.3 Objects of locative P require overt D . . . . . 3506.4.3 D and locatives in other Bantu languages . . . . . . . 3556.4.3.1 Bemba/Kinande-type: objects of locative Pprohibit overt D . . . . . . . . . . . . . . . . 355xiii6.4.3.2 Zulu/Luganda: objects of locative P permitovert D sometimes . . . . . . . . . . . . . . . 3576.5 Summary and conclusion . . . . . . . . . . . . . . . . . . . . 361Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365xivList of TablesTable 1.1 Useful orthographic symbols . . . . . . . . . . . . . . . . . 15Table 1.2 The Nata N-prefixes . . . . . . . . . . . . . . . . . . . . . . 21Table 1.3 Nata syntactic categories and augment possibilities . . . . 23Table 1.4 Agreement paradigms . . . . . . . . . . . . . . . . . . . . . 25Table 1.5 The Nata verb template . . . . . . . . . . . . . . . . . . . . 26Table 2.1 Similarities and differences between Zulu and Nata Ds . . 40Table 2.2 Three-way distinction of deictic demonstratives in Nata . 52Table 2.3 Distribution of definite strong and weak Ds and Nata Ds . 70Table 2.4 Similarities and differences in the three languages . . . . . 84Table 3.1 Operators that license the polarity-sensitive D∅ . . . . . . 175Table 4.1 Requirements for use of the existence Ds . . . . . . . . . . 184Table 4.2 Summary of correlated diagnostics . . . . . . . . . . . . . 192Table 4.3 Locus of variation between St’át’imcets and Nata . . . . . 202Table 5.1 Different D distinctions among augment languages . . . . 240Table 5.2 R/Rukiga Ds and their correlation with other D systems . 242Table 5.3 Haya Ds and their correlation with other D systems . . . . 249Table 5.4 Luganda Ds and their correlation with other existence Ds . 259Table 5.5 Kinande Ds and their correlation with other existence Ds . 266Table 5.6 Xhosa Ds and their correlation with other existence Ds . . 275Table 5.7 Zulu Ds and their correlation with other existence Ds . . . 281Table 5.8 Bemba Ds and their correlation with other existence Ds . . 292xvTable 5.9 Dzamba Ds and their correlation with existence Ds . . . . 300Table 6.1 Parametric variation in the syntax . . . . . . . . . . . . . . 319Table 6.2 Syntactic-semantic similarities and variation . . . . . . . . 362xviList of FiguresFigure 1.1 003-Not-All-Birds-In-Trees . . . . . . . . . . . . . . . . . . 8Figure 3.1 Pitch pattern for regular sentences . . . . . . . . . . . . . 159Figure 3.2 Pitch pattern for sentences with covert negation . . . . . 159xviiList of AbbreviationsACC accusativeAOE assertion of existenceAPL associative pluralAPPL applicativeAUG augmentBOE belief of existenceC noun classCAUS causativeCOMP complementizerCOP copulaDAT dativeDEF definitenessDEIX deicticDEM demonstrativeD(ET) determinerDST distalEC Expletive constructionERG ergativexviiiEXIS assertion-of-existenceFEM feminineFOC focusFUT futureFV final vowelGEN genitiveIPFV imperfectiveINCIP incipientINF infinitiveINTERM intermediateLOC locativeMASC masculineMOD modalNEG negationNMLZ nominalizerNOM nominativeNT number neutralOBJ objectOBL oblique markerOCC occupationOM object markerPART partitivePASS passivePROG progressivexixPROX proximalPST pastPFV perfectivePL pluralPOSS possessivePRES presentPROX proximalRED reduplicantREFL reflexiveREL relativeREM remoteSA subject agreementSG singularSUB subjectSUBJV subjunctivexxAcknowledgementsI cannot exhaust a list of people who helped me achieve this remarkablestage of my life. My greatest thanks is to my two co- supervisors Dr. LisaMatthewson and Dr. Rose-Marie Déchaine, for their constructive criticisms,insightful suggestions, support and courage in every part of this thesis. Dr.Matthewson has tirelessly read many drafts of this thesis and provided on-time feedback every time. Her deep knowledge of both St’át’imcets andformal semantics became a keystone to my semantic analysis. Dr. Matthew-son and Dr. Henry Davis have played a vital role in my exploration of theBantu–Salish connection. I thank Dr. Déchaine for always taking me outof my comfort zone, challenging me to be a critical thinker and less of ahasty writer. She has challenged me on how to present the data, to makeargumentation and connections, to go back, to restart: all these were justto make my work better even when she knew doing that would take extratime and pain. The late Dr. Michael Rochemont was a member of my for-mer committee and he contributed greatly to the earlier drafts of this thesis.His belief in me, and that my research could be something extraordinaryto the field was awakening. His memory will live with me as long as I shalllive. Any errors are solely mine but any exposition embodied in this thesisis largely due to my committee.I should also thank my two amazing university examiners, Dr. HotzeRullmann and Dr. Gloria Onyeoziri-Miller, for their comments that raisedthe bar of the quality of this work.I thank several language consultants who accepted to share their beau-tiful languages with me. From Nata, my fellow native speakers: Sab-xxihiti Winyanya, Mnata Sarota, Nyabhikwabhe Yati (Baunsa), Mugesi Ma-chota, Peter Kishora and my sister Wasato Gambarage. I thank my cousinWinyanya Marobhe and his wife Nyambura for allowing me and my col-league to stay at their house for two months (July and August, 2016) forno cost during my fieldwork in Nata. Thanks for their generosity and forfeeding us so well: Muuchọmiri bhọọsẹ chwẹẹ! I thank the many otherlanguage consultants and/or speaker-linguists I have worked with: Man-disa Ndlovu and Lutsha Bata (Xhosa); Mthuli Buthelezi (Zulu); Dr. PhillipMutaka and Jack Mutaka (Kinande); Edie Amaitum and Richard Musoka(Luganda); Musonda Chilengwe (Bemba); Dr. Eyamba Georges Bokamba(Dzamba); Dr. Asiimwe Allen (Runyankore-Rukiga); and Angela Katabaro(Haya). This dissertation could not be what it is without them.I am grateful to the following people for providing feedback, encourage-ment and insights on various aspects of my work: Manfred Krifka, Larry Hy-man, Vicki Carstens, Malte Zimmerman, Talmy Givón, Georges Bokamba,Phillip Mutaka, Claire Halpert, Michael Diercks, Patricia Schneider–Zioga,and Holly Higgins. This research has also benefited from the commentsfrom several members of the UBC Linguistics department who were alsomy early teachers: Hotze Rullmann, Martina Wiltschko, Doug Pulleyblankand Henry Davis. Thank you Molly Babel, Bryan Gick, Kathleen Hall andin loving memory, Eric Vatikiotis-Bateson, for introducing me to the the-oretical linguistics world, along with others mentioned above. Thanks tothe audiences of ACAL47 and ACAL50 for their valuable comments andquestions.Thanks to my friend and colleague James Gauthier, my first friendin Canada. I also thank my classmates in the 2010 cohort: James Crip-pen, Sihwei Chen, Ella Fund-Reznicek, Tianhan Liu, Michael McAuliffe,Stacey Menzies, Kevin McMullin, and Lauren Quinn. My first TAs Jen-nifer Glougie, Jennifer Abel, Patrick Littell and Heather Bliss made myfirst classes less stressful. Thanks Jen Abel for proofreading this work.Thanks to all my colleagues: Andrei Anghelescu (with whom I conductedthe field work in Nata), Erin Guntly, Emily Sadlier-Brown, Adriana Osa-Gomez, Herman Keupdjio, Sam Akinbo and all the members of the NataxxiiWorking Group. Thank you Edna Dharmaratne and Shaine Meghji for be-ing extraordinary staff members.I am highly indebted to my family for their support throughout thislife. Thanks to my sons “the boys” Jones and Eli. You two were absolutelyamazing. I will forever appreciate your prayers every night and your funnyjokes that put a smile on my face during downtimes. I owe my wife Neemaenormous thanks for always being on my side, in the hills and valleys ofmy academic journey. She agreed to take on many family responsibilities;she provided moral and financial support throughout my studies here atUBC. Words just can’t be enough. Thanks to Neema’s friends behind thescenes: Julie, Kashmira and Nancy at Westcoast Suites. Thanks to PennyDesjarlais and Nancy Moseley for being so special to my family. Thanksto mom and dad for their upbringing and giving me the gift of speakingNata–the language I love so dearly. Thanks to my second family, OakridgeSeventh-Day Adventist Church, and to all my friends in the Afro-CaribbeanSabbath school group. I have always felt a sense of belonging.I wish to thank and acknowledge funding sources that made my PhDstudies possible: (i) the UBC Four Year Fellowship (2010-2014); (ii) In-ternational Partial Tuition Scholarship (2011-2018); (iii) Faculty of Grad-uate studies Award (2015/2018); (iii) UBC Emergency Fund for develop-ing country students (2015/2016). (iv) Stipend for Teaching Assistantshipfrom the UBC Linguistics Department (2011-2016). (v) Salary as a NataLanguage consultant to the 2013/2014 field methods class. (vi) SSHRC In-sight Grant to Douglas Pulleyblank (2011-2016). (vii) SSHRC Insight Grantto Lisa Matthewson and Hotze Rullmann (2016-2017).Last but not least, thanks to the Almighty God for providing all of thesewonderful people and resources listed above, and everything else besides.xxiiiDedicationTo my beautiful wife Neema and to my two awesome boys: Jones and Eli.xxivChapter 1IntroductionIf I have seen farther it is by standing on the shoulders of giants.— Sir Isaac NewtonThis thesis explores the syntactic and semantic properties of the aug-ment (a.k.a the pre-prefix) in Nata (Eastern Bantu, E45). The augment isthe leftmost element of the noun domain labelled as AUG in (1a). The aug-ment precedes the noun-class prefix (the element marked as C; see §1.5.2for details about noun classes). Thus, traditionally the Nata noun templateis presented as in (1b).(1) a. o=mú-nwaAUG=C3-mouth‘a/the mouth’b. AUGMENT=CLASS PREFIX–NOUN STEMIn this thesis I investigate the syntactic status of the augment in casessuch as (2), where nominals denoting an entity appear with the augment,but nominals denoting a property lack the augment, (3).[Context: Bahati is a gender neutral name. A woman and a man are stand-ing before us. M is wondering which person is Bahati:]1(2) a. BahatiBahatiBahatin=o=mo-subhen=o=mo-súbheCOP=AUG=C1-man‘Bahati is the man.’b. #Bahati#BahatiBahatim=mo-subhen=mo-súbheCOP=C1-man‘Bahati is a man.’(3) [Context: M is describing Bahati’s gender:]a. #Bahati#BahatiBahatin=o=mo-subhen=o=mo-súbheCOP=AUG=C1-man‘Bahati is the man.’b. BahatiBahatiBahatim=mo-subhen=mo-súbheCOP=C1-man‘Bahati is a man.’Largely, in this thesis, I investigate the semantic principle that forcesNata speakers to use an overt augment in sentences such as (4), and not incontexts such as (5a).(4) a. n-a-a-rọch-en-á-á-rotʃ-eFOC-PST-see-FVo=mo-subheo=mo-súβeAUG=C1-man‘S/he saw a/the man.’2b. ta-a-rọch-eta-a-rotʃ-éNEG-PST-see-FVo=mo-subheo=mo-súβeAUG=C1-man‘S/he didn’t see a/the man.’(5) [Context: The speaker does not believe that X saw any man:]a. ta-a-rọch-eta-a-rotʃ-eNEG-PST-see-FVmo-subhemo-súβeC1-man‘S/he didn’t see a/any man.’b. *n-a-a-rọch-e*n-a-a-rotʃ-eFOC-PST-see-FVmo-subhemo-súβeC1-manIntended: ‘S/he saw a/*any man.’In order to provide an analysis of the Nata augment, I adopt the hypoth-esis that nominal arguments are DPs, and predicate nominals are φP/NP(Stowell 1989; Longobardi 1994, 2001, 2008; Déchaine 1993). I provideevidence that the augment instantiates the category of D, whose surfaceform may vary according to whether D is overt as in (4) or covert (D∅) asin (5a) (see Longobardi 1994, 2001, 2008; Déchaine 1993; Déchaine et al.2018). I claim in this thesis that the cases in (2) and (3) involve a partitionbetween a DP argument and a nominal predicate which lacks a D layer. Iwill therefore gloss the augment as D from now on.While I argue in this thesis that syntactically the augment is D, the Nataaugment does not seem to encode definiteness or specificity as widely as-sumed in the literature on the semantics of Ds. Rather, the contrast betweenthe overt D in (4) and the covert D in (5a) is forced by the speaker’s com-mitment to existence of a referent for the NP. I show that the Nata augmentsystem is strikingly similar to the D system of St’át’imcets (Salish) whichencodes the notion of ‘assertion-of-existence’ (see Matthewson 1998, 1999).3The rest of Chapter 1 highlights the goals of this thesis and the the-oretical framework in which the notion of the augment/D is couched. Iintroduce the Nata language and people and discuss why studying Nata isimportant.1.1 The goals of the thesisThis thesis explores various syntactic and semantic properties of the aug-ment in Nata. The thesis has four main goals:(6) a. To present a syntactic-semantic analysis of the augment/D systemin Nata.b. To compare Nata’s augment/D system with the strikingly similarD system of St’át’imcets (Salish) and other Bantu languages.c. To revisit previous semantic hypotheses and show that none ofthe previous semantic accounts are capable of accounting for whatforces augment/D choices in Nata and Bantu more generally.d. To account for the locus of parametric variation between the Nataaugment/D and other Ds (in Bantu and in Salish).1.2 Theoretical assumptionsI briefly review the relevant frameworks for the current proposal about theaugments in Nata. I employ the following frameworks: (i) the DP hypothe-sis, (ii) the assertion of existence account, (iii) syntactic licensing accounts,and (iv) the choice function account. While I will not attempt to reviewall the details of each account, I do present the main insights that resonatewith my proposal. I start with the DP hypothesis.1.2.1 The DP hypothesisThe DP hypothesis was developed in the wake of Brame (1982), Szabolsci(1983), and Abney (1987); its basic assumption is that the determiner is4the functional category D that selects an NP as its complement, (7a). Thisview is consistent with the current proposal that the augment is the mor-phosyntactic realization of the functional category D. As such, I will notadopt the earlier version of the internal structure of the NP which assumesthe noun is the head of the phrase and the D sits in the specifier of the NP(Jackendoff 1977; Chomsky 1981), (7b).(7) a. DPSpecifier D′D NPb. NPD N′NIn Chapter 3, I show that the DP hypothesis is supported by the internalstructure of Nata DPs.1.2.2 The predicate/argument hypothesisLongobardi (1994), following Stowell (1989), explicitly argues that a com-mon noun must have a D to function as an argument. This assumptionmakes a distinction between argument and predicate nominals:(8) D and argumenthood (Longobardi 1994: 620, 628)a. A ‘nominal expression’ is an argument only if it is introduced bya category D (p.620).b. DP can be an argument, NP cannot (quoting Stowell 1989).I adopt Longobardi’s predicate/argument contrast and argue that, in Nata,Ds are not only necessary for projecting a DP but are also required forargumenthood (i.e., making an entity-denoting argument of type e).1.2.3 The assertion-of-existence hypothesisMatthewson (1998) unearths a new typology of indefinite Ds in St’át’imcetswhose function is encoding assertion of existence. She illustrates thatSt’át’imcets Ds encode the following distinction:5(9) Determiners in St’át’imcets (Matthewson 1998: 53-54)a. The non-polarity Ds (X...a) encode assertion of existence.b. The polarity D encodes failure to assert existence.Matthewson illustrates that the overt Ds, X...a (where X refers to differ-ent elements encoding deictic distinctions), consistently encode assertionof existence, while the D ku fails to encode existence. A major motivationfor adopting the assertion-of-existence account is the observation that Dsin both Nata and St’át’imcets encode whether the speaker believes the NP’sreferent exists. I provide a semantic analysis of the D system in Nata andcompare it with the strikingly similar D system in St’át’imcets.1.2.4 Choice function accountsI claim that Nata Ds require an analysis involving choice functions whichis consistent with various works on indefinite Ds (Reinhart 1997; Winter1997; Kratzer 1998; Matthewson:1999; and others). The definition of achoice function is as follows:(10) A choice function definition:A function f is a choice function (CH(f)) if it applies to any non-emptyset and yields a member of that set. Reinhart (1997: 372)I argue in Chapter 4 that Nata augments are all analyzable as indefinites.Thus, in Reinhart’s (1997) formulation, D is a function, f<e> whichtakes a nominal expression of type , and yields an entity from the NPset it is applied to. The preview of my analysis for an example like (11a) isgiven in (11b):(11) a. MakuruMakurúMakurua-ka-ghoor-aa-ka-ɣór-aSA1-PST-buy-FVe=ghi-tabhoe=ɣí-taβoD=C7-book‘Makuru bought a book.’6b. 9f [CH(f) & [Makuru bought f(book)]]There is a choice function (f) which picks out a book from theset of books and Makuru bought the book chosen by f.In chapter 4, I discuss the specific properties of choice functions that Iadopt, and the implication of choosing one approach over the other.1.3 MethodologyThough I speak Nata natively, to avoid building a theory based only onmy own judgments, I adopted Featherston’s (2007) standards for collectinglinguistic data by verifying attested data with multiple informants. I con-ducted fieldwork in the village of Nata Mbisso (the heartland of the Natalanguage) in July and August 2016. I worked with 6 speakers (3 malesand 3 females). Five of the speakers were born and raised in Nata village,and one female speaker was born and raised in Mugeta village. Their agesranged between 40 and 60 and they all spoke Nata natively1.Semantic fieldwork is a challenging enterprise. Therefore, well-designed elicitation techniques have to be employed, which include but arenot limited to spontaneous discourse and direct elicitations (Dimmendaal2001; Matthewson 2004; and others). Matthewson (2004) argues furtherthat one cannot gather adequate information about meaning from sponta-neous discourse alone, but must also include direct elicitation, asking con-sultants for grammaticality, felicity and truth value judgments. The findingsof the current research resulted from these methodological practices.In spontaneous elicitations, speakers were asked to tell their ownstories and narratives like folktales. In directed elicitations, I primarilyused two methods. One was setting up a relevant context of use. Forinstance, to verify whether the augment can be used in new discoursecontexts, the consultants in their individual elicitation sessions were1. One of the language consultants was my older sister, whom I grew up with in the samehousehold until her late teenage years when she got married and moved to Nata-Mbisovillage. The reason I included her is because I was interested to see if there would be anydifferences between her grammaticality and felicity judgements and mine. I did not comeacross any.7asked to pay attention to the context of use and provide a relevant re-sponse. An example of a context used in directed elicitations is given below:[Context: Suppose Wanchota goes to school and on her way back shemeets a man who stops her to ask for directions. When she gets home, thefirst thing she says to her mom is about what happened on her way home.She goes: Mama, guess what?...(please continue)].I also employed the option of using the Totem Field storyboard model(cf. Burton and Matthewson 2015), and Bruening’s (2008) storyboards forquantifier elicitation. These were presented as powerpoint slides on a lap-top. For instance, using Bruening’s image 003 (http://udel.edu/~bruening/scopeproject/scopeproject.html), I asked the consultants if it was at all pos-sible to say all (the) birds are sitting in the tree in Nata given the scenariobelow. In turn, as the answer was no, I asked them to give the correctsentence one would use in this scenario:Figure 1.1: 003-Not-All-Birds-In-TreesMy role as a speaker-linguist was clear in this undertaking. I producedsentences and asked the consultants to give their judgements in relation tothe context. This was fun! I also noticed that when some consultants had a‘foggy head’, they would give responses that were infelicitous and counter-8intuitive; however, they rejected such responses in the following elicita-tions as if they had never said anything like that. As a speaker-linguistI could easily tell that a consultant’s mind was not on task, and I wouldask them to take a break, or drink something refreshing to help them keepfocussed on the task.For the other Bantu languages I report on in Chapter 5–Runyankore-Rukiga, Kinande, Xhosa, Zulu, Bemba, and Dzamba–I used phone inter-views and asked speakers to help with checking my transcriptions. ForHaya and Luganda, the elicitations were done here at UBC with nativespeakers who are UBC students.1.4 The language and the peopleNata is an endangered Bantu language spoken in Tanzania by approxi-mately 7000 speakers. Speakers of the language call themselves a=βanáataand refer to their language as e=kináata. In the context of Swahili, whichdoes not have augments, all languages are referred to by using the Ki- pre-fix; thus, Nata is called Kináata. In the English text, I have omitted theaugment e= and the ki- prefix to be consistent with the meta-language(English). Thus, throughout I refer to the language under study as ‘Nata’.Nata is spoken in several villages in the two districts of the Mara Region,namely Serengeti and Bunda. The majority of Nata speakers are found inthe Mara Region, particularly in the villages of Nata-Mbiso, Nata-Motukeri,Makondose and Bwanda. Nata people claim that their first ancestors livedin Bwanda (a.k.a Rakana), but today, the heartland of Nata is the villagesof Nata-Mbiso and Nata-Motukeri. There are a number of villages whichcombine Nata speakers and speakers of other languages. These include thevillages of Mugeta and Kyandege in the Bunda district, and the villages ofNyiichoka, Ikoma and Burunga in the Serengeti district.91.4.1 Language classificationIn the wider typology of African languages, Greenberg (1963) shows thatNata belongs to the Bantu family, which falls under the Niger-Congo family.The latter falls under a major group known as the Benue-Congo group.There have been attempts to classify Bantu languages into zones of re-lated languages. Guthrie (1948) places Nata under group E.40 (Ragoli-Kuria Group) where Nata and Ikoma are coded as E45 (Zone E, Group 40,Index 5). The languages under the Ragoli-Kuria Group are listed in (21):(12) Ragoli-Kuria Group (E.40)E.41 Logoolį (Ragoli)E.42 Gųsįį (Kisii)E.43 KorįaE.44 ZanakįE.45 Nata (Ikoma)E.46 Sonjo (Sonyo)Guthrie (1971) maintains the above classification in which Nata andIkoma are treated as one and the same language. A number of classifica-tions treat Nata and Ikoma as the same language, where Nata and Ikomaconsistently belong to the East Nyanza subgroup which is comprised ofGusii, Kuria, Zanake, Ngoreme and Shashi (see Nurse (1977), as well asNurse and Philippson (1980)). Nata, Isenye and Ikoma are also treated asone language in Heine’s (1976) classification.Maho (2003) attempts a new classification where he moves some lan-guage groups placed in Guthrie’s (1948) zone E, to the new group namedJE.40, which has a total of 19 languages. For instance, Guthrie’s E.20(Kwaya-Ruuri) and E.40 (Lagooli-Kuria) groups are now in the JE.40 group.Surprisingly, Isenye, which is a sisterlect to Nata, is only mentionedin Heine’s classification. In the most current classification by Ethnologue,Nata, Isenye and Ikoma share the same code (E.45) and ISO number (ISO-639-3: NTK). These three lects together constitute an east-west Ikoma-10Nata-Isenye cline as shown in Map 1 below. Present-day Nata is borderedto the west by Isenye, to the east by Ikoma, to the north by Ngoreme andKuria, and to the south by the Gurumeti river and the Serengeti NationalPark.Map 1: The Ikoma-Nata-Isenye cline1.4.2 Dialect continuum and Nata neighboursDuring Nata data collection, I noticed that speakers exhibited some mi-nor variations in certain sounds (e.g. o=ɣóɣoro vs o=kóɣoro ‘leg’), and inH-tone placements (e.g. e=keróóngoori vs e=kéroongoori ‘porridge’) . Theelders remarked that the differences stem from a split between a Southerndialect and a Northern dialect. However, these differences are not pre-dictable based on a Southern/Northern split.Nata is mutually intelligible with languages of zone E which are found inthe Kuria Group, e.g. Kuria, Isenye, Ikoma, Zanaki, Ikizu, Sizaki and Shashito mention a few. There is controversy about whether these are dialects ordifferent languages. I hold the view that Nata, Ikoma, and Isenye are closelyrelated dialects, and are distinct from the rest. Hill et al. (2007:42) show,11for instance, that Ikoma and Nata share 89% lexical similarity, Ikoma andIsenye 85%, and Nata and Isenye 88%. It should be noted, however, thatthe major linguistic differences between the three lects (Nata, Ikoma, andIsenye) do not relate to lexical items, but rather to the phonology; e.g.Dahl’s Law (a voicing dissimilation rule where a voiceless obstruent be-comes voiced when immediately followed by a syllable with another voice-less obstruent, Meinhof (1932)) applies differently, and tonal melodies andvowel harmony rules are also different (cf. Aunio 2010 and Higgins 2011).Shetler (1996) gives two views on migration and settlement of the KuriaGroup in the present-day Mara Region. One view is that south Mara groupsall came together to Mara from the Great Lakes (Nyanza or Lake Victoria).According to this view, the Great Lakes family was one big language familythat included the Haya, Kerewe and the Jita-Kwaya-Ruuri languages, whichnow share few cognates with Nata–Ikoma–Isenye. Shetler contends thatdue to increasing pressure on the land, the Kuria group started moving eastand split into small groups. The groups which crossed the Mara river andproceeded to the east became what is today the east Mara, Lagooli-Kuria;and the group that proceeded south of the Mara river formed the Nata-Ikoma-Isenye cline. Shetler also presents an alternative view supported bythe indigenous people that the Ikoma and Temi (Sonjo) were one group thatlived at the eastern side of the Serengeti Plain (Arusha), and that the Ikomabroke off and traveled west across the Serengeti Plain. According to thisview, later the group was further divided into the present-day Ikoma, Nata,Isenye and Ngoreme tribes. This happened around 300-400 AD. Maho’s(2003) classification, where he collapses Guthrie’s E.20 and E.40 groupsinto JE40, may have possibly followed this historical path.There are notable differences in cultural practices among these groups,especially as regards marriage ceremonies, male circumcision, dance forms,and musical instruments. Shetler (1996:12) argues that although the Bantupeople speaking these lects claim that they are separate languages, linguis-tically and culturally they are closely related to each other, suggesting thatthe Nata-Ikoma-Isenye people had a common heritage in the past.121.4.3 Language endangermentNata is classified by the Ethnologue as a language that is in trouble or”threatened”, as the intergenerational transmission is in the process of be-ing broken. This corresponds to yellow on the language cloud display athttp://www.ethnologue.com/cloud/ntk.Although Nata is presently used for communication at a family level,the language is partially used by the child-bearing generation but not bychildren. Children and adolescents can understand Nata but they alwaysspeak Swahili to their parents and peers. There are a number of factors thatpresent a threat to the future of Nata. One is language contact, particularlywith Swahili, which is a dominant language. The second one is the lack ofpolicies that support minority language learning. This has led children andadolescents to pay no attention to minority languages; as a consequencethey only speak Swahili both when socializing with their peers and at home.As noted in Mekacha (1993) and Hill et al. (2007), Nata children growup speaking Swahili as their first language. This change has been drasticand rapid: within the same family, siblings in their 30s have Nata as theirfirst language, while siblings in their late teens have Swahili as their firstlanguage. Hill et al. (2007:34) note that Nata “has a certain amount ofprestige attached to it by its speakers, but […] its functional value anduse are declining, possibly signalling the death of the language variety inthe coming generations.” Nata is at a tipping point, and timely documen-tation is needed to stabilize and revitalize the language. In terms of theendangerment status of related lects, the rate of endangerment is higherfor Isenye than Ikoma, based on the fact that Ikoma has many more speak-ers (19,000), than Isenye (8,000). By comparison, Nata has only 7,000speakers (see Muzale and Rugemalira 2008).1.4.4 Previous literature on NataNata is an understudied language. There are three theses on Nata that Iknow of: Mekacha’s (1985; 1993) on the phonological processes affect-ing Nata vowels and the sociolinguistic impact of Swahili on Nata, respec-13tively, and Johannes (2007) on the Nata nouns structure. Currently, AndreiAnghelescu is writing a PhD thesis on Nata tone and vowel harmony.Other existing publications on the language include a few papers pub-lished as Qualifying Papers (from UBC) or conference proceedings. Theseinclude: Gambarage (2012) and Gambarage (2013) on augmented andunaugmented N-stems in Nata and vowel harmony in Nata, respectively;The morphophonology of tone in Nata by Anghelescu (2013); Verbal mor-photonology of Nata by Lam (2013); and Quantification and Freedom ofChoice in Nata by Osa-Gomez (2016). There are also several book chap-ters in the recently published OUP volume Africa’s Endangered Languages:Documentary and Theoretical Approaches. This book included four chapterson Nata: Gambarage and Pulleyblank (2017) on vowel harmony; Déchaineet al. (2017) on deverbal nouns; Anghelescu et al. (2017) on tone; and Gam-barage et al. (2017) on documenting Nata. Lastly, the UBC Nata WorkingGroup (NaWoG) is working to produce a record of the grammar of Nata inthe form of a descriptive grammatical sketch.1.4.5 Orthography and transcriptionsNata does not have a standard orthography. As a member of the Nata com-munity, I am making efforts to assist the Nata community to develop andstandardize an orthography which would best represent the sound systemof the language. As noted by Mekacha (1993), Swahili has a detrimen-tal effect on local languages, such that speakers of Nata think that Natasounds are similar to Swahili ones. Since tense /e,o/ and lax mid /ɛ,ɔ/vowels contrast and have a high functional load, I have proposed to markthe lax mid vowels with a subdot in the orthography, hence <ẹ,ọ>. Thesubdot system is very common amongst Africanists from west Africa. Thisis in line with the Summer Institute of Linguistics (SIL) efforts in the Mararegion to promulgate the spelling systems for the local languages (see Hig-gins 2010). Special graphemes used in Nata are presented in Table 1.1; allother graphemes are common Roman forms.14Table 1.1: Useful orthographic symbolsOrthography IPA. Ex. Orthography Ex. IPA Glossgh ɣ ghóra ɣóra buybh β bhára βára countng’ ŋ ang’amá aŋamá cystng ŋg anguhá aŋguhá tickny ɲ anyáma aɲáma meatọ ɔ rọra rɔŕa seeẹ ɛ rẹra rɛra babysitNata is a tone language. I use the diacritic ( ´ ) to mark high tone, asseen in the examples in Table 1.1. For a discussion of tone realization inNata see Anghelescu et al. (2017).Lastly, in this thesis I use a four-line system when presenting data. Theseconsist of the orthography, a broad phonetic transcription, a morphemegloss and a free translation. Any relevant context of use will appear insquare brackets before the relevant example.1.5 Why study Nata?Nata is a language that is woefully under-researched (compared to the lan-guages of more economically-developed regions). This research presentsan in-depth examination of the syntactic and semantic properties of DPs.Unearthing these aspects will contribute greatly to the typological and theo-retical understanding of the structure of DPs and the syntactic and semanticfeatures encoded in the Nata D system. This dissertation is the first explo-ration that reports on the syntax and semantics of the D in Nata or anyrelated Zone E45 language.Furthermore, Nata augment data provide new evidence for augmentsas Ds that encode existence, as opposed to being inherently definite or spe-cific or lacking a semantic function as claimed in previous accounts. Thissuggests a deep typological split between Bantu languages and languages15like English, whose D system contrasts for definiteness. Crucially, Nata aug-ments compare with D systems in some Salish languages, which introducea whole new area of inquiry regarding the “Bantu-Salish connection”2.Moreover, the current Nata work is not only an investigation and docu-mentation of a particular linguistic phenomenon related to augments, butis also a contribution to the linguistic theory of the human faculty of lan-guage, hence it shapes our understanding of universal grammar (UG). Fur-thermore, the diagnostic tests used in this thesis could be used to test forthe linguistic function of augments in other Bantu languages.Finally, as a Nata speaker-linguist I have always wished to contribute toNata language documentation. This special motivation is based on the factthat I was born and raised in the Nata-speaking community, where for thepast two decades I have witnessed drastic social changes that pushed Natato the verge of extinction. Writing a thesis on Nata is a fulfilment of oneof the many promises I have committed myself to through Nata languagedocumentation.In the rest of this section I outline some basic phonological, morpho-logical and syntactic characteristics which define Nata typologically as aBantu language. Familiarity with these properties will facilitate our under-standing of the proposal I present in the later chapters.1.5.1 Phonology of the augmentHere I discuss the different phonological forms/structures of the augmentas the left-most element of the nominal structure. Unlike Bantu augmentlanguages like Luganda which have only simple V(owel)-type augments(a.k.a initial vowels, see Hyman and Katamba 1993), Nata augments comein different syllable structures. I discuss four augment structures: the V–type augment, the VV-type, the CV–type, and the CVV–type. I start withthe V-type.2. This is not the first work to claim a linguistic connection between widely-separatedlanguage families: there is work on the ‘Romance-Bantu connection’, De Cat and Demuth(2008), as well as a ‘Salish-Japanese connection’ Kratzer (2005), Kiyota (2009).161.5.1.1 The V–type augmentsExamples of V–type augments are given in (13). As one can see, the V-typeaugments are manifested in seven surface/phonetic forms /i e ɛ a ɔ o u/,consistent with Nata as a seven vowel system3.(13) V-augment typesa. i=kj-ɔɔ́de ‘a/the honey badger’ C7b. i=βj-andá ‘(the) intestines’ C8c. e=me-kéra ‘(the) tails’ C4d. e=βí-taβo ‘(the) books’ C8e. ɛ=kɛ-rɛɛ́rɔ ‘a/the relish’ C7f. a=βa-aná ‘(the) children’ C2g. ɔ=mɔ-́rɔrɔ ‘a/the fire’ C3h. o=mo-síimo ‘(the) bone marrow’ C14i. o=βu-sɔɔ́hu ‘(the) greediness’ C14j. u=mw-aaká ‘a/the year’ C3k. u=βw-ɔɔŋgɔ́ ‘a/the brain’ C14Note that augments are manifested with different vowel heights (high,mid and low) and/or tongue root qualities (advanced [ATR] or retracted[RTR]) based on vowel harmony phonotactics. For discussions of Natavowel harmony rules see Gambarage (2013) , Gambarage and Pulleyblank(2017), and Anghelescu (2019).1.5.1.2 The VV–type augmentThe next case concerns the Nata VV augment form which is found withmonomoraic/monosyllabic class 9 nouns, (14a-c) but not with longer N-stems (i.e., disyllabic or polysyllabic), (14d-f):3. In Johannes (2007), it was claimed, based on impressionistic vowel quality, that Nataaugment vowels preceding the prefixes (Cj/Cw) (where C is a consonant) are variants ofhigh vowels (i.e., are the lax high vowels [ɪ ʊ]). Further research has confirmed that thereare no lax high vowels in Nata (see Gambarage (2013, 2017); Gambarage and Pulleyblank2017; Higgins 2011).17(14) The C9 VV vs V augmentsa. áa=n-da ‘a/the stomach’b. aa=n-dá ‘a/the lice’c. aa=ø-swɛ́ ‘a/the fish’d. a=m.borí ‘a/the goat’e. a=ø-kaβirá ‘a/the tribe’f. a=ø-kurú ‘a/the tortoise’For class 9 monosyllabic N-stem cases such as in (14a-c), I propose thatthe augment is a long vowel due to the word minimality constraint in (15):(15) Noun minimality constraint in Nata:A noun must consist of at least three moras.The minimal noun size requirement is three moras in Nata. However,when the interpretive component requires the covert augment to be used,for the reasons I make clear in Chapter 4, the minimal noun size require-ment cannot over-rule this semantic requirement:(16) a. MariaMariáMariat-a-it-iret-a-it-ireNEG-PST-kill-PFVaa=swéaa=ø-swéDET=9-fish‘Maria did not kill any fish.’b. MariaMariáMariat-a-it-iret-a-it-ireNEG-PST-kill-PFVswéø-swé9-fish‘Maria did not kill any fish.’1.5.1.3 The CV–type augmentThe CV-type augment is found with class 5 nouns; class 5 nouns; it has theform ri-, beginning with a trill or tap, (17)4:4. A few monosyllabic class 5 nouns (about 3 in number) occur with the V-type augmentand a long form of the C5 prefix rii-, as in (i).18(17) The CV form ri- with mono- and polymoraic stemsa. rí=i-to ‘a/the leaf’b. ri=i-rú ‘a/the knee’c. ri=i-sɛ ́ ‘(the) cow dung’d. rí=í-βohe ‘a/the stone’e. ri=i-βurúuŋga ‘an/the egg’f. ri=i-kuβáte ‘a/the stalk’In Bantu historical linguistics, it is claimed that the augments and pre-fixes were CVs in Proto-Bantu (see Meinhof 1932; Guthrie 1967-1971;Maho 1999; see Diercks 2010 on Bukusu which still has a CV augmentin most noun classes with an exception in classes 1 and 9 which have a V-type.). If this is correct, then the cases in (17) retain the old augment shape.The prefixes seem to have lost their initial consonant, -r- (cf. De Blois 1970;De Wolf 1971; Williamson 1993; Maho 1999; Ndayiragije et al. 2012)5. Aswe saw for the VV-type, the CV type also would be realized as a covert Din contexts where the covert D is required semantically. See Chapter 4 fordiscussion of such contexts.1.5.1.4 The CVV–type augmentThe CVV-type augment occurs with the class 10 nouns and begins witha voiceless palatal affricate tʃ, (18). While the VV-type only occurs withmonosyllabic stems as a result of the noun minimality constraint given in(15), the CVV prefix is invariable as it occurs with all sorts of class 10 nouns:(i) The V-type with the V-rii- in monomoraic noun stemsa. e=rii-nɔ́ ‘a/the tooth’b. e=rii-kɔ́ ‘a/the kitchen’c. e=rii-sɔ́ ‘an/the eye’It is not clear to me why these nouns still maintain a V-type augment and not the CV-typeas we saw above. Apparently, some grammaticalization process is happening with the class5 augment and prefix but the direction of the change is unknown. Since when the covertaugment is semantically required, it is the only the initial vowel which deletes and not therii prefix, as I will show in Chapter 4, I will treat the rii- element in (i) as a prefix and notpart of the augment.5. In Nata, if there are two syllables containing an /r/ sound, the second /r/ will delete.19(18) The CV form tʃaa with mono- and polymoraic stemsa. tʃáa=ø-ka’ ‘(the) lions’b. tʃáa=n-dá ‘(the) lice’c. tʃaa=ŋ-gɔkɔ́ ‘(the) chickens’d. tʃaa=m-baráhɛ ‘(the) Thomson’s gazelles’The tʃaa= augment will also be realized as a covert augment when usedin semantic contexts requiring the use of a covert D, as (19b) below shows:(19) a. MariaMariáMariat-i-it-iret-a-it-ireNEG-PST-kill-PFVtʃáa=katʃáa=ø-kaD=ø-lions‘Maria did not kill the lions.’b. MariaMariáMariat-i-it-iret-a-it-ireNEG-PST-kill-PFVkakalions‘Maria did not kill any lions.’As I show in this thesis, regardless of the augment structure, all theaugment structures discussed here behave the same syntactically and se-mantically, consistent with the proposal I make in Chapters 3 and 4.1.5.2 Morphology of the augmentIn this section, I present the Nata noun class prefixes, the prefixal materialfollowing the augment. As with many other Bantu languages, noun stemsin Nata are marked with prefixal morphology that corresponds to the nounclass of the nominal. Following the Bantu tradition, a list of the noun classeswith their prefixes in Nata is given in Table 1.2 below (cf. Meeussen 1967;Denny and Creider (1986); Maho 1999; and others). I have included theversion of each nominal without an overt augment, which occurs in certainsyntactically and semantically-defined environments. In chapters 3 and 4, Iwill argue that such nominals need to be licensed by a non-factual operator.20Table 1.2: The Nata N-prefixesCL N-prefix Overt AUG+N Gloss Covert AUG+N Gloss1 /mo-/ o=mo-súβe ‘a/the man’ mo-súβe ‘any man’2 /βa-/ a=βa-súβe ‘(the) men’ βa-súβe ‘any men’3 /mo-/ o=mu-tɛŕɛβi ‘a/the ladle’ mu-tɛŕɛβi ‘any ladle’4 /me-/ e=mi-tɛŕɛβi ‘(the) ladles’ mi-tɛŕɛβi any ladles’5 /re-/ ri=i-βáβa ‘a/the wing’ i-βáβa ‘any wing’6 /ma-/ a=ma-βáβa ‘(the) wings’ ma-βáβa ‘any wings’7 /ɣe-/ e=ɣe-túumbe ‘a/the chair’ ɣe-túumbe ‘any chair’8 /βe-/ a=βe-túumbe ‘(the) chairs’ βe-túumbe ‘any chairs’9 /N-/ a=m-bɔrɛt́ɛ ‘a/the goat’ m-bɔrɛt́ɛ ‘any goat’10 /N-/ caa=m-bɔrɛt́ɛ ‘(the) goats’ m-bɔrɛt́ɛ ‘any goats’11 /ro-/ o=ro-síri ‘a/the rope’ ro-síri ‘any rope’10 /N-/ caa=∅-sirí ‘(the) ropes’ ∅-sirí ‘any rope’12 /ka-/ a=ɣa-síri ‘a/the rope’ (dim.) ɣa-síri ‘any rope’ (dim.)13 /to-/ o=to-sirí ‘(the) ropes’ (dim.) to-sirí ‘any ropes’ (dim.)14 /βo-/ o=βu-kaanɔ́ ‘(the) sesame’ βu-kaanɔ́ ‘any sesame’15a /ɣo-/ ɣo-tuka ‘to dig’ ɣo-tuka ‘to dig’15 /ɣo-/ o=ɣo-twe ‘a/the ear’ ɣo-twe ‘any ear’16 /ha-/ a=há-se ‘a/(the) place(s)’ há-se ‘any place(s)’19 /he-/ e=he-sirí ‘(the) ropes’ (dim.) he-sirí ‘any ropes’ (dim.)20 /ɣo-/ o=ɣo-síri ‘a/the rope’ (aug.) ɣo-síri ‘any rope’ (aug.)21 /ɣe-/ e=ɣesirí ‘(the) ropes’ (dim.) ɣesirí ‘any ropes’ (dim.)Nata has a total of 19 N-classes, but traditionally 21 if the classes 17 koand 18 mo, which I analyze here as prepositions, would be added6. Class15a is an infinitive class; as a nomininal/verbal predicate it appears herewith no augment (see Schadeberg 2006). Note that the augment vowel al-ways agrees with the [+/-back] feature value of the N-prefix vowel. Obvi-ously, this rule does not apply with defective prefixes in C9 and C10 nouns,whose noun prefix has no vowel. The chart does not present all prefixalallomorphs resulting from vowel phonotactic conditions (see Gambarageand Pulleyblank 2017 for a fuller vowel harmony account).6. Classes 17 and 18 in Nata behave as prepositions rather than regular N-classes or locativeclasses. They do not trigger concord or number morphology, and they do not take modifierslike regular N-classes. See chapter 6 for further discussion.21The N-prefix has been argued to have both a descriptive and an evalu-ative function (cf. Fortune 1984; Déchaine et al. 2014). In the descriptivedimension, the N-prefix hosts some semantic concepts (class features) (cf.Denny and Creider 1986; Maho 1999; Contini-Morava 2000 and others)and it also marks number (see Carstens 2001; Déchaine et al. 2014). Inits evaluative dimension, the N-prefixes encode the speaker’s perspective(Déchaine et al 2014, Déchaine and Gambarage (2016), see also Potts 2007for a general discussion of evaluatives). I do not discuss the properties ofthe N-prefixes any further (see Déchaine and Gambarage 2016 for discus-sion). In chapter 3, I return to the issue of noun decomposition and thesyntactic function of the augment.1.5.3 SyntaxFor the purpose of the discussion of augments in Chapter 3, in §.1.5.3.1,I briefly outline what syntactic categories are present in Nata and whichones may take an augment. Then, in §.1.5.3.2 I briefly touch on the issueof agreement, which is crucial in understanding both phrasal and sentencestructures. Finally, in §.1.5.3.3 I finish with the verb structure, which willhelp us to understand the various morphemes that may appear in the verbcomplex.1.5.3.1 Syntactic categoriesNata appears to have the following syntactic categories: nouns, adjectives,possessives, demonstratives, numerals, adverbs and verbs. In Table 1.3, Ilist each category with their possibility of taking an augment. Note thatnouns, adjectives, and possessives may or may not occur with the augmentdepending on the syntactic status as shown in Table 1.37.7. In Chapter 3, I will show that weak quantifiers behave syntactically as a subclass ofadjectives, but strong quantifiers behave differently, eg., consistently they do not take anaugment. Additionally, proper names and pronouns do not take augments. In Chapter 6, Iwill return to the issue of proper names and pronouns where I will discuss the implicationof my proposal for these categories.22Table 1.3: Nata syntactic categories and augment possibilitiesCategory With AUG Gloss Syntactic statusNouns o=mo-súβe ‘a/the man’ argument DPAdjectives o=mo-koro ‘a/the big one’ argument DPPossessives u=w-ane ‘mine’ argument DPCategory Without AUG Gloss Syntactic statusNominals mo-súβe ‘a man’ predicateAdjectives mo-koro ‘big’ modifierPossessives w-ane ‘my’ modifierDemonstratives -nɔ ‘this’ modifierNumerals i-βere ‘two’ modifierAdverbs iɣoro ‘up’ modifierVerbs -tuka ‘dig’ predicateIn Chapter 3, I will give arguments that the syntactic categories occur-ring with the augment, i.e., those labelled traditionally as nouns, adjectives,and possessives, are argument DPs; and categories without an augment arepredicates and/or modifiers.1.5.3.2 AgreementAs an agglutinating language, Nata exhibits exuberant agreement phenom-ena. I discuss the three types of agreement. The first one is concordialagreement, a system of feature sharing (gender and number/φ) betweennouns and modifiers, (20a). The second one is subject-verb agreement–a syn-tactically conditioned co-variation between the subject and the predicateusually involving gender, number, and in some cases person, (20b). Thethird one is object-verb agreement, which involves an anaphoric reading orTopichood, (20c) (see Bresnan and Mchombo 1987b; Contini-Morava 2000;Corbett 2000, 2006; Aikhenvald 2000; Carstens 2000, 2008; Baker 2003;23Zamparelli 2008; Henderson 2006; Giusti 2008; Diercks 2010; Carstensand Diercks 2011; and others)8.(20) Types of agreementa. Concordial agreement[e=ki-ghẹsọ[e=ki-ɣɛsɔ[D=C7-knifeki-rẹ]ki-rɛ]C7-long]ghi-ka-gw-aɣi-ka-gw-áSA7-PST-fall-FV‘A long knife fell.’b. Subject-verb agreemente=ki-ghẹsọe=ki-ɣɛsɔD=C7-knifeghi-ka-gw-aɣi-ka-gw-áSA7-PST-fall-FV‘A knife fell.’c. Object-verb agreementa-gha-(ke)-gw-isia-ɣa-(ke)-gw-isiSA7-PST-(OM7)-fall-FVe=(ki)-ghẹsọe=(ki)-ɣɛsɔD=C7-knifeLit: ‘S/he dropped (it) the knife.’The one difference between subject-verb agreement and object-verbagreement is that the latter is optional while the former is not. However, Iwill not focus on the subject-verb and object-verb agreement types as theyare not central to the current discussion. The N-prefixes and the differentagreement targets on various syntactic categories (adjectives (A), numerals(Num.), quantifiers (Quant.), demonstratives (Dem.), and possessive pro-nouns (Poss.pron.)) are given in Table 1.4 below.8. Carstens (2001, 2008) tries to unify these agreement types using a Chomskyan Agreeaccount.24Table 1.4: Agreement paradigmsI II III IV V VI VII VIIICL N A Num. Quant. Dem. Poss.pron. V.SA V.OMC1 mo- mo- u- u- u- u- a- mo-C2 βa- βa- βa- βa- βa- βa- βa- βa-C3 mo- mo- u- u- u- u- o- o-C4 me- me- e- ɣi- ɣi- ɣi- ɣe- ɣe-C5 (r)i- (r)i- ri- rj- ri- ri- re- re-C6 ma- ma- a- ɣa- ɣa- ɣa- ɣa- ɣa-C7 ke- ke- ki- ki- ki- ki- ke- ke-C8 βe- βe- βi- βi- βi- βi- βe- βe-C9 N-/ø- N/ø- i- i- i- i- e- e-C10 N-/ø- N/ø- i- tʃa- tʃi- tʃi- tʃe- tʃe-C11 ro- ro- ru- ru- ru- ru- ro- ro-C12 ka- ka- ka- ka- ka- ka- ka- ka-C13 to- to- tu- tu- tu- tu- to- to-C14 βo- βo- βu- βu- βu- βu- βo- βo-C15 ko- ko- ku- ku- ku- ku- ko- ko-C16 ha- ha- ha- ha- ha- ha- ha- ha-C19 he- he- hi- hi- hi- hi- he- he-C20 ɣo- ɣo- ɣu- ɣu- ɣu- ɣu- ɣo- ɣo-C21 ke- ke- ki- ki- ki- ki- ke- ke-The chart does not present all allomorphs resulting from vowel phono-tactic conditions (see Anghelescu 2019 for further discussion). Also, in thistable, various agreement prefixes may reflect Dahl’s Law, i.e., some obstru-ents such as k are realized as ɣ before a voiceless obstruent.1.5.3.3 The verb complexThe Nata verb template shown in Table 1.5 is similar in many respectsto that of other Bantu languages. Some exceptions include two elements,25namely the ‘nasal clitic’ (Nasal) and the second object marker (OM2). Dou-ble object constructions are found in a few Bantu languages; and the nasalclitic is only found in Zone E languages. In Table 1.5, slots I and II can befilled by a nasal clitic or NEG.Table 1.5: The Nata verb templateI II III IV V VI VII VIII IXa. (Nasal=) –– SA Tense (OM1) (OM2) Verb (Ext) Asp/FVb. –– (NEG) SA Tense (OM1) (OM2) Verb (Ext) Asp/FVWe will see consistently that object DPs can be dropped when there arecorresponding OMs in the verb, in which case the referents must be famil-iar to the discourse participants. Secondly, as Table 1.5 shows, the nasalclitic and negation must be in complementary distribution as exemplifiedin (21c):(21) NEG and the nasal clitic cannot co-occur [Nata]a. o=mo-subheo=mo-subheD=C1-mann=a-ku-gha-mu-ret-er-an=a-ku-gha-mu-ret-er-aN=SA-FUF-OM6-OM1-bring-APL-FV‘A/the man will bring it to him/her.’b. o=mo-subheo=mo-subheD=C1-manta-a-ku-gha-mu-ret-er-ata-a-ku-gha-mu-ret-er-aNEG-SA-FUF-OM6-OM1-bring-APL-FV‘A/the man will not bring it to him/her.’c. *o=mo-subhe*o=mo-subheD=C1-mann=ta-a-ku-gha-mu-ret-er-an=ta-a-ku-gha-mu-ret-er-an=NEG-SA-FUF-OM6-OM1-bring-APL-FVIntended: ‘A/the man will not bring it to him/her.’26Initial diagnostics show that the nasal has some left-periphery func-tions, such as being a strong assertion marker or sureness marker (see Kotaniand Gambarage 2016). It is also possible that the nasal has some otherfunctions, eg., Focus marking as Brown (2013) observes, or some modalityfunction as Francis (2014) observes. More research is needed to pin downits function.1.6 Thesis outlineThe structure of the remainder of this thesis will be as follows. In Chapter2 I will review the literature in relation to the augment/D phenomena.I revisit various D contrasts and show that none of these are capable ofexplaining the core function of the augment/D in Nata. This leads to myproposal.The syntactic-semantic analysis of the Nata augment is presented in twodistinct chapters. Chapter 3 presents the syntactic proposal for the augmentas D. Chapter 4 presents the semantic analysis of the augment where I claimthat the augment requires a choice function analysis following Matthewson(1999). I compare the Nata augments with the strikingly similar determinersystem in St’át’imcets (Salish) (Matthewson 1998; 1999).In Chapter 5, I extend my analysis and argue that the notion of existenceis pertinent to other Bantu languages with augments also. I conclude that ofthe nine languages I investigated, D systems in eight languages encode thespeaker-oriented notion of existence, namely, Nata, Runtankore-Rukiga,Haya, Luganda, Kinande, Xhosa, Zulu and Bemba. One language, Dzamba(spoken in the Democratic Republic of Congo), encodes novelty-familiarity,hence the notion of existence is not applicable to Dzamba.Chapter 6 is my last chapter in which I discuss the parametric variationof augments in Bantu. I show that some variations can be explained underthe current theory and some augment aspects may be reducible to inde-pendent syntactic variation. This chapter also gives the direction of futureresearch.27Chapter 2The Nata augment: now yousee it, now you don’t!This chapter reviews the literature and establishes the groundwork of themorphosyntactic and semantic features that are traditionally assumed tobe hallmarks of Ds, that may be linked to the Nata augment. The augment,as I argue in future chapters, is a morphosyntactic head instantiatingthe category D (cf. Halle and Marantz 1993; Embick and Noyer 2001;Hornstein et al. 2005). I hold the view that cross-linguistically features oroperations involving Ds may vary from language to language depending onD-feature composition. In natural languages, the D position can be a locusfor: (a) a mass-count distinction (Déchaine et al. 2018), (b) Case (Halpert2012), (c) deictic force (Gillon 2006; Guillemin 2009), (d) definiteness(Heim 1988; 2011; Schwarz 2009; 2012), (e) specificity (Enç 1991) ornone of the above1. Using data from Nata, I argue in this chapter that thefollowing generalizations are upheld in Nata.1. The Bantu overt augment has previously been linked to features of gender and num-ber (see Ndayiragije et al. 2012), Topicality (see Petzell 2003), and Focality (Hyman andKatamba 1993). Ndayiragije et al argue that in Kirundi, vowel harmony between the aug-ment and the N-prefix vowel is a result of an AGREE relation. For a discussion of Natavowel harmony rules between the augment and the prefix see Gambarage and Pulleyblank(2017), also refer to Chapter 1. In Chapters 5 and 6, I return to the notions of Topicalityand Focus after presenting my proposal in Chapters 3 and 4.28(22) Generalizations about the Nata augment:a. The Nata augment is not conditioned by the mass-count contrast.b. The Nata augment is not conditioned by Case.c. The Nata augment is not conditioned by deixis.d. The Nata augment is not conditioned by definiteness.e. The Nata augment is not conditioned by specificity.f. The Nata augment is not a domain restriction element.I show that mass-count, Case, deixis, definiteness, specificity or domainrestriction do not condition the selection of the augment in Nata. In Chap-ters 3 and 4, I will present my grand proposal about the syntactic-semanticfunction of the Nata augment.The chapter is organized as follows. In §2.1 I present the problem, thepuzzling behaviour of Nata augments. This leads to §2.2 where I investi-gate possible accounts and why they do not work for Nata. In §2.2.1 discussthe possibility that the mass-count distinction may force augment choice,in a manner similar to the distinction between count nouns with a non-expletive D and mass nouns with an expletive D in Old French. I show thatthe overt augment cannot be an expletive element. In §2.2.2 I explore thenotion of Case and rule out the hypothesis that the augment can assign Caseto its complement NP/φP. In §2.2.3-§2.2.5 I investigate the semantic fea-tures of deixis, definiteness, and specificity and show that Nata speakers donot switch augments based on deixis, (in)definiteness, or (non-)specificitycontrasts. In §2.2.6 I discuss the notion of domain restriction where I showthat DPs containing an augment may or may not be interpreted via do-main restriction, unlike Skwxwú7mesh deictic Ds and the non-deictic D inOkanagan (Gillon 2006, Lyon 2011). In §2.3 I give a roadmap on how the29augment puzzles can be resolved, which leads to my proposal in Chapters3 and 4. In §2.4 I give a summary and conclusion.2.1 The puzzling behaviour of the Nata augmentI seek to provide a definitive answer to the question of what features un-derlie the contrast between Nata argument DPs with an overt augment andargument DPs without one. Below I provide initial data showing the con-texts in which Nata speakers switch nominal expressions.2.1.1 Sometimes the Nata augment is thereIn declarative sentences, the Nata overt augment is obligatorily presentwhen used in an argument position, such as in (23a); or in argument posi-tion in embedded clauses, such as in (23b).(23) a. o=mo-subheo=mo-súβeD=C1-mana-gha-sẹk-aa-ɣa-sɛk-aSA1-PST-laugh-FV‘A/the man laughed.’b. N-ka-rọr-aN-ka-rɔŕ-a1sg–PST-see-FV[o=mo-subhe[o=mo-súβe[D=C1-mana-ra-sẹk-a]a-ra-sɛḱ-a]SA1-PST-laugh-FV]‘I saw [a/the man laughing].’When the covert augment is used in the cases in (23) they turn out tobe ungrammatical, as (24) shows.(24) a. * mo-subhe* mo-súβeC1-mana-gha-sẹk-aa-ɣa-sɛḱ-aSA1-PST-laugh-FVIntended: ‘A/the man laughed.’30b. *N-ka-rọr-a*N-ka-rɔŕ-a1sg–PST-see-FV[ mo-subhe[ mo-súβe[ C1-mana-ra-sẹk-a]a-ra-sɛḱ-a]SA1-PST-laugh-FV]Intended: ‘I saw [a/the man laughing].’Another environment where the overt augment may be used is in nega-tive sentences with the felicity condition that the DP is associated with anexistential interpretation, as in (25).(25) [Context: The speaker believes that a man who did not laugh exists]a. o=mo-subheo=mo-súβeD=C1-manta-a-sẹk-ireta-a-sɛk-iréNEG-PST-laugh-PFV‘A/the man did not laugh.’b. a-ka-bhugh-aa-ka-βuɣ-aSA1-PST-say-FV[ango[aŋgo[thato=mo-subheo=mo-subheD=mo-subhet-a-a-sẹk-ire]t-a-a-sɛk-ire]NEG-SA1-PST-laugh-PFV]‘S/he said [that a man didn’t laughed].’Finally, the overt augment is also used in post-copula environmentswhere the nominal expression taking the augment always denotes anindividual.[Context: Bahati is a gender neutral name. A woman and a man are stand-ing before us. M is wondering which person is Bahati:](26) a. BahatiBahatiBahatin=o=mo-subhen=o=mo-súβeCOP=D=C1-man‘Bahati is the man.’31b. o=mo-subheo=mo-subheD=C1-manni-weni-weCOP-RELBahatiBahatiBahatiLit: The man is the one who is Bahati.’The puzzle with the cases in (25) and (26) is that the overt augmentmay be not used if these structures have different interpretations, as I showin the following subsection.2.1.2 Sometimes the Nata augment isn’t thereWhile we saw that in negative sentences the overt augment is possible, ina restricted set of negative environments the overt augment is unavailable,and speakers switch to using the covert augment. Note that for the covertaugment to be used, there must be some kind of a non-factual operator to li-cense this element (compare with example (24) above) (see Progovac 1993;Gambarage 2012; 2013; Carstens and Mletche 2015 and many others).(27) [Context: The speaker does not believe that a man laughed]a. mo-subhemo-súβeC1-manta-a-sẹk-ireta-a-sɛk-iréNEG-PST-laugh-PFV‘No man laughed.’b. a-ka-bhugh-aa-ka-βuɣ-aSA1-PST-say-FV[ango[aŋgo[thatmo-subhemo-subhemo-subhea-gha-sẹk-a]a-ɣa-sɛk-a]SA1-PST-laugh-FV]‘S/He said [that a man laughed] (I doubt it).’32(28) [Context:The speaker does not believe that a man laughed]a. #o=mo-subhe#o=mo-súβeD=C1-manta-a-sẹk-ireta-a-sɛk-iréNEG-PST-laugh-PFVIntended: ‘No man laughed.’b. #Makuru#MakuruMakurua-ka-bhugh-aa-ka-βuɣ-aSA1-PST-say-FV[ango[aŋgo[thato=mo-subheo=mo-suβeD=mo-subhea-gha-sẹk-a]a-ɣa-sɛk-a]SA1-PST-laugh-FV]Intended: ‘Makuru said [that a man laughed] (I doubt it).’Finally, the last case involves post-copula environments where the aug-ment seems to be obligatorily absent. The nominal expression in (29a) onlydenotes a property. This contrasts with the nominal argument with a D in(29b), which denotes an individual or entity and is marked as infelicitousin this this context2.(29) [Context: M is describing Bahati’s gender...]a. BahatiBahatiBahatim=mo-subhen=mo-súβeCOP=C1-man‘Bahati is a man.’b. #Bahati#BahatiBahatin=o=mo-subhen=o=mo-súβeCOP=D=C1-man‘Bahati is the man.’2. Note that the copula nasal, which is underlyingly n, may be homorganic to the im-mediately following nasal or consonant if no vowel intervenes between, as (29a) shows.Compare with the example in (26), where no copula nasal assimilation takes place.33In summary, the data presented above give us three classes of nominalexpressions: argument DPs with an overt augment, examples (23)-(26); ar-gument DPs with a covert augment, example (27); and nominal expressionswithout an augment, example (29). These are summarized below:(30) a. Argument DPs with an overt augment: [DP D[...]]b. Argument DPs with a covert augment: [DP D∅[...]]c. Non-argument nominals (no augment): [φP φ[...]]Factors underlying the distribution of nominals above will be exten-sively discussed in Chapters 3 and 4. At this moment we remain agnos-tic about what these nominal distinctions follow from. At the end of thischapter I will give a roadmap on how my proposal in Chapters 3 and 4will explain the syntactic-semantic factors forcing augment choices in Nata.Before presenting such a proposal, here, I consider a range of possible ac-counts to show that none of the previously proposed D accounts can explainwhat conditions the choice of augment in Nata.2.2 Possible accounts and why they don’t workI start each section by giving a brief review of the relevant literature onhow these features are encoded in other D-systems, then I turn to Nata toshow why these accounts cannot explain what conditions the presence orabsence of the overt augment as summarized in (30a)-(30b). I start withthe mass-count distinction.2.2.1 The mass-count contrast does not condition theaugmentI consider whether the partition of the Nata augment system between theovert augment and the covert/null augment is conditioned by factors thatpartition nouns like mass-count or non-count–count as is the case in manylanguages. I show these factors do not force augment choice in Nata. In34some languages, English, for instance, an overt D is obligatory with sin-gular count nouns, (31). However, a bare noun is possible with countplurals (32a), abstract nouns, (32b), mass nouns, (32c), when used in non-unique/non-familiar contexts:(31) Alternation in overt D bare nouns [English]a. A boy is climbing up the treeb. The boy is climbing up the treec. *Boy is climbing up the tree(32) Appearance of plural/abstract/mass nounsa. (The) boys are climbing up the tree.b. (The) truth can be painful to hear.c. (The) milk is white.In some Romance languages, an expletive D–a D with no semanticfunction–is inserted in mass or abstract nouns only to satisfy a syntacticrequirement (Longobardi 1994; Kyriakaki 2014; Déchaine et al. 2018).Déchaine et al. (2018) show that in some varieties of Old French, abstractnouns and mass nouns (count nouns) may take an expletive D. The alterna-tions involve DPs with an expletive D (the cases in a) and DPs with covertD/D-drop (the b cases). D.M and D.F stand for masculine and feminine Ds,respectively:35(33) Abstract Ns in Old French [Déchaine et al., 2018: 171]a. Mesbutsiifvus2PLplestpleasequethatjeo1SGvus2PLdietell//La veritéD.FEM truthvus2PLcunterai1SG.will.tell‘If it interests you, I will tell you my adventure.’(lit.‘tell the truth.’[From Lais de Marie de France, Guigemarv.312–13]b. Entrebetweeneus3PLmeinentmaintainjoiejoymutmuchgrant.great‘They are happy to finally be together.’(lit.‘maintain much joy.’) [From Lais de Marie de France,Chievrefueil v.94](34) Mass Ns in Old French [Déchaine et al., 2018: 171]a. Eandparbysunhisdundonationunt3PL.havele cunreiD.MASC provision‘He had furnished them with provisions.’(lit.‘have the provision’) [From Le voyage de saint Brendan,v.582]b. MaisbutDeusGodnenotvoltwantquethatplusmoredeof.theforsoutsideVenistcamecunreidprovisionpurforsulonlymunmycorsbody‘But God did not want to bring from outside provisions destinedonly to feed me.’[From Le voyage de saint Brendan, v.1583-4]Adopting the hypothesis that nominal arguments are DPs (Longobardi1994, 2001, 2008), Déchaine et al. analyze the overt D which is used tomark definite descriptions here as a non-referring D as it does not involve36a definite interpretation3.Expletive Ds (non-denoting Ds) are obligatory with generic (kind-denoting) expressions in a number of languages. In Greek, (35a), Italian,(35b), and French, (35c) the definite D is used as an expletive D where itreceives a generic interpretation.(35) Expletive D in generics [Kyriakaki 2010: 255; 263]a. *(I)the.MASC.PLelefand-eselephant.MASC-PLlatrev-unadore.PRES-3PL*(ta)the.NEU.PLfistikj-a [Greek]peanut.NEU-PL‘Elephants adore peanuts.’ (Generic subject and object).’b. *(I)the.PLcastor-ibeaver-PLsonobe.3PLintelligent-i. [Italian]intelligent-PL‘Beavers are intelligent.’c. Lesthedodosdodossontareéteints [French]extinct‘Dodos are extinct.’In Nata, the overt augment occurs with count Ns (36)-(37), abstractNs (38), and mass Ns, (39)-(40). Recall from Chapter 1 that the classprefix expresses number morphology (singular/plural), and number mark-ing on mass nouns is interpreted as follows: ‘some amount of X’ appear-ing with singular morphology, and an interpretation akin to ‘X’ or ‘lots3. There are different expletive Ds discussed in the linguistic literature that I will not ex-haust here. One kind is that found in Catalan, German and Brazillian Portugese whichmark body part nouns to express (extended) inalienable possession (see Vergnaud and Zu-bizarreta 1992; Longobardi 1994.). Nata does not have this type of Ds. Another categoryis the polydefinite Ds in Greek and related languages, which I relate to Nata D-doublingconstructions which I discuss in Chapter 3. Expletive Ds are also obligatory with generic(kind-denoting) expressions in a number of languages: Greek, Italian, French etc where thedefinite D receives a generic interpretation (see recently Espinal (2017); Kyriakaki 2014).Obviously, Nata overt Ds do have a semantic function as I argue in Chapter 4.37of X’ appearing with plural morphology (see Borer 2005; Chierchia 1998;Schwarzschild 2006; Wiltschko 2009; Gillon 2010 for discussion in otherlanguages).(36) a. [a=ø-ka-raam][o=ø-mu-ráam][D=C9-pen]i-ka-bhunek-ai-ka-βunek-aSA9-PST-break-FV‘A/the pen broke.’b. *[ø-karaam]*[ø-karaam]C1-pen]i-ka-bhunek-ai-ka-βunek-aSA9-PST-break-FVIntended: ‘A/the pen broke.’(37) a. [a=bha-kári][a=βa-kari][D=C2-women]bha-ka-het-aβa-ka-het-aSA2-PST-pass-FVha-nọha-nɔC16-here‘(The) women passed here.’b. *[bha-kári]*[βa-kari][C2-women]bha-ka-het-aβa-ka-het-aSA2-PST-pass-FVha-nọha-nɔC16-hereIntended: ‘(The) women passed here.’(38) a. [o=bho-bhiihi][o=βo-βíihi][D=C14-lie]bhu-ka-mw-oor-aβu-ka-mw-óor-aSA14-PST-cost-FV‘(The) lies cost him/her.’b. *[bho-bhiihi]*[o=βo-βíihi][D=C14-lie]bhu-ka-mw-oor-aβu-ka-mw-óor-aSA14-PST-OM1-cost-FVIntended: ‘(The) lies cost him/her.’38(39) a. [ri=i-sahẹ][rí=í-saahɛ]D=C5-bloodri-ko-om-ari-ka-om-áSA5-PST-dry-FV‘(An/the) amount of blood dried up.’b. *[i-sahẹ]*[í-saahɛ]C5-bloodri-ko-om-ari-ka-om-aSA5-PST-dry-FVIntended: ‘(An/the) amount of blood dried up.’(40) a. [a=ma-saahẹ][a=ma-saahɛ]D=C6-bloodgha-ghi-itek-aɣa-ɣa-iték-aSA6-PST-spill-FV‘(The) blood/lots of blood spilled.’b. *[ma-saahẹ]*[ma-saahɛ]C3-bloodgha-ghi-itek-aɣa-ɣa-iték-aSA6-PST-spill-FVIntended: ‘(The) blood/lots of blood spilled.’The fact that covert augments are ruled out in contexts which in En-glish, allow bare plurals and bare mass nouns, and which in Old French,allow bare count nouns, provides an argument for lack of alternation be-tween mass and count nouns and/or count vs. non-count nouns in Nata.Furthermore, I do not analyze Nata overt augments as expletive Ds due tothe semantic claim I articulate in Chapter 4, that the overt augments havea particular semantic function. One way to diagnose expletive Ds in Ro-mance and related languages is to use the definite D in contexts in whichDPs do not refer to unique or familiar referents. This cannot serve as a diag-nostic for Nata given that overt augments are not definites as I argue below.Finally, analyzing overt augments as expletive Ds does does not explain thesemantic contrast between overt augments and covert augments in Nata.392.2.2 Case does not condition the augmentIn this section I consider Halpert’s (2012; 2015) arguments that Case con-ditions the augment in Zulu. Points of convergence and departure betweenthe Case analysis and the behaviour of augments in Nata are summarizedbelow:Table 2.1: Similarities and differences between Zulu and Nata DsProperty Halpert’s analysis Nata augmentsLicensing is vP internal 3 3Non-overt AUG are licensed 3 3Augment choice by Case 3 7Licensing by CAUS or APPL 3 7AUGs are semantically vacuous 7 3I review arguments for the Case account then show that Case does notforce augment choice in Nata.Halpert (2012, 2015), following Schütze’s (1997) analysis for Icelandicargument nominals, argues that Zulu has a system of Case corresponding tothe inherent, structural, and quirky Case found in languages like Icelandic.According to Halpert, examples of inherent Case include argument expres-sions with overt augments (41a), and elements that replace the augment,such as pre-nominal demonstratives (41b) or the oblique element thatmarks benefactive objects, kwa-, (41c). Halpert claims that examples ofstructural Case are augmentless ([-A]) nominals, as in (42a). Finally, basedon Halperts, quirky Case (an unpredictable Case morphology which marksnominals but does not license them) corresponds to augment-permittingstructures (those with vowel coalescence); this includes objects of comita-tives, (43a) and certain temporal adverbials, (43b).40(41) Halpert’s inherent Case [Zulu, Halpert 2012: 237; 130; 212]a. A-ngi-m-bon-iNEG-1SG-see-NEGu-muntuAUG-1person‘I don’t see the person.’b. lo1DEMmntwana1childu-ya-ganga1S-YA-misbehave‘This child is misbehaving.’c. u-SiphoAUG-1Siphou-zo-pheka1S-FUT-cookukudlaAUG.15foodkwa-zinganeKWA-10child‘Sipho will cook food for the children.’(42) Halpert’s structural Case [Zulu, adapt. Halpert 2012: 91]a. A-ngi-bon-iNEG-1SG-see-NEGmuntu1person‘I don’t see anybody.’b. *ngi-bona1SG-seemuntu1personIntended: ‘I see a/the person.’(43) Halpert’s quirky Case [Zulu, adapt. Halpert 2012]a. u-MfundoAUG-1Mfundou-dlala1S-playi-bholaAUG-5ballno-muntu [>na+u-muntu]NA.AUG-1person‘Mfundo is playing soccer with someone/the person.’b. u-MlungisiAUG-1Mlungisiu-zo-fika1S-FUT-arrivenga-sonto [>nga+i-sonto]NGA.AUG-5sunday‘Mlungisi will arrive on Sunday.’41Halpert argues that overt augments function as morphological Case li-censors, while augmentless nominals (covert augments in the current anal-ysis) are restricted to vP-internal positions where they can be structurallylicensed via a local relation with a Licensing head L. Thus, according toHalpert’s analysis, in constructions with zero or one external argument,only one nominal (the highest argument) may be licensed, (44).(44) Augmentless licensing via L [Zulu, Halpert 2012: 94]Here L asymmetrically c-commands and licenses the highest (the mostlocal) augmentless nominal argument inside vP. Halpert argues that in neg-ative constructions involving heads that take a specifier argument – CAUSor APPL, as in (45a) – L can license one more augmentless nominal argu-ment. Thus, there is a 1-to-1 mapping between augmentless nominal ar-guments and the licensing heads, which means three or four augmentlessnominals are impossible in Zulu as the ungrammaticality of (45b) shows.Curiously, the second augmentless nominal has to be licensed by V0 throughCAUS or APPL, as illustrated in (46). Halpert argues that the highest aug-mentless argument is licensed by L and Case introduced by APPL/CAUSis passed down to V0 (under feature inheritance (Chomsky 2008; Asarina2011))4.4. In Bantu APPL and CAUS may co-occur. For this Halpert argues that the appearance ofboth CAUS and APPL does not mean that each will independently license an augmentlessnominal; rather, she argues that there is one V0, hence only one argument will be licensedby them.42(45) No triple augmentless nominals [Zulu, Halpert 2012: 108; M.B]a. uSiphoAUG.1Siphoa-ka-fundis-el-iNEG-1SG-teach-APPL-NEGmuntu1persona-bantwanaAUG-2childlutho7thing‘Sipho doesn’t teach (the) kids anything for anyone.’b. *uSiphoAUG.1Siphoa-ka-fundis-el-iNEG-1SG-teach-APPL-NEGmuntu1personbantwana2childlutho7thingIntended: ‘Sipho doesn’t teach any kids anything for anyone.’(46) Case licensed via APPL/CAUS [Zulu, Halpert 2012: 94]Halpert argues further that augmentless nominals are licensed in aprobe-goal (Agree(ment)) fashion with the effect that all augmentless nom-inals must be vP internal. Extending this to conjoint/disjoint data, sheargues that the disjoint morpheme YA in non-negative data is a morpho-logical spell out of L. Just as L probes for augmentless nominals for Caselicensing, Halpert argues that L probes the vP content for an XP to agreewith, and if the vP has no argument, YA spells out as a result of the probefailing to find its goal:43(47) Disjoint morphological marker YA [Halpert 2012: 166]Halpert claims that evidence for (47) comes from predicates with no the-matic subject like weather predicates, (48), as well as constructions whereall of the arguments of the verb have moved out of vP, leaving it empty(i.e., after A-movement has occurred), (49a), in contrast with (49b) whichhas a vP-internal argument:(48) Weather predicates [Adapt. Halpert, 2012: 181]a. ku-ya-banda17S-YA-cold‘It’s cold.’b. *ku-banda17S-coldIntended: ‘It’s cold.’(49) Disjoint/conjoint [Adapted from Halpert, 2012: 142; 194]a. i=qandakD=C5.eggu=SiphoiD=C1.Sipho[u-ya-li-pheka1S-YA-5O-cookti tk VP] (disjoint)‘As for the egg, Sipho is cooking it.’b. u=SiphoD=C1.Siphou-(*ya-li)-pheka1S-(YA-5O)-cooki=qanda (conjoint)D=C5.egg‘Sipho is cooking an egg.’44Halpert concludes therefore that augmentless nominals are argumentsthat need structural Case licensing, (50a); the counterpart augmented Nsare inherently/intrinsically case-marked KPs, (50b)5.(50) Structural and inherent K [Halpert, 230]a. DPD NPb. KPKaugDPD NPHalpert’s assumption that there is morphological Case in Zulu that headsa ‘Kase Phrase’ (KP) is novel to Bantu, but not to other languages (see Lam-ontagne and Travis 1986; Loebel 1994 in the analyses of German, Finnish,Russian, and Spanish DPs)6. However, I do not extend Halpert’s analysisof Zulu augments to Nata for three major reasons: (i) The Nata augmentis not conditioned by Appl/Caus heads (§2.3.2.1); (ii) The elements thatintroduce Halpert’s Case system are missing (§2.3.2.2), (iii) The Nata aug-ment is not semantically vacuous (§2.3.2.3).2.2.2.1 Verb extensions do not license the augmentWhile it is true that the augmentless nominal argument must be licensedvP-internally, the treatment of CAUS and APPL as Case licensers is cross-linguistically puzzling as similar heads cannot be treated as forming a syn-tactic/semantic class with operators that license polarity elements. APPL/-CAUS do not play any licensing role in Nata, which I also believe is the case5. Halpert discusses restrictions on expletive constructions and conjoint/disjoint morphol-ogy in Zulu as providing evidence for her account. Nata lacks such structures. However,see Carstens and Mletshe 2016 for a review of Halpert’s account where they argue that suchrestrictions do not derive from Case but from Focus.6. The KP was proposed first by Lamontagne and Travis (1986) and advanced by Loebel(1994) in the analyses of German, Finnish, Russian, and Spanish, in which Loebel arguesthat K selects for a DP. For instance, she splits up the German article/D der into two sep-arate functional categories: the D d-, which hosts some semantic content (i.e., referentialfeatures), and er, which has a syntactic function, namely K.45in Zulu7. Rather, as I argue in Chapter 3, augmentless nominal argumentsare licensed by a non-factual operator (NEG and others) (see also Chapter6 for various licensing parameters in Bantu).In addition, licensing of covert augments in Nata does not seem tokeep track of either locality or a 1-to-1 mapping between the non-factualoperators and covert augments to be licensed. A single operator in casessuch as (51) can license multiple covert augments in an unselective bindingfashion:[Context: A mentally confused person always makes up stuff. You hearhim hallucinating saying A man is teaching kids some language for a visitorjust at the front door of your house. You go outside to calm him down. Yousay, that’s not true, (51)].(51) Multiple DPs with a polarity D∅ [Nata]mo-subhemo-súβeC1-mant-a-kw-eegh-er-it-a-kw-eeɣ-éer-iNEG-SA1-teach-APPL-FVmu-ghenimú-ɣeniD=C1-visitorbha-anaβa-anáD=C2-childki-ghambọkí-ɣambɔD=C7-language‘No man is teaching any kids any language for any visitor.’In this case, a single NEG c-commands/takes scope over the entire propo-sition in LF where it licenses all the covert augments in the proposition.Nata DP arguments containing a covert augment are at odds with the gen-eral licensing condition proposed by Halpert, (52):(52) [Halpert, p.93]Augmentless nominal generalization (Final):An augmentless nominal argument must be local to a nominal-licensing head.7. See Hyman and Katamba (1993); Carstens and Mletshe (2016); and Cheng and Downing(2009) who define locality as a function of the Focus parameter in Zulu.46I rule out the APPL and CAUS as licensors in Nata. The treatment ofCAUS and APPL as Case licensers is cross-linguistically puzzling and similarheads cannot be treated as forming a syntactic/semantic class with non-factual operators that license polarity elements like NEG.2.2.2.2 No evidence for morphological Case in NataElements used in Halpert’s account as evidence for Case are missing in Nata.Halpert takes the surfacing of the disjoint morphology YA in weather pred-icates as the correlate of L which probes to license structural Case. Datawith weather predicates in Nata do not show signs of probe-goal relationsthat may substantiate a Case theoretic–account in Nata. (53), for instance,contains weather predicates which also lack thematic subjects, however,there is no morphology to signal the probe-goal activity8.(53) Weather predicates [Nata]a. ko-mitit-irekoo-mítit-ireSA20-cold-PFV‘It’s cold.’b. ko-rinde-erekoo-ríínda-ireSA20-gloom-PFV‘It’s gloomy.’Additionally, according to Halpert’s analysis, augment-permitting struc-tures (comitative objects, instrumentals, etc.,) are an exceptional type ofmorphological Case, namely quirky Case, because they have an unusualand unexpected agreement morphology and they do not license nominals.Nevertheless, these structures involve vowel coalescence phonological pro-cesses, with an underlying augment vowel, which happens in many lan-guages irrespective of their Case morphology (see Casali 2003 and others).In Nata, hiatuses involving objects of comitatives are resolved by way ofvowel harmony, and have no bearing on Case, as these are cross-linguisticphonological processes:8. It is not clear whether the subject agreement in (53) is a class 15 or 17 or 20. I will useclass 20 since class 15 in Nata is the infinitive class and 17 is prepositional; see Chapter 6.47(54) Augment-permitting structures [Nata]a. MakuruMakurúMakurun-a-ku-bhar-an-an-aa-ku-bhar-an-aMN-SA1-IMPF-RECIP-FVnu=u=mw-aanana=u=mw-aanawith=D=C1-child‘Makuru is playing with a child.’b. *Makuru*MakurúMakurun-a-ku-bhar-an-an-aa-ku-bhar-an-aFOC-SA1-IMPF-RECIP-FVna= mw-aanana= mw-aanawith= C1-childIntended: ‘Makuru is playing with a child.’We can see that the hiatus (a+u) formed of the oblique structure andthe prepositional vowel is resolved by right-to-left harmony resulting in thehomophonous vowels (u+u) (see Anghelescu 2019 for more vowel hiatuscontexts). This has nothing to do with Case.Halpert remarks that augment-replacing structures such as argumentnominals occurring with pre-nominal DEMs cannot occur with the aug-ment. This seems to support both De Dreu’s (2008) and Carstens and Mlet-she’s (2016) observation that the augment in Zulu has D-like properties.If the augments were Case, we would predict that they could co-occurwith pre-nominal DEM, but they do not. In Nata pre-nominal DEMs alsodo not occur with the augment and the language shows no signs of mor-phological Case. Halpert seems to correctly locate overt augments in thesame structural position as augment-replacing material (e.g., pre-nominalDEMs), where they are in complementary distribution (see Chapter 3), con-sistent with the cross-linguistic treatment of pre-nominal DEMs and Ds, buther analysis fails to identify the augment as a D element:...augment-replacing morphology is selected for in certain con-structions and carries some semantic content. Selection ofaugment-replacing morphology eliminates the need for an aug-ment. [Halpert 2012: 222]48There are two choices for Halpert for augmentless nominal arguments:either they do not contain a D (in which case we get the problem of pred-icate nominals which cannot equal argument DPs (see Zerbian and Krifka2008), or they do (but she does not adopt a covert D option). Halpertmakes an implicit assumption that there is a DP-shell but it is not clear inher analysis what fills the D slot in her analysis in (50) above. In this thesis,I will demonstrate, consistent with previous literature on Zulu (see de Dreu2008; Adams 2010; Carstens and Mletshe 2016; and others), that the Nataaugment is linked to various D-like properties syntactically (see Chapter 3)and semantically (see Chapter 4).2.2.2.3 The augment is not semantically vacuousI agree with Halpert that augments in Zulu, like in Nata, do not contrast fordefiniteness or specificity, as I show in this thesis. However, I do not treatovert augments in Nata or Zulu as semantically vacuous as Halpert claims:I have argued that the augment vowel is essentially a mor-phosyntactic default that does not correspond to any particularmeaning or syntactic configuration. [Halpert 2012: 222]I argue that in the syntax, a covert augment must be licensed by a higheroperator, but that this has a reflex in the semantics. For instance, the inter-pretive contrast between (55a) and (55c) seems to be that the augmentlessnominal in (55c) must be interpreted under the scope of the non-factual op-erator where it yields a non-existential interpretation, while the overt aug-ment in (55a) is associated with an existential interpretation (cf. Matthew-son 1998; Giannakidou 1998, 2006):(55) Interpretive contrast [Zulu, Halpert 2012: 214-215]a. u-XolaniAUG-1Xolaniu-dlala1S-playno-mfana [>na+umfana)]NA.AUG-1boy‘Xolani is playing with a boy.’49b. *u-XolaniAUG-1Xolaniu-dlala1S-playna-mfanaNA-1boyIntended: ‘Xolani is playing with a boy.’c. u-XolaniAUG-1Xolania-ka–dlal-iNEG-1S-play-NEGna-mfanaNA-1boy‘Xolani is not playing with any boy.’In these cases what forces the choice of an augmentless nominal doesnot seem to be structural Case; rather it is when Ds are interpreted withscope under a non-factual operator like NEG. Halpert admits that the con-ditioning factor is NEG but provides no explanation for the effect of NEG:The environment in (281)[55a] is an environment where a coreargument, such as a subject or direct object, would also be re-quired to bear an augment... By contrast, if we place nominalswith augment-permitting prefixes in negated sentences, as in[55c], the augment may now be dropped.[Halpert 2012: 222]Halpert seems to acknowledge the role of non-factual operators suchas NEG in the distribution of augments but adopts a different analysis thatthere is an abstract L (apparently distinct from NEG) that does the licens-ing. It indeed seems that NEG is the relevant operator that licenses theaugmentless nominal. Similar examples are available in Nata as I show inChapter 3. Note that under Halpert’s analysis of oblique arguments, caseslike (55a) are treated as exceptions (i.e., quirky Case). The Case analysisthus fails to provide a unified account for simple cases such as these.Since I cannot find any connection between Case and negation in Nata, Itake non-factual operators as the overt realization of L in Halpert’s account.The implication of my analysis is that L/NEG not only accounts for thesyntactic distribution but also for the interpretive contrast between the twoaugment choices, as I show in the future chapters. I therefore rule out the50Case analysis for Nata augments. In Chapter 5, I provide an alternative andunified analysis for Zulu and Nata augments.2.2.3 Deixis does not condition the augmentIn this section I explore whether the Nata augment is the locus of deic-tic features, namely spatial/temporal and visibility features. I argue thataugments in Nata do not encode deixis.2.2.3.1 Defining deixisIn other languages D is the locus of deictic force (see Hanks 2005; Gillon2006; Guillemin 2007; and many others). Deixis is a way in which “the ref-erence of certain elements in a sentence is determined in relation to either adiscourse participant, or to a specific time and/or place of the discourse (orutterance)”, (Guillemin 2009: 9). In some languages deictic features aremarked on D (Gillon 2006 on Skwxwú7mesh; Guillemin 2007 on MauritiusCreole), while in others, they are introduced by a demonstrative (Hanks2005 on a typology of deictic elements; Lyon 2011 on Okanagan; Gam-barage 2012 on Nata). In yet other languages, deictic features are markedboth on D and on demonstratives (Matthewson 1998 on St’át’imcets).Typically, deictic distinctions include proximity, and this can be eitherspatial or temporal, and visibility (see Matthewson 1998; Gillon 2006). De-spite the range of distinctions among deictic Ds, all deictic elements locatea referent in space/time in relation to discourse participants. In the nextsection I show that Nata augments do not encode deictic features.2.2.3.2 The Nata augment does not encode deixisAs I argued in Gambarage (2012), Nata augments are not deictics; spa-tial deixis is fixed by the deictic demonstratives (DEMs) occurring in post-nominal position9. A table below shows three types of spatial DEMs in9. Nata has a pre-nominal DEM which replaces the augment and appears with no H-tone.The element always marks referents mentioned in the previous discourse, unlike deicticdemonstratives. See Chapter 3 for further discussion.51Nata: proximal, intermediate, and distal. DEMs are marked with a classprefix, which means they agree with the head noun10.Table 2.2: Three-way distinction of deictic demonstratives in NataDemonstrative Agree with singular N Agree with plural Neg. o-mu-kári ‘woman’ C1 e.g. a-βa-kári ‘women’ C2Proximal: -nɔ ú-nɔ ‘this’ βá-nɔ ‘these’Intermediate: -jɔ ú-jɔ ‘that’ βá-jɔ ‘those’Distal: -ri uu-ri ‘that over there’ βáa-ri ‘those over there’As I argued in Gambarage (2012), Nata augments are not deictics, andshow no contrast between proximal and distal spatial features as I showbelow (see also Lyon 2011 for a similar claim about Okanagan Salish).A DP with an overt augment can introduce a referent that is proximal orintermediate to the speaker or distal from the speaker:(56) [Nata]Deictic demonstrative (Subject position)a. o=mu-kario=mu-káriD=C1-womanu-nọú-nɔC1-PROX.DEMa-kaa-n-dọr-a [Proximal]a-kaa-n-dɔŕ-aSA1-PST-1sg-see-FV‘This woman saw me.’b. o=mu-kario=mu-káriD=C1-womanu-yọú-jɔC1-INTERMa-kaa-n-dọr-a [Intermediate]a-kaa-n-dɔŕ-aSA1-PST-1sg-see-FV‘That woman saw me.’c. o=mu-kario=mu-káriD=C1-womanuu-riuu-ríC1-DIST.DEMa-kaa-n-dọr-a [Distal]a-kaa-n-dɔŕ-aSA1-PST-1sg-see-FV‘That woman over there saw me.’10. In the singular forms, but not in plural, the DEMs have lost the prefix consonant, there-fore they only appear with a vowel prefix not a CV one (see de Wolf 1971; Hyman 1999).52(57) Deictic demonstrative (Object position)a. N-ka-rọr-aN-ka-rɔŕ-a1sg-PST-see-FVo=mu-kario=mu-káriD=C1-womanu-nọ [Proximal]ú-nɔC1-PROX.DEM‘I saw this woman.’b. N-ka-rọr-aN-ka-rɔŕ-a1sg-PST-see-FVo=mu-kario=mu-káriD=C1-womanu-yọ [Intermediate]ú-jɔC1-INTERM.DEM‘I saw that woman.’c. N-ka-rọr-aN-ka-rɔŕ-a1sg-PST-see-FVo=mu-kario=mu-káriD=C1-womanuu-ri [Distal]uu-ríC1-DIST.DEM‘I saw that woman over there.’Notice that the longer the demonstrative vowel, the farther the referent isfrom the speaker (e.g., uuu-ri is ‘way over there’).Furthermore, Nata augments do not encode a contrast in visibility asin Downriver Halkomelem determiners (cf. Wiltschko 2009 and referencestherein) or the St’át’imcets D system (Matthewson, 1998) which contrastfor a discourse referent present and visible at the location of the utterancevs. a referent not visible at the location of the utterance. Nata overt aug-ments can introduce a referent that is visible to the speaker (58a), or totallyinvisible to the speaker (58b).(58) a. o=mu-sẹkẹẹnyao=mu-sɛkɛɛnyáD=C1-sando-ri-itek-ireo-ri-itek-ireSA3-PST-spill-PFVha-nọ [Visible]há-nɔC16-here‘Sand spilled over here.’53b. o=mu-sẹkẹẹnyao=mu-sɛkɛɛnyáD=C1-sandw-o-ɲíw-o-ɲíC-SA-bemuu-mw-ẹẹrí [Invisible]muu-mw-ɛɛríC18-C3-moon‘There is sand on the moon.’These examples show that Nata augments do not encode a visible-invisible distinction. The augments also do not encode sensory evidence(i.e., visual) (cf. Chafe 1986; Garrett 2001; Déchaine 2007; Matthewson2011; Matthewson et al. 2007; Waldie 2012, and others) as this is done bya special particle, ka-, which is adjoined to the demonstrative as shown in(59)11.(59) a. o=mu-kario=mu-káriD=C1-womanka-u-nọka-u-nɔ́VIS-C1-PROX.DEMa-ra-yaar-a [Proximal]a-ra-yáar-aSA1-PRES-run-FV3‘Here comes a/the woman running!’ (Visual)7‘Here comes a/the woman running!’ (Auditory)b. o=mu-kario=mu-káriD=C1-womanka-uu-rika-uu-ríVIS-C1-DIST.DEMa-ra-yaar-a [Distal]a-ra-yáar-aSA1-PRES-run-FV3‘I see a/the woman over there running!’ (Visual)7‘I hear a/the woman over there running!’ (Auditory)I argue that it is the particle ka- that encodes visual evidentiality andnot the augment. The evidence for this is that the augment can be used incontexts where the referent is not visible as I showed above.Finally, I show that temporal deixis is not encoded by Nata augments.An augmented NP can refer to an entity from the past (60a), an entity inthe present (60b) or an entity in the future (60c).11. Some Nata speakers seem to use another strategy which is to replace ka- with oŋ- (C1).In the plural ka- may be replaced with βa-. However, the meaning is the same across thesevariants. Speakers think that the source of variation lies in dialectical differences, i.e.,south vs north. However, the demarcation between the two dialects overlaps considerablyin many respects, as I hinted in chapter 1.54(60) a. KarẹKárɛin.the.pasta=bha-bhyem-ia=βá-βjeem-iD=C2-hunt-FVm-ba-arẹm-ba-arɛ ́C-C2-werebha-aruβa-arúC2-many‘In the past there were many poachers.’b. chaa-ø-sikótʃaa-ø-sikóD=C10-daychi-nọtʃí-nɔC10-thisa=βa-βyem-ia=βá-βjeem-iD=C2-hunt-FVm-ba-suuhum-ba-suuhuC-C2-few‘Nowadays there are few poachers.’c. čaa-ø-sikóčaa=ø-sikóD=C10-dayčí-nɔčí-nɔC10-thisče-kúu-č-ače-kúu-č-aC10-INF-come-FVa=βa-βyem-ia=βá-βjeem-iD=C2-hunt-FVm-ba-ko-bhor-am-ba-ko-bhor-aC-C2-FUT-end-FV‘In the coming days, poachers will be rare.’To sum up, until now we have seen that Nata augments do not encodefeatures encoded in the D systems of other languages such as mass-countdistinctions, Case and deixis. This leads to the conclusion that Nata aug-ments differ from D systems that have those features as part of their Dspecification.2.2.4 Definiteness does not condition the augmentDefiniteness has been claimed to be a universal semantic feature whereit is either expressed within the D system or elsewhere in the grammar(Lyons 1999; Guillemin 2007). While definiteness is a phenomenon foundwith Ds in many languages, including English, I show that Nata augmentsdo not encode definiteness. I argue that definiteness in Nata comes fromelsewhere.552.2.4.1 Defining definitenessThe notion of definiteness is defined variably in the semantic literature.One standard view of definiteness is a novelty–familiarity contrast. I con-sider several other views related to definiteness–presupposition and/or as-sertion of uniqueness, presupposition and/or assertion of maximality–toshow that all these notions do not force augment choice in Nata. Eachsection starts with a literature review.2.2.4.2 The Nata augment does not encode novelty-familiarityThe novelty-familiarity hypothesis introduced by Christophersen (1939)and adapted by Hawkins (1978) and Heim (1982) is one way of definingdefiniteness. Indefinite DPs are argued to be novel to the common groundof the discourse while familiar DPs are familiar to the common ground ofthe discourse (Christophersen 1939; Hawkins 1978; Heim 1982; Ladusaw1979; Matthewson 1998; Schwarz 2009; and others). English is one of thelanguages that encode this distinction in its D system as the examples belowillustrate. Assume (61b) is a follow-up to (61a):(61) a. A/#the police officer stopped me today. [novel]b. What did #a/the police officer say to you? [familiar]The use of the English novel/unfamiliar D a in (61a) does not depend onthe addressee having background information about the referent (i.e., itis not used in discourse anaphoric contexts); however, the familiar articlethe does, as (61b) shows. When this article is used, both the speaker andthe hearer must have access to/knowledge of the discourse antecedent(i.e., the referent can have been mentioned in the previous discourse).Therefore, English the accesses the common ground of the discourse whilea does not. This shows that novelty-familiarity is overtly expressed and iscrucial for D choice in English, as indicated by the infelicitous use of the in(61a) and a in (61b).56Nata augments do not encode this distinction. The only possible can-didate that occurs in affirmative/declarative familiar contexts is the overtaugment; its counterpart, the covert augment, must be conditioned by somenon-factual operator as we saw at the outset. The overt augment appearsboth in familiar and novel contexts, hence there is no familiarity require-ment associated with the use of overt augments. In narratives, the overtaugment is the same in both novel and familiar contexts. Assume (62b) isa continued story from (62a):(62) a. (novel)HayọHayotherekarẹkárɛlong.agon=aarẹ-họn=aarɛ-́hɔSAM=be-LOCo=mu-tẹmio=mu-tɛḿiD=C1-chief‘Long ago there was a chief’‘Once upon a time there was a chief.’b. (familiar)mbembesoo=mu-tẹmio=mu-tɛḿiD=C1-chiefa-ra-kom-aa-ra-kóm-aSA1-PST-gather-FVa=bha-toa=βáa-toD=C2-people‘So, *a/the chief gathered the people.’In these examples there is no familiar-novel distinction that is expressedby the overt augment, as the same augment is used both in novel and indiscourse anaphoric/familiar contexts. I show that the same generalizationholds for non-narrative data; there is a lack of a familiar-novel distinctionin the Nata augment system:[Context: In (63a) a girl is telling her friends about what happened to hertoday. The day after, her friends follow up with (63b) or (63c).](63) a. (novel)Weeche?W-éech-e?2sg-knowRẹẹrọRɛɛrɔ́Todayo=mo-sirikareo=moo-sirikaréD=C1-policea-kaa-ny-imeereria-kaa-ɲí-imeereriSA1-PST-OM-stop‘You know what? A police officer stopped me today.’57b. (familiar)o=mo-sirikareo=mo-sirikaréD=C1-policeu-you-yoC3-DEMni-hení-heCOP-WHa-a-ru-ure?a-a-ru-iré?SA1-PST-come-PFV‘Where did the/that police officer come from?’c. (familiar)u-ka-mu-rọr-au-ka-mú-rɔr-a2SG-PST-OM1-see-FVo=mo-sirikareo=mo-sirikaréD=C1-policewi-ichọ?wa-itʃɔ?of-yesterday‘Dd you see (him) the police officer from yesterday?’We see that the same augment is used regardless of whether the policeofficer is a novel referent or is mentioned in the previous discourse. Theovert augment does not place any constraints on the common ground ofdiscourse. The implications of the lack of the semantic features of defi-niteness is that the Nata grammar uses other tools to express such features.Argument DPs are interpreted as definite when used with materials rein-forcing definiteness like DEMs, as in (63b), or object markers (OMs), suchas in (63c), which shows that the overt D is not a definite D (see Carstens2001, 2008, Adams 2010; and others).2.2.4.3 The Nata augment does not presuppose existenceThe notion of presupposition of existence is cross-linguistically analyzed ona par with the notion of familiarity (Stalnaker 1974; Diesing 1992; Heim1982; Chung and Ladusaw 2004, Matthewson 1998; and others). Presup-position of existence, like the notion of familiarity, also heavily relies onthe common ground of the discourse. This is precisely what automaticallyrules out Nata overt augments as they are not presuppositional elements,unlike Ds in systems like English.One type of presupposition I consider here is logical presupposition de-fined in terms of a semantic relation between propositions as in Soames(1989: 556) quoted in Matthewson (1998: 92)12:12. I will not discuss various versions of presupposition of existence. For instance, logicalpresupposition is also informally defined in terms of a presupposition being able to ‘survive58(64) [Matthewson 1998: 92]Logical presuppositionA proposition P logically presupposes a proposition Q iff the truth ofQ is a necessary condition for P to be true or false.Matthewson (1998) gives the example in (65), again from Soames(1989: 557), which shows that the proposition Q must be true in orderfor P to be assigned a truth value (either true or false).(65) a. P: The queen of England is popular.b. Q: England has a (unique) queen.In (65), if Q is false, the truth value of P cannot be determined (cf. Heimand Kratzer 2010). Presupposition of existence in English is exemplified bythe definite D that must access the common ground of the discourse.Nata augments are not definite Ds, unlike the English definite D. Firstnote that the overt augment can be used in contexts in which a referent ispresupposed to exist, as in (66).(66) [Context: There is only one Ikizu chief known]a. o=mu-tẹmio=mu-tɛmiD=C1-chiefwawaofa=bhi-ikiichoa=βi-ikitʃoD=C2-Ikizu.peoplem-mo-bhem=mo-βéFOC=SA1-cruel‘The chief of the Ikizu people is cruel.’b. Q: Ikuzu people have a chiefMy strong argument for not analyzing overt augments as presupposingexistence comes from the fact that the overt augment can also be used innovel/non-presuppositional discourse contexts, as in (67) repeated fromabove, where the chief is not in the common ground of the discourse:negation’ (see Frege 1997; Soames 1989; Aldridge 1992 and others). Another notion ofpresupposition is expressive presupposition or Strawson’s presupposition, which describes arelation between a sentence S and a proposition Q in context C; see particularly Strawson(1950) and other recent works. For a helpful review of the different types of presuppositionssee van der Sandt (2019).59(67) HayọHayotherekarẹkárɛlong.agon=aarẹ-họn=aarɛ-́hɔSAM=be-LOCo=mu-tẹmio=mu-tɛḿiD=C1-chief‘Long ago there was a chief.’‘Once upon a time there was a chief.’If Nata overt augments are not pressupositional we correctly predictthat they will be used in a variety of other non-presuppositional contexts.Certainly, in (68), the speaker presents their hope that they will find abow to buy, but the utterance does not hold any presupposition of a bow,and yet the overt D is used:[Context: M broke B’s bow and she cannot find a similar bow to replaceit. She articulates her compensation plan:](68) N=ne-gho-ko-ghor-er-aN=ne-ɣo-ko-ɣor-er-aSAM=1sg-2sg-buy-APPL-FVo=bhu‐tao=βu- tae=C14-bow‘I will buy you a bow.’Presuppositional Ds are absent in Nata. The lack of presuppositional Ds inNata can be derived from a negative setting of a Common Ground Param-eter, consistent with Matthewson (1998). Next I show that presuppositionof uniqueness is not part of semantics of the Nata augment.2.2.4.4 The Nata augment does not presuppose uniquenessThe Nata overt augment does not presuppose a unique referent which satis-fies the nominal property. If augments in Nata were Fregean definites thatpresuppose uniqueness, we might expect presupposition failure in contextswhere a referent referred to by the speaker was not the same one in thehearer’s mind. For instance (69a) is challenged in (69b) as the hearer doesnot have a unique man in his/her mind, leading to presupposition failure:60(69) a. A: The man is standing at the door.b. B: I have no idea which man you are talking about.If overt augments presupposed uniqueness, it would not be possibleto have the overt augment used felicitously in context where there is nounique referent under discussion, and yet the overt augment is used, (70).Since the Nata overt D does not introduce referents that are presupposedto be unique, it is infelicitous to reply to () with () as if the D is projectinga presupposition.(70) a. o=mo-subheo=mo-suβeFOC-SA1-be-LOCn-i-i-meer-iren-a-imeer-ireD=C1-manmo=o-ghe-sekumó=ó-ɣe-sekuFOC-SA1-stand-PFV‘A man is standing at the door.’b. #N-ty-eeche#N-tj-eech-é1sg-NEG-know-FVm=mo-subhe=kem=mo-súbhe=keCOP=C1-man=WHo-ku-ghamb-ao-ku-ɣámb-a2SG-IMPFV-talk-FV‘I don’t know which man you are talking about.’c. Ne-we?Ne-weCOP=WH‘Who is it?’The examples in (70) are incompatible with a uniqueness presupposi-tion. The overt augment is felicitous when the hearer is not aware thatthere is only one man in the context. We will see throughout this thesisthat Nata overt augments are freely used in non-unique contexts such asin existential sentences which introduce indefinite DPs or weak quantifiers(see Milsark 1974).612.2.4.5 The Nata augment does not assert uniquenessIt is also plausible to test for definiteness using a Russellian account of as-sertion of uniqueness. Some semanticists consider uniqueness assertionsas hallmark properties of definiteness (Russell 1905; Sharvy (1980); Lyons1999; and many others). First, we find that Nata unique singular referents,such as the sun and moon, are introduced using DPs containing overt aug-ments. In (71) ‘sun’ or ‘moon’ denotes a singleton set, which is implicitlypart of the interlocutors’ common ground knowledge.(71) a. u=mw-ẹẹríu=mw-ɛɛríD=C5-moono-ri-ibhís-ireo-ri-iβís-ireSM3-C5-hide-PFVmu-u-ma-saaromu-u-ma-sáaroLOC-PPF-C6-cloud‘The moon is hiding behind the clouds.’b. a=βa-náataa=βa-naataD=C2-Natam-ba-haa-sáásaam-am-ba-haa-sáásaam-aSAM-C2-HABT-worship-FVi=ry-oobhai=rj-ooβáD=C5-sun‘Nata people worship the sun.’Evidence for the fact that overt augments do not assert uniquenesscomes from considering further Nata data. In (72b), an overt augmentis used in answers that do not assert a unique referent; and in (73) it isbeing used in contexts where there is more than one contextually-salientelement satisfying the nominal property.[Context: The questioner sees the addressee opening the cupboard with avariety of utensils, glasses, and cups and has no idea what the addresseewants to take from the cupboard](72) a. Ne-keNe-keCOP-WHu-kwẹẹnd-á?u-kwɛnd-á?2sg-want-FV‘What do you want?’62b. Ni-kwẹẹnd-áNi-kwɛnd-a1s-want-FVe=ghi ‐kọọmbẹe=ɣi- kɔɔ́mbɛD=C7-cup‘I want a cup.’[Context: There are several cups on a table, equidistant from speaker, Lyon2015: 130, originally from Gillon 2006: 88]13(73) Nu-u-h-ẹNe-u-h-ɛ ́1sg-2sg-give-FVe=ɣi- ‐kọọmbẹe=ɣi- ‐kɔɔ́mbɛD=C7-cup‘Give me a cup.’[Consultant’s comment: I’d pass you one of the cups]In example (73) the Nata language consultants responded by saying thatthey would grab any one of the cups on the table. Similar responses havebeen reported in Gillon’s (2006) and Lyon’s (2015) works. Both contextsshow that the addressee is not talking about a unique cup in context, hencea uniqueness assertion is not encoded. Having seen that Nata augments forsingular referents do not either presuppose or assert uniqueness, now weturn to what happens when plural and mass Ns have overt augments.2.2.4.6 The Nata augment does not presuppose/assert maximalityAfter considering uniqueness for singular DPs, I consider the notion ofmaximality for plural and mass entities (Link 1983; Krifka 2003; andothers) and show that maximality is not a presuppositional or an assertivecomponent of the denotation of augments. If augments presupposed max-imal individuals/entities, they would not be felicitous with plural/massreferents that are not maximal to the hearer without the hearer challengingthe presupposition, (74):13. Note that the Nata consultant’s comment was exactly the same as what Lyon’s consul-tant also said.63[Context: It is very rare to see police officers in this village. Makuru tellshis mom what happened today, (74a). Mom follows up with (74b).].(74) a. RẹẹrọRɛɛrɔ́Todaya=bha-sirikarea=βa-sirikaréD=C2-policebha-ka-ny-imeerer-iβa-ka-ɲ-imeerer-iSA2-PST-1SG-stop-CAUS‘Police officers stopped me today.’b. ni-hẹni-hɛCOP-WHbha-ru-ure?βa-ru-ire?SA2-from-PFV‘Where did they come from?’The augment in the nominal a=bha-sirikare ‘police officers’ does not pre-suppose maximal plural individuals. If this was the case then mom wouldhave challenged the presupposition with a reaction such as All of them? or Ihave no idea which police officers you are talking about, but as mom’s responsein (74b) shows, she is oblivious about where the officers came from.Furthermore, augments used with plural entities do not assert maxi-mality. Nata data show that speakers can use overt augments with pluralDPs to refer to a non-maximal subset of the contextually salient individualssatisfying the nominal property, (75). With (75), the consultants commentthat they would respond by grabbing some of the cups or seek clarificationfrom the speaker on the number of cups the speaker wants, which shows anon-maximal plural sum of cups. Note that class 8 bhi marks plurality.[Context: There are several cups on a table, equidistant from the speaker.Context adapted from Lyon (2011:8)].(75) Nu-u-h-ẹNu-u-h-ɛ ́1sg-2sg-give-FVe=bhi‐kọọmbẹe=βi ‐kɔɔ́mbɛe=C8-cup‘Give me cups.’64With mass nouns, overt augments do not presuppose or assert max-imality either. In (76) the DP o=mo-sori ‘soup’ appears in existentialsentences which introduce indefinite DPs or weak quantifiers:[Context: B just arrived and he wants to eat his lunch first. He says: I hateeating dry rice. His friend has served some soup and left some in the pot.He surprises B by saying (76):](76) o=mo-sorio=mo-sóriD=C1-soupw-o-nyiw-o-ɲiFOC-SA3-bemo=o=nyongomo=a=ɲonŋóLOC=D=C3-pot‘There is soup in the pot.’In (76), the overt augment is used even though the soup is not maximal(the speaker is not talking about all the soup in the context, which rulesout the possibility that overt augment may assert or presuppose maximalreference). Note that overt augments will be fine to be used in contextswhere the speaker might in fact be referring to all the soup in the context(e.g., in contexts where the soup in the pot was the only soup in the con-text), and the speaker may wish his friend to have it all. This propertyaligns with the neutral status of augments regarding definiteness as I arguefurther below. Throughout this thesis I will argue that augments in Nata arenot definites and show that they can only access the speaker’s knowledgebut not the hearer’s, as proposed for many indefinite Ds (see Ionin 2006;Lyon 2011, 2013; Matthewson 1998, 1999, 2001; Gillon 2006; and others).Next, I compare Nata with the weak or strong definite Ds in German.2.2.4.7 Augments are not weak/strong German definite DsSome dialects of German possess a distinction between weak and strong def-inite Ds (see for instance Schwarz 2009, 2012 and references therein, andWiltschko 2013 and references therein)14. Weak definite Ds (or reduced Ds)14. Schwarz (2009) and Wiltschko (2013) list dialects which contrast for weak-strong in-cluding: the Rhineland dialect, Mönchen-Gladbach dialect, Viennese dialect, Cologne di-65are used to refer to unique referents (e.g., sun, moon), as in (77a), or are in-terpreted under situational uniqueness, (78a). The examples come from anAustro-Bavarian dialect (see Wiltschko 2013 and references therein). (I usestrong-weak instead of other labels e.g., full vs. reduced Ds, respectively):(77) Weak D contexts [Austro-Bavarian, Wiltschko 2013: 171]a. DaDETweakMondmoonisisheittodaynetnotzumto.DETweaksegnseen‘The moon isn’t visible today.’b. #DeaDETstrongMondmoonisisheittodaynetnotzumto.DETweaksegnseenIntended: ‘The moon isn’t visible today.’(78) Weak D contexts [Austro-Bavarian, Wiltschko 2013: 170]a. DaDETweakHonsHansisisimin.DETweakSpitoihospital‘Hans is in the hospital.’b. #DeaDETstrongHonsHansisisimin.DETweakSpitoihospitalIntended: ‘Hans is in the hospital.’Strong Ds (full form Ds), on the other hand, have to be used in deicticdemonstrative/pointing contexts, as in (79a), or be used in discourseanaphoric contexts, as in (80b)15.alect, Swiss German dialect, Bavarian dialect, Hessian (Frisian) dialect, Austro-Bavarian,and Standard German.15. Both Schwarz and Wiltschko agree that the choice of definite Ds also correlates withother syntactic factors, not semantics only, e.g., the type of relative clauses. In many, if notall German dialects, the strong Ds are used on restrictive relative clauses while the weak Dsare used on non-restrictive clauses. Nata shows no variation of augments with restrictiveor non-restrictive relative clauses.66[Context: A points to a house (the only one in the immediate surrounding)and asks B:](79) Strong D contexts [Austro-Bavarian, Wiltschko 2013: 173]a. GfoitLikedayoudesDETstrongHaus?house‘Do you like this house?’b. #GfoitLikedayous’DETweakHaus?house‘Do you like this house?’(80) Anaphoric contexts [Standard German, Schwarz 2009: 30]a. HansHanshathaseinen Schriftstellera writerundandeinen Politikera politicianinterviewtinterviewed‘Hans interviewed a writer and a politician.’b. ErHehathas#vom/von demfrom theweak/from thestrongPolitikerpoliticiankeinenointeressanteninterestingantwortenanswersbekommengotten‘He didn’t get any interesting answers from the politician.’The Nata augment system does not seem to encode the weak-strongdistinction for two main reasons. First, as Schwarz illustrates, German Dscontrast for novelty-familiarity, which is not the case with the augment inNata as we saw earlier. Second, considering all the contexts that force Dchoice in German dialects, the overt augment is used in all these contextsin Nata:(81) [Context involving unique referents: overt augment is obligatory]67a. u=mw-ẹẹríu=mw-ɛɛríD=C5-moono-ri-ibhís-ireo-ri-iβís-ireSM3-C5-hide-PFVmu-u-ma-saaromu-u-ma-sáaroLOC-PPF-C6-cloud‘The moon is hidden behind the clouds.’b. * mw-ẹẹrí* mw-ɛɛríC5-moono-ri-ibhís-ireo-ri-iβís-ireSM3-C5-hide-PFVmu-u-ma-saaromu-u-ma-sáaroLOC-PPF-C6-cloudIntended ‘The moon is hidden behind the clouds.’[Context: My friend is wondering where my sister went. I say:](82) In situational unique contexts: overt augment is obligatorya. m-mo-rw-iire.n-mo-ro(r)-ire.FOC-3sg-be.sick-PFV.a-ka-ghia-ka-ɣi3sg-PST-gokw=o=mu-ghabhokw=o=mu-ɣaβoLOC=o=C1-doctor‘She is sick. She went to the doctor.’b. *m-mo-rw-iire.*n-mo-ro(r)-ire.FOC-3sg-be.sick-PFV.a-ka-ghia-ka-ɣi3sg-PST-gokw= mu-ghabhokw= mu-ɣaβoLOC= C1-doctorIntended: ‘She is sick. She went to the doctor.’In discourse anaphoric contexts, as in (84), the same overt augment isused as in discourse new contexts, (83).[Context: B reports to her friends what happened to her on the way home:](83) a. w-eche?w-etʃe?2sg-knowrẹẹrọrɛɛrɔ́todayo=mo-sirikareo=mo-sirikaleD=C1-policea-ka-ny-iimereri (novel)a-ka-ɲi-imerer-iSA1-PST-stop-FV‘You know what? Today a police officer stopped me.’68b. *w-eche?*w-etʃe?2sg-knowrẹẹrọrɛɛrɔ́todaymo-sirikaremo-sirikareC1-policea-ka-ny-iimereria-ka-ɲi-imerer-iSA1-PST-stop-FVIntended: ‘You know what? Today a police officer stopped me.’[Context: The following day, one of her friends follows up:](84) a. u-ka-mu-rọr-au-ka-mu-rɔr-a2sg-PST-OM-see-FVo=mo-sirikareo=mo-sirikareD=C1-policeghwiiki? (familiar)ɣwíiki?again‘Did you see the police officer again?b. *u-ka-mu-rọr-a*u-ka-mu-rɔr-a2sg-PST-OM-see-FVmo-sirikaremo-sirikareC1-policeghwiiki?ɣwíiki?againIntended: ‘Did you see the police officer again?’We saw that strong Ds in German and related dialects are used withdemonstrations, however, in Nata the same overt D is being used.(85) In deictic contexts: overt augment is obligatorya. o=mo-sirikareo=mo-sirikaréD=C1-police.officeru-nọu-nɔC1-DEMm=mu-tatam=mu-tatáCOP=C1-bad‘That police officer is troublesome.’b. * mo-sirikare* mo-sirikaréC1-policeu-nọu-nɔC1-DEMm=mu-tatam=mu-tataCOP=C1-troublesomeIntended: ‘That police officer is troublesome.’We can therefore summarize the differences between the two systemsas follows:69Table 2.3: Distribution of definite strong and weak Ds and Nata DsVariation German German Nata NataWeak D Strong D Overt AUG Covert AUGDefinite D 3 3 7 7Assertion of uniqueness 3 7 3 7Situationally unique 3 7 3 7Anaphoric contexts 7 3 3 7Deictic contexts 7 3 3 7Despite the fact that overt augments are used in discourse-anaphoricenvironments and in demonstration/deictic contexts like strong Ds in Ger-manic dialects, the fact that the same augment is used in weak definitecontexts rules out the possibility that the weak-strong distinction is respon-sible for forcing augment choice in Nata.2.2.5 Specificity does not condition the augmentIt is not the case that Nata augments encode specificity. I present datato show that Nata augments surface in a variety of non-specific contexts,hence ruling out the specificity hypothesis.2.2.5.1 Defining specificityOne of the most common definitions of specificity states that a DP is specificwhen the speaker has a specific individual/entity in mind (Givón 1970,1978; Kamp and Reyle 1993; Hedberg et al. (2009); Matthewson 1998;and others). Compare the cases in (86), which can be said to be specific,with the non-specific cases in (87):(86) a. A friend (of mine) gave me this fidget spinner.b. I will be giving a student a make-up midterm.70(87) [Context: speaker is not referring to any specific entity]a. A wrench will do that job nicely.b. I need a horse for my rodeo practice.The subject DP a friend (of mine) in (86a) and the object DP a student in(86b) refer to a specific individual in the speaker’s mind16. In most lin-guistics literature, the DPs in (87) are not treated as specific indefinites.They have been given various analyses. In some accounts such DPs are in-terpreted with the universal quantifier reading (Eisner 1994; Dayal 1998,2004, Menéndez-Benito 2010); in others they are interpreted under do-main restriction (e.g., the DP denotation is implicitly restricted to a horsewith some identifying property that the speaker has in mind (see Portnerand Yabushita 2001; Schwarzschild 2002; Breheny (2003); Gillon 2006).Yet in others they are accounted for under a choice function approach (seeEbert to appear for a helpful review). I show below that the overt augmentis used in both specific and non-specific contexts such as these, consistentwith the proposal I articulate in Chapter 4.I consider three tests for (non)specificity that can be used to show thatNata augments do not encode specificity. The first one is Enç’s (1991)which is based on the notion that specific DPs introduce referents knownto the speaker (see also Hedberg et al. 2009; Matthewson 1998). Enç arguesthat the specific reading of the DP in the second sentence in (88) is that theDP will pick out two girls who are members of the set of children introducedin the preceding sentence, while the non-specific reading is one in which theDP picks out any two girls. The example in (88) is taken from Matthewson(1998: 41); emphasis is mine:16. Fodor and Sag (1982) claim that the indefinite a in English sentences such as (i) isambiguous between the quantificational reading (where it obeys scope islands) and thereferential (specific) reading and hence they are unusual indefinites (see also Kratzer 1998):(i) Each teacher overheard the rumour that a student of mine had been called before thedean..71(88) A lot of children came in. I knew two girls.The second test is for non-specific DPs which are found in free choiceenvironments, i.e., contexts that give the addressee freedom of choice(89) (see Kadmon and Landman 1993; Alonso-Ovalle and Menéndez-Benito2003; Osa-Gomez 2016). Note that in (89c) the Free Choice Item (FCI) anyis used.(89) [Context: There are a bunch of cups lying on the table]a. M: Pass me a cup.b. W: Which do you want?c. M: Just pass me ANY cup.The third test is to see if augments can be used in characterizing state-ments (i.e., statements that express generalizations about sets of entitiesor situations believed to exist without asserting that such entities exist,(90) (see Carlson and Pelletier 1995; Krifka 2003; Déchaine and Tremblay(2011)).(90) Owls hunt butterflies.I show that augments are fine with all these interpretations hence theycannot be treated as specific.While the English indefinite a seems to be used both in specific and non-specific contexts, and English therefore shows no overt contrast of speci-ficity in its D system, languages like Turkish or Persian allow a specificitycontrast to be encoded morphologically (Enç 1991; Hedberg et al. 2009).The following examples from Hedberg et al. (2009) indicate that Turkishmarks specific direct object DPs with the accusative case marker ; withoutthis marking, object DPs get a non-specific interpretation.(91) [Turkish, Hedberg et al. 2009:4-5]Specificity marking72a. Bugüntodaybironeavukat-lawyer-ACCgör-üyor-umsee-PROG-1SG‘I am seeing a (particular) lawyer today.’b. Bugüntodaybironeavukatlawyergör-üyor-umsee-PROG-1SG‘I am seeing a lawyer today (some lawyer or other).’In a way similar to Turkish, Persian also marks specific direct objectswith the suffix -RA, which is realized as either -o or -ro due to vowel har-mony (see Hedberg et al. 2009):(92) [Persian, Hedberg et al. 2009:5]Specificity markinga. Emruztodayyea/onevakil-(i)-olawyer-I-RAmi-bin-amDUR-see-1SG‘I am seeing a (particular) lawyer today.’b. Emruztodayyea/onevakillawyermi-bin-amDUR-see-1SG‘I am seeing a lawyer today (some lawyer or other).’I show Nata overt augments are neutral with respect to specificity, hencethey are fellicitous in specific and non-specific contexts. Unlike Turkish orPersian-style Ds, augments in Nata are not switched based on the notion ofspecificity17.17. Thanks to Rose-Marie Déchaine (p.c) for shedding more light on this question. Sheobserves that the specificity/referentiality feature is also marked in the Niger-Congo lan-guage family, pointing me to references like Aboh (2004), Ajiboye (2005) and others. Someresearch has shown other semantic features are also encoded. For instance, see Arkoh andMatthewson (2013) and references therein for discussion of the familiar definite D nʊ inAkan (Kwa) but see Bombi (2018) and Owusu and Korsah (2019) for responses to this paper.See also Chapter 5 for discussion about the definite D in Dzamba.732.2.5.2 The Nata augment does not encode specificityNata overt augments appear in specific contexts, as shown in (93), butagain the same augments can appear in non-specific contexts as in (94).The context in (94) is from Ferch (2012):[Context: B: I wish I had a cup. I would drink from this stream. A replies:](93) e=ghi-kọmbẹe=ɣi-kɔɔmbɛD=C7-cupn-ke-nyin-ke-ɲiFOC-C7-bemu-u=n.dọbhọmu-a=n.dɔβɔ́LOC-D=C9.bucketi-yọ [Nata]i-jɔC9-DEMLit: ‘A cup is in that bucket.’‘There is a cup in that bucket.’[Context: Before going to the store, I confirm my shopping list with myroommate].(94) N-to-ko-ghor-aN-to-ko-ɣor-aFOC-2pl-FUT-buy-FVe=ghi-kọmbẹe=ɣí-kɔɔmbɛD=C7-cupnenaande=bhe-chiikọe=βe-tʃíikɔD=C8-spoon‘We need a cup and some spoons.’Nata overt augments fail Enç’s (1991) specificity test which distin-guishes between specific (under discussion) and non-specific (not underdiscussion) DPs. The overt augment contained in the DP that picks out thetwo specific girls (under discussion), (95b), is the same augment used withthe non-specific DP (for not under discussion children), (95a).(95) a. [a=bha-ana[a=βa-ana[D=C2-childbha-mwe]βa-mwe]C2-some]bha-gha-sọh-aβa-ɣa-sɔh́-aSA2-PST-come.in-FV‘Some children came in’.74b. N-ka-mẹɲa=muN-ka-mɛɲ́-a=mu1sg-PST-know-FV=among[a=bha-ana[a=βa-aná[D=C2-childbha-bhere]βá-βere]C2-two]‘I knew two children among them.’As we saw in the previous section about definiteness, example (95)demonstrates that Nata overt augments are neutral with respect to speci-ficity also. The overt augment in a=bha-ana ‘children’ is used in both thespecific and non-specific DP (see similar results in St’át’imcets, a languagewith existence Ds (Matthewson 1998)).Nata augments appear in sentences that allow them to be associatedwith a specific interpretation, (96); however, the same augments also fea-ture with Free Choice Items (FCIs), as in (97), in which a specific readingis unattainable.(96) e=ghi-tabhoe=ɣí-taβoD=C7-bookki-nọki-nɔC7-RELn-gor-irene-ɣor-ire1sg-bought-PFVn-ke-bhen-ke-βéCOP-C7-bad‘A book that I bought is bad.’(97) [Context: A bunch of books are lying on the table]a. chaghor-atʃaaɣor-achoose-FVe=ghi-tabhoe=ɣí-taβoD=C7-book‘Choose a book.’b. chaghor-atʃaaɣor-achoose-FVe=ghi-tabhoe=ɣí-taβoD=C7-bookkyɔ-ky-ɔśɛkjɔ-kj-ɔśɛRED-C7-any‘Choose any/ANY book.’In example (96) e=ghitabho ‘a book’ has an existential interpretation.Note that the DP in (96) is modified by a relative clause which narrows75the domain of the NP and reinforces specificity. In example (97) the sameaugment is used but the referent is non-specific. As Osa–Gómez (2016) il-lustrates, the ambiguous NPI/FCI jọ-j-oọsẹ ‘any’ in (97b) is a FCI functioningas a maximal domain widener, in which case it requires the overt augment.As I show in Chapter 4, this is consistent with my analysis that the NP hasa non-empty reference, hence the overt augment is required18.Lastly, Nata DPs containing overt augments receive a generic interpre-tation when used in generics or characterizing statements, which expressgeneralizations about sets of entities or situations (Krifka 2003; Déchaineand Tremblay 2011). Thus, the DPs in (98) have a non-specific reading:(98) a. cha=ny.ahuumetʃa=ɲ.ahuuméD=C10.owln=chi-haa-byeman=tʃi-haa-βjemaSAM-SA10-HAB-hunte=bhi-bhabhayoe=βi-βaaβáyoD=C8-butterfly‘Owls hunt butterflies.’b. a=bha-kária=βa-káriD=C2-womanm=ba-haa-atim=ba-haa-atiSAM-SA10-HAB-chopchaa=∅-kwétʃaa=∅-kweD=C10-wood‘Women chop wood.’I show in Chapter 4 that the use of overt augments in generics is consistentwith my analysis that DPs containing overt augments may be associatedwith a specific or non-specific interpretation. This explains why genericNP expressions like ‘owls’, ‘butterflies’, ‘women’ and ‘wood’ in (98) take theovert augment even though they do not denote specific referents. Note thatin Nata, and Bantu more generally, generic statements require the habitualaspectual marker, which is argued in Déchaine and Tremblay (2011) to bethe overt realization of the covert generic operator proposed by Chierchia18. For a discussion of the item -ɔ-ɔśɛ ‘any’ see Osa-Gómez (2016), and the discussion inChapter 3 indicating that this item can also be used as an NPI forcing the use of the polarityD. See Krifka (1995) for a related claim that only the stressed ANY in English is a domainwidener.76(1998) and Krifka (2003), and argued to select for the generic reading19.I thus rule out the specificity hypothesis. Having shown that definitenessand specificity are not encoded in the Nata augment system, I turn to theneutral status of the augment with respect to these features.2.2.5.3 The augment is not the English this-specific indefiniteIonin (2006) argues that the English this-indefinite involves a felicity con-dition focusing on the knowledge state of the speaker, i.e., the speakerconsiders only her own view of what’s noteworthy, and not the state of herlistener’s knowledge. Ionin adds that the this-indefinite is specific in thesense of encoding ‘noteworthiness’, a notion similar to referentiality (seeFodor and Sag’s (1982)) or presupposition of existence along the lines ofDiesing (1992). See also Kratzer (1998). Ionin gives the crucial examplefrom Maclaran (1982)):(99) There is this man who lives upstairs from me who is driving me madbecause he jumps rope at 2 a.m. every night (Maclaran:1982:85).I argue that the Nata overt augment is used in many contexts wherethe English this-indefinite is used, however, there are differences betweenthe two. Consider, for instance, the differences and similarities of the twosystems presented below:19. Krifka (2003) discusses the possibility that generic statements also have a kind readingwhich is specific to a genus, e.g., by owls in (98a) the speaker maybe referring to a specificgenus of the Tyto type. This is consistent with the use of the overt augment, although I willnot address the kind analysis in this thesis.77Summary of contexts English thisref Nata overt augmentThere-insertion context 3 3Speaker-knowledge 3 3Used in specific contexts 3 (3)Scope under a non-factual Op 7 7Definite contexts 7 (3)Noteworthy statement required 3 7Uniqueness implicature contexts 7 (3)Since the this-indefinite is essentially a non-deictic demonstrative, wemay think that at least one Nata demonstrative would be its counterpart.On the contrary, it appears that both the Nata pre-nominal and post-nominal demonstratives are ruled out in similar there-insertion contexts,(101), but the overt augment is allowed, (100)20:(100) N=a-nyi-hoN=a-nyi-hoSAM=SA1-be-LOC[o=mo-subhe][o=mo-suβe][D=C1-man]a=bhaa-toa=βaa-toD=C2-peoplebha-ko-bhugh-aβa-ko-βuɣ-aSA2-IMPF-say-FVn-a-haa-turutumb-an-a-haa-turutumb-aSAM-HAB-be.witcho-bho-tikoo-βo-tikoD=C14-night.time‘There is a man; the people say he plays black magic at night.’20. It appears that in Nata, thisref readings are fixed by the element hano ‘here’, which mustoccur with the overt D:(i) N=e-nyiN=e-nyiFOC-1sg-havena-[a=-ø-singorina-[a=-ø-singoriwith-[D=C9-songha-nọ]ha-nɔ]C16-here]yo=o=ko-bhin-er-ayo=o=ko-bhin-er-aof=D=C15-callinga-bha-toa=βaa-toD=C2-people‘I have a song here for calling/attracting people’78(101) a. *N=a-nyi-ho*N=a-nyi-hoSAM=SA1-be-LOC[o=mo-subhe[o=mo-suβe[D=C1-manu-nọ]u-nɔ]C1-this]a=bhaa-toa=βaa-toD=C2-peoplebha-ko-bhugh-aβa-ko-βuɣ-aSA2-IMPFsay-FVn-a-haa-turutumb-an-a-haa-turutumb-aSAM-HAB-be.witcho-bho-tikoo-βo-tikoD=C14-night.timeIntended: ‘There is this man; the people say he plays blackmagic at night.’b. *N=a-nyi-ho*N=a-nyi-hoFOC=SA1-be-LOC[unọ=mo-subhe][unɔ=mo-suβe][DEM=C1-man]a=bhaa-toa=βaa-toC1-this]bha-ko-bhugh-aβa-ko-βuɣ-aD=C2-peoplen-a-haa-turutumb-an-a-haa-turutumb-aSA2-IMPFsay-FVo-bho-tikoo-βo-tikoSAM-HAB-be.witchIntended: ‘There is the/this man; the people say he plays blackmagic at night.’This prohibition for Nata demonstratives must be derived from the factthat Nata demonstratives are definite while the this-indefinite is not.What sets apart the English this-indefinite and the Nata overt augment isthe fact that the former involves a noteworthy property, i.e., a statement ofsomething noteworthy about the individual denoted by the indefinite. Thisis similar to adding descriptive content about the referent of the indefinite.For instance, below the noteworthy property is added by a relative clause(RC):(102) Noteworthy property by RC-modification [Ionin 2006:7]a. #I want to see this new movie.b. I want to see this new movie that my friends have been recommend-ing to me for ages.79c. I want to see this new movie – it’s one that my friends have beenrecommending to me for ages.According to Ionin, (102a) is infelicitous because it does not have anoteworthy statement. The Nata overt augment does not require this con-dition. This follows from the fact that the English this-indefinite is specificwhile the overt augment is neutral with respect to specificity, which cor-rectly predicts that the overt augment will be used both with noteworthystatements and without. I thus rule out the possibility that the Nata overtaugment can be analyzed as a specific indefinite.2.2.5.4 The augment is not the English indefinite aWe saw above that the English indefinite D a can be used in both specificand non-specific indefinite contexts like the Nata overt augment. At facevalue, one may conclude that the augment is the equivalent of indefinitea in English. For instance, one may argue that both the English a, (103),and the Nata augment, (104), are used in non-coreferential contexts wherea speaker is referring to two different entities/individuals.(103) I saw a raccoon in the playground, and I saw another/a raccoon inthe backyard.[Context: Nata and Tiriina are separate locations. Context was adaptedfrom Matthewson (1999) and Lyon (2013)].(104) N-ka-rọr-aN-ka-rɔŕ-a1sg-PST-see-FVa=ma-yaania=ma-yaaníD=C6-gazelleTiiriina,Tiiriina,Tiriina,nanaandn-ka-rọr-an-ka-rɔr-a1sg-PST-see-FVa=ma-yaanía=ma-yaaniD=C6-gazelleNataNaataNata‘I saw a gazelle in Tiriina, and I saw a/another gazelle in Nata.’80However, while in coreferential contexts in English, the definite D is re-quired when referring to the antecedent as in (105b), (see Heim 2011), inNata, the overt augment is used both in non-coreferential (104) and coref-erential contexts, (106), which indicates that the augment does supportcoreference. In (106), there is no different augment choice. As we saw in§.2.2.4, for familiar referents, the OM morphology must be used to markreferents interpreted anaphorically21.(105) a. *John opened a doori. And Sophie closed a doori.b. John opened a doori. And Sophie closed the door/iti.(106) MakuruMakuruMakurua=ki-ighor-aa-ka-iɣ.or-aSA1-PST-open.REVS-FVe=ghe-sek-u,e=ɣe-sek-u,D=C7-door,MasatoMasatoMasatoa-ka-ghi-sẹnchẹk-a,a-ka-ɣi-sɛɲtʃɛk-a,SA1-PST-OM-close.REVS-FVe=ghe-sek-ue=ɣe-sek-uD=C7-door‘Makuru opened a door, and Masato closed (it) the door.’The second argument against analyzing the Nata augment as the coun-terpart of English indefinite a comes from the fact that a can be used eitherunder the scope of non-factual operators, paralleling the NPI any, or it canbe used outside the scope of negation, where it yields an existential inter-pretation:(107) a. I didn’t bring a pen. I forgot it in my drawer.b. I didn’t bring a/any pen. I forgot to buy one.Unlike English, which maintains the same indefinite in both contexts, Nataforces different augment choice: the argument with the overt augment is21. Speakers prefer to drop the overt object DP when the OM is used. But they also acceptthe overt DP if it is preceded by a pause, which suggests that the object DP is dislocated/ina topic position.81used for existential wide scope (108a); the argument nominal with thecovert augment is used for existential narrow scope (108b).(108) a. MakuruMakuruMakuruta-a-ghor-ireta-a-ɣor-iréNEG-PST-buy-PERFe=ghi-tabhoe=ɣí-taβoD=C7-book‘Makuru did not buy a book.’b. MakuruMakuruMakuruta-a-ghor-iréta-a-ɣor-iréNEG-PST-buy-PERFɣí-etaβoɣí-etaβoC7-book‘Makuru did not buy a/any book.’I rule out the hypothesis that the Nata augment is like the English in-definite a. Next, I discuss the notion of domain restriction.2.2.6 The Nata augment is not a ‘domain restrictor’Gillon (2006) and Lyon (2011, 2013) present empirical claims that Ds inSkwxwú7mesh and Okanagan are interpreted via domain restriction: thenotion that the interpretation of a DP or NP which provides the range for aquantifier is contextually restricted (see Westerstahl 1984; von Fintel 1994,1998; Matthewson 2000, 2001 and many others). For instance, both theDPs the man/men in (109) and the QP everyone in (110) are interpreted withrespect to a context set (C).(109) Contextually salient man/men [Gillon 2006: 70]a. The man was laughingb. The men were laughing(110) Contextually salient Q NP [von Fintel 1998: 2; 1994: 28]82a. Everyone had a good time.b. The dinner guests had rhubarb pie for dessert. Everyone de-veloped a rash.With the men, Gillon argues that the speaker does not refer to all the menin the world but rather the men in the context set. Likewise the man doesnot refer to a singleton man in the whole world; rather, it refers to a uniqueman in the discourse context. With (110), von Fintel (1994, 1998) arguesthat the quantifier everyone does not quantify over all the individuals inthe world, but is restricted to individuals who attended the relevant event,(110a) or who had the rhubarb pie, (110b)22.Gillon on deictic Ds in Skwxwú7mesh and Lyon on the determiner iʔin Okanagan argue that the Ds are sensitive to the context of use (domainrestriction). They argue also that the Ds are associated with an implicatureof uniqueness or maximality. Under the Cooperative Principle of conver-sation, conversational implicatures are inferences arising from pragmaticsand are not tied to any structural configurations (Grice 1975; Levinson2000). When speakers are having a conversation, they are tied to conver-sational principles (maxims) which they may either ‘flout’/violate or obey(see Grice 1975; Levinson 1983 for discussion of conversational maxims).While I show that the Nata overt augment is consistently used in all thediscourse contexts supporting Gillon’s (2006) and Lyon’s (2013) analysesof Ds, I show that the Nata overt augment behaves differently from Ds inthese two languages. Consider first the similarities and differences betweenthe Nata augment, the Skwxwú7mesh deictic Ds (d-Ds), and the Okanagannon-deictic context-sensitive domain restrictor D, iʔ.22. I do not explore all the versions/extensions of domain restriction analyses. See West-erstahl 1984 who argues that the itself is a domain restrictor. Gillon (2006) argues thatdefiniteness is not necessary, and that a uniqueness/maximality implicature also providesdomain restriction. Von Fintel (1994, 1998) argues with quantifier data that domain re-striction is provided by a quantificational D. Matthewson (2000, 2001) provides data fromSt’át’imcets arguing that domain restriction is done by a non-quantificational determiner,which co-occurs with the quantificational element, as in “all the students”.83Table 2.4: Similarities and differences in the three languagesAugment/D properties Nata AUG Skw’sh d-Ds Okan. iʔEncodes definiteness 7 7 7Encodes specificity 7 7 7Interpreted via domain restriction (3) 3 3Carries /MAX-implicature 7 3 3Gillon specifically argues that Skwxwú7mesh deictic Ds are sensitive tothe context of use to the effect that they are able to access the commonground of the discourse via domain restriction. Nata augments are fine tobe used with referents established in the context set, however, they do notaccess the addressee’s knowledge in any way. I propose instead that unique-ness or maximality in Nata arises from the morphology (e.g., by using OMs,or DEMs) or is purely pragmatic but does not come from the semantics ofthe augment itself. Another point of departure from treating the augmentas a domain restrictor, which I consider here, is that the Nata augmentsystem exhibits a semantic contrast that is ruled out in Skwxwú7mesh andOkanagan. First I compare then contrast data from Nata and Ds that arealways interpreted via domain restriction.2.2.6.1 Similarities between the Nata Ds and domain restriction DsAt first, the Ds in these languages appear to behave like the Nata overtaugment. Gillon and Lyon illustrate that Ds in these two languages do notpresuppose/assert uniqueness or familiarity and are used in a variety ofcontexts: definite, indefinite, specific, and non-specific. For instance, bothGillon and Lyon show that speakers can use the same Ds in both novel andfamiliar contexts, which is the case we saw for Nata above. A representativeexample comes from Skwxwú7mesh, (111):84(111) Skwxwú7mesh Ds [Squamish, Gillon 2006: 5]a. [Novel]Chen1sg.skw’ách-nexwlook-tr(lc)ti/ta/kwa/kwidetswí7kaman‘I saw a man.’b. [Familiar]Narlkwʼáyhungryti/ta/kwa/kwidetswí7kaman‘The man is hungry.’Furthermore, Gillon and Lyon argue that Ds in these languages areanalyzed with respect to the contextually restricted set (domain restric-tion) with a referent matching the NP description. The following contextadapted from Lyon (2013: 143-144) shows that the Nata overt augment canalso be used in contexts where the DP picks the only referent in the context.[Context: My friend and I are tossing two balls, and my friend throwsthem to me when I go inside. When I came back the balls were gone, I ask,(112a), and he answers, (112b):](112) a. e=me-bhiirae=me-βiiráD=C4-ballni-hẹne-hɛCOP-WHghe-nyí?ɣe-ɲí?SA4-be‘Where are #(the) balls?’b. N-ne-ghi-rẹki-ireN-ne-ɣi-rɛkɛ(r)-ireFOC-1sg-OM8-throw-PFV‘I threw them.’For (112a) to be felicitous the DP in (112a) must be referring tothe balls in context, hence the DP in (a) picks the only balls intro-duced by the context set. The interpretation of the DP e=mebhiira ‘theballs’ in (112a) arises from the contextually salient set (domain restric-85tion). Another example comes from data involving non-interrogative cases:[Context: Makuru is babysitting his younger sibling. Busy mom gets an-noyed by Makuru consistently reporting what the child is doing. Mom: Youkeep pestering and calling (saying)...](113) a. u=mw-aanau=mw-aanaD=C1-childa-ri-iti-irea-ri-ité(r)-ireSA1-PST-spill-PFVa=ma-bhẹẹrẹa=ma-bhɛɛ́rɛD=C6-milk‘The/#a child spilled the milk.’b. u=mw-aanau=mw-aanaD=C1-childa-ri-iti-irea-ri-ité(r)-ireSA1-PST-spill-PFVe=ke-roongoorie=ke-róóŋgooriD=C7-porridge‘The/#a child spilled the porridge.’c. u=mw-aanau=mw-aanaD=C1-childa-a-kọr-irea-a-kɔr-iréSA1-PST-do-PFVhang’u.haŋú.PART.NahọNahɔ́why.noto-ko-mu-tẹm-a?ó-ko-mú-tɛm-a?2sg-FUT-OM1-hit-FV‘The/#a child did this... Why don’t you spank him?’Given this context, the mom must be referring to the same trouble-making kid, the one in the context. Notice that the OM in the last sen-tence, (113c), co-refers with the DP denoting the child in the context, whichmeans this is the same child mentioned in the previous discourse. However,as I have argued in §2.2.4, the augment can in fact be used in contexts thatdo not refer to unique/maximal referents in the context.In quantificational environments in Skwxwú7mesh, Gillon argues thatdomain restriction is provided by the deictic D appearing in the universallyquantified DP:86(114) Squamish Ds [Skwxwú7mesh, Gillon 2006: 100]S-ennom-1sg.sbjmenjustkwélash-tshoot-trí7xwalltadetmex-míxalhredup-bear‘I shot all of the bears.’The Nata overt augment is used in a variety of contexts, therefore is pre-dicted to be used with a contextually restricted set denoted by a universallyquantified DP, such as in (115). Here the DP a=bhaana ‘(the) children’ re-stricts the domain of the universal quantifier, by denoting a contextuallysalient set of children and not all the children in the world. Notice thatwhen the quantifier -ọsẹ is in singular form it denotes the universal quanti-fier every and when it is in plural it is all, and both must co-occur with theaugment23.(115) a. [u=mw-ana[u=mw-ana[D=C2-childw-ọọsẹ]w-ɔsɛ]́C2-all]a-ghi-itẹm-ẹr-aa-ɣa-itɛm-ɛr-aSA2-PST-enter-FVmo=o=nyumbamo=a-ɲuumbáLOC18=D=house‘Every child entered into the house.’23. Nata quantifiers do not sit in D0 position (i.e., do not create a generalized quantifier oftype <t>) as in English (see Chapter 4). However, Nata speakers use the Swahiliform kila ‘every’, pronounced as kira in Nata, which seems to replace the augment andsits in D in object positions, as in (ia). This produces a QP structure as the English ‘every’in every child. Speaker marginally accept the singular Nata quantifier SG-ọsẹ in an objectposition, i(b).(i) a. N-ka-ghamban-aN-ka-ɣámban-a1sg-PST-talk-FVnanawithkira[kira[everymw-aana]mw-aná]C1-child]‘I talked with every child.’b. %N-ka-ghamban-a%N-ka-ɣámban-a1sg-PST-talk-FVnanato[u=mw-aana[u=mw-aná[D=C1-childw-ọọsẹ]w-ɔsɛ]́C1-all]‘I talked to every children.’It seems that the quantifier SG-ọsẹ is undergoing some changes resulting into subject-objectasymmetry but it is not clear to me why.87b. [a=bha-ana[a=βa-ana[D=C2-childbh-ọọsẹ]βa-ɔsɛ]́C2-all]bha-ghi-itẹm-ẹr-aβa-ɣa-itɛm-ɛr-aSA2-PST-enter-FVmo=o=nyumbamo=a-ɲuumbáLOC18=D=house‘All #(the) children entered into the house.’c. N-ka-ghamban-aN-ka-ɣámban-a1sg-PST-talk-FVnanato[a=bha-aana[a=βa-aná[D=C2-childbh-ọọsẹ]βa-ɔsɛ]́C1-all]‘I talked to all #(the) children.’Here, the domain of the quantifier is contextually restricted. The quan-tifier then quantifies over elements/subsets of the range (a DP of type e)which picks out a restricted domain (see Matthewson 2001; Lyon 2013).Finally, the Ds in Skwxwú7mesh and Okanagan are used in sentencesthat may carry an implicature of uniqueness/maximality. Both Gillon andLyon give examples showing that domain-restriction Ds by default carry animplicature of uniqueness/maximality that can be cancelled.(116) Skwxwú7mesh Ds [Skwxwú7mesh, Gillon 2006: 89]Chen1sg.skwélash-tshoot-trta/tsidetmíxalhbearkwidetcheláklh.yesterdayChenrlkw’ách-nexwlook-tr(lc)ta/tsidetchánatthreemíxalh,bearswelhconjnarltl’íw’-numut-witescape-refl-3pl‘I shot a bear yesterday. I saw three bears, but some escaped.’[Context: There was a bowl of berries on the table, but now it is gone. Iask “What happened to the berries?” You reply:](117) Okanagan Ds [Okanagan, Lyon 2013: 143]a. ʔiɬ-əneat-[DIR]-1SG.ERGiʔDETs-p’y’q-aɬqNOM-ripe-fruit‘I ate (all) the berries.’88b. ʔiɬ-əneat-[DIR]-1SG.ERGiʔDETs-p’y’q-aɬq,NOM-ripe-fruitnáxə̌mɬCONJilíʔDEMk’m-xt-m-nexcept-BEN-2SG.OBJ-1SG.ERGiʔDETs-p’y’q-aɬqNOM-ripe-fruit‘I ate some/#the berries, but I saved you some.’Both Gillon and Lyon argue that when the domain restrictor Ds inSkwxwú7mesh and Okanagan are used, the hearer expects the referent(s)to be unique/maximal in any contexts, unless the context rules the uniqui-ness/maximality interpretation out, or the implicature is cancelled. As theyargue, the Ds in Skwxwú7mesh and Okanagan imply the conversational im-plicatures of uniqueness/maximality as their default semantics.In Nata, the overt augment can be used in sentences that may implyuniqueness, (118) or maximality of referents/entities, (119), which maybe cancelled. The example in (118) refers to one gazelle; it is also possiblefor the addressee to think that this was the only gazelle in context, however,the second part cancels that implicature. In (119a) the addressee may thinkthat the speaker is referring to the maximal soup; however, the proportionalreading of (119b) cancels any maximality implicature that the speaker atethe entire bowl of soup.(118) N-ka-ras-aN-ka-rás-a1sg-PST-see-FVa=ma-yaania=ma-yaaníD=C6-gazelleTiiriina,Tiiriina,Tiriina,n-ka-rọr-an-ka-rɔr-a1sg-PST-see-FVe-che-nde,e=tʃe-nde,D.PART=C10-othermaremarebutchi-ka-ng’ọs-achi-kayaaraSA10-PST-escape-FV‘I shot a gazelle in Tiriina, I saw others but they escaped.’[Context: Your friend is preparing some soup. There was a bowl of soupon the table, but now it’s gone. I ask what happened to the soup? Yourfriend replies (replicated from Lyon 2013)]89(119) a. o=mo-sorio=mo-sóriD=C3-soupn=ne-nyw-iren=né-ɲw-ireFOC-1sg-ate-PFV‘I ate some/the soup.’b. o=woo-ndéo=woo-ndéD=C9-anothern=né-wi-it-iren=né-wo-ite(r)-ireFOC-1sg-OM3-spill-PFV‘The rest I spilled (it).’The cancellation of the implicature really shows that the overt augment isnot a maximality operator, otherwise it would force the infelicitous read-ing as with the English determiner the in #I ate the soup, the rest I spilled.In spite of the parallels in the data so far between the three languages,based on some differences shown immediately below I will propose a dif-ferent analysis of Nata, according to which the D does not enforce domainrestriction.2.2.6.2 Differences between the Nata augment and domainrestriction DsWhile the overt augment in Nata is used in domain restriction contexts andin sentences that may imply unique or maximal entities, I do not think thatdomain restriction is part of the representation of the Nata augment as isargued to be the case for the Ds in Skwxwú7mesh and Okanagan. I argueinstead that that the value for the interpretation of the DPs containing theovert augment in these contexts is supplied by the context and it does notcome from the augment itself. I give reasons why the Nata augment is nota domain restrictor element like Ds in Skwxwú7mesh and Okanagan.First, if we consider further Nata data we see that the augment doesnot always refer to referents in a contextually restricted domain. In (120),for instance, the speaker presents their hope that they will buy a cup,even if they do not know where they will find one to buy. Similarly,in Nata, the overt augment is used in cultural assumptions that only90surmise that a referent exists but such referents cannot be analyzed ascoming from a contextually restricted set, (121). The same is true withgenerics/characterizing statements presented above, which use referentsthat do not have context-dependent meaning24.[Context: M broke B’s cup and she is hoping to go to the store. She artic-ulates her compensation plan:](120) N=ne-gho-ko-ghor-er-aN=ne-ɣo-ko-ɣor-er-aSAM=1sg-2sg-buy-APPL-FVe=ɣi‐kọọmbẹe=ɣi- ‐kɔɔ́mbɛe=C7-cup‘I will buy you a cup.’[Context: B is chewing and she bites her lip. B says:](121) o=mo-too=móo-toD=C1-personn-aa-ku-n-gaamb-an-a-ku-ŋ-gaamb-aSAM-SA1-PROG-1SG-talkbhwahẹẹnẹβwahɛɛ́nɛwell/good‘Some person is speaking well of me.’I argue that the augment cannot be a domain restrictor element, sinceupon saying (121), the speaker may have an unrestricted set of possibili-ties about who may be talking about them. For instance, the speaker onlybelieves that such an individual exists but has no idea where they are lo-cated, i.e., whether or not they live in Nata, in Canada or in any part ofthe world. Thus context-dependence cannot be at issue here. In chapter4, I present a proposal that explains the role of the augment in all thesepuzzling contexts25.My strongest reason for parameterizing Nata apart fromSkwxwú7mesh/Okanagan Ds is not because Nata augments utilize24. Nata augments also do not support Etxeberria and Giannakidou’s (2010) view that theD head provides domain restriction or bears deictic features.25. In my understanding Gillon or Lyon do not discuss data involving cultural assumptions.This is a new area of inquiry in relation to D meanings and domain restriction.91domain restriction only in a subset of contexts, but rather because Okana-gan and Skwxwú7mesh Ds do not encode the contrast that forces augmentchoice in Nata. As I argue more elaborately in Chapters 3 and 4, theNata augment has a contrast between the overt augment and the covertaugment relative to non-factual operators. Both Gillon and Lyon illustratethat Ds in these languages can be used in declarative sentences as well asin non-factive environments. A classic example comes from the Okanagandomain-restriction determiner iʔ which can be used under the scope of anon-factive operator as well as outside the scope of such an operator26.(122) The Okanagan iʔ determiner [Okanagan, Lyon 2011: 26)]a. iʔDetsqəltmíxʷmanlutaʔNEGkaʔkíc-ísfind.(DIR)-3SG.ERGiʔDetsənklʼcaʔsqáxaʔhorse‘The man didn’t find any horses.’b. iʔDetsqəltmíxʷmanlutaʔNEGkaʔkíc-ísfind.(DIR)-3SG.ERGiʔDetsənklʼcaʔsqáxaʔhorse‘The man didn’t find the horses.’Gillon (2006) also shows that all deictic Ds in Skwxwú7mesh are equallyavailable in declarative sentences as well as in those with non-factual op-erators. She argues that the non-deictic D, kwi, is the closest candidateto polarity sensitive Ds. However, kwi, like all the other Ds ti/ti/kwa, canoccur in non-factive environments (123b), but can also be used in factiveenvironments (123a)27.(123) Skwxwú7mesh kwi determiner[Skwxwú7mesh, Gillon 2006: 6-7]26. I do not talk about the Okanagan oblique element t, which according to Lyon (2013),is not a determiner but a semantically vacuous morphological reflex of semantic incorpo-ration. This is not relevant for Nata.27. Gillon shows that different DPs in Skwxwú7mesh can take different scope with respectto an operator.92a. Chen1sg.ssilh7-ánbuy-trti/ta/kwa/kwidetsts’úkwi7fish‘I bought a/the fish.’b. Núrl.Qchexw2sg.ssilh7-ánbuy-trkwidetsts’úkwi7?fish‘Did you buy a fish?’c. Háw,Negháwkbe.notsts’úkwi7fish‘No there weren’t any fish.’Gillon concludes that non-factual operators do not force D choice inSkwxwú7mesh, and there is no polarity sensitive D. In contexts where the(strong) NPI reading applies, like (123c), apparently, none of the Ds isused, which is interesting given Gillon’s claim that argument nominals inSquamish are DPs.Nata speakers switch augments in these contexts, as I illustrate moreelaborately in Chapters 3 and 4. In the examples below, Nata differs cru-cially from domain restrictor Ds which are insensitive to the interpretivecontrast I reveal in Chapter 4.(124) a. N-ka-ghor-aN-ka-ɣor-a1sg-PST-buy-FVa=∅.swe [Nata]a=∅.sweD=C9.fish‘I bought a/the fish.’b. N-ty-a-a-ghor-ireN-tj-a-a-ɣor-ire1sg-NEG-SA1-PST-buy-PFVa=∅.swea=∅.sweD=C9.swe‘I did not buy a/the fish.’93(125) a. N-ty-a-a-ghor-ireN-tj-a-a-ɣor-ireFOC-NEG-SA1-PST-buy-PFV∅.swe∅.sweC9.fish‘I did not buy any fish.’b. *N-ka-ghor-a*N-a-ka-ɣor-aFOC-1sg-SA1-PST-buy-FV∅.swe∅.sweC9.fishIntended: ‘I bought a/the fish.’The differences in the semantics of the Nata augments and domain-restriction Ds introduced in the data above separate the Nata augment sys-tem from Skwxwú7mesh/Okanagan Ds. I argue in this thesis that it is ageneral property of Nata overt Ds that they can appear in a variety of con-texts: in specific, non-specific, familiar, novel, context-dependent, or inpragmatic contexts that give rise to conversational implicatures of unique-ness or familiarity. This behaviour is predicted under the analysis I developin Chapter 4, which has to do with speaker-oriented belief in the existenceof a referent.2.3 Solving the Nata puzzle: the two ingredientsTo account for the distribution of both nominals appearing with the aug-ment on the surface, and those without the augment on the surface, I adoptthe hypothesis that the D category in Nata is instantiated by the augment.Two ingredients will be needed. The first ingredient comes from Longob-ardi’s hypothesis that nominal arguments are DPs and predicate nominalsare NPs (Longobardi 1994, 2001, 2008). Longobardi’s proposal is also inline with the second ingredient I propose here; that in the overt syntax andat the meaning level, nominals may vary according to whether the aug-ment is overt or covert (D∅). By covert, I mean that the augment has nophonological content, but is not semantically vacuous. Thus, to understandwhat is forcing augment choice in the puzzling data presented in the outsetof this chapter, three nominal distinctions must be understood: argument94DPs with an overt augment, (126a); argument DPs with a covert augment,(126b); and nominal expressions without an augment, (126c):(126) a. Argument DPs with an overt augment: [DP D[...]]b. Argument DPs with a covert augment: [DP D∅[...]]c. Non-argument nominals (no augment): [φP φ[...]]The two ingredients for solving the Nata augment puzzles derive thesedistinctions. One is the contrast between nominal arguments versus nom-inal predicates; and ingredient two is the semantic distinction between ar-gument nominals appearing with overt augment vs those with the covertaugment. I briefly discuss both ingredients below.2.3.1 Ingredient 1: argument vs predicate nominalsThe hypothesis I adopt is that Nata augmented nominals are DP arguments;they denote entities of type e, (127). On the other hand predicate nominalsdenote a property and lack a DP shell, (128).[Context: Bahati is a gender neutral name. A woman and a man are stand-ing before us. M is wondering which person is Bahati](127) a. BahatiBahatiBahatin=o=mo-subhe [DP=Argument]n=o=mo-subheCOP=D=C1-man‘Bahati is the man.’b. #Bahati#BahatiBahatim=mo-subhe [NP=Predicate]n=mo-súβeCOP=C1-man‘Bahati is a man.’95(128) Context: M is describing Bahati’s gender...]a. BahatiBahatiBahatim=mo-subhe [NP=Predicate]n=mo-súβeCOP=C1-man‘Bahati is a man.’b. #Bahati#BahatiBahatin=o=mo-subhe [DP=Argument]n=o=mo-subheCOP=D=C1-man‘Bahati is the man.’I will argue in the subsequent chapter that the proposal that argumentsare DPs in Nata makes certain correct predictions about the syntactic dis-tribution of augments.2.3.2 Ingredient 2: overt versus covert augmentIn Chapter 4, I will present my core semantic proposal that the choice be-tween the overt augment and the covert augment depends on whether thespeaker believes the noun phrase’s referent exists or not. I will show thatthe Nata overt augment commits speakers to belief-of-existence, (129); andthe covert augment does not commit speakers to belief of existence, (130a).(129) a. o=mo-subheo=mo-súβeD=C1-mana-gha-sẹk-aa-ɣa-sɛk-aSA1-PST-laugh-FV‘A/the man laughed.’b. o=mo-subheo=mo-súβeD=C1-manta-a-sẹk-ireta-a-sɛk-iréNEG-PST-laugh-PFV‘A/the man did not laugh.’96(130) a. mo-subhemo-súβeC1-manta-a-sẹk-ireta-a-sɛk-iréNEG-PST-laugh-PFV‘No man laughed.’b. * mo-subhe* mo-súβeC1-mana-gha-sẹk-aa-ɣa-sɛḱ-aSA1-PST-laugh-FVIntended: ‘A/the man laughed.’The overt augment contrasts with a phonologically null augment whichrequires syntactic licensing by a non-factual operator. Thus, an affirma-tive sentence like (130b) will always be ungrammatical. The details of myproposal are given in Chapters 3 and 4. In Chapter 4, I will argue that theinterpretation of the augment requires an analysis involving choice func-tions (cf. Reinhart 1997; Matthewson 1999).2.4 Summary and conclusionIn this chapter I have argued that the following proposals about the aug-ments are upheld in Nata:(131) a. The Nata augment is not conditioned by mass-count distinc-tions.b. The Nata augment is not conditioned by Case.c. The Nata augment is not conditioned by deixis.d. The Nata augment is not conditioned by definiteness.e. The Nata augment is not conditioned by specificity.f. The Nata augment is not a domain restriction element.97I have ruled out various hypotheses. First, Nata augments are not of theRomance-type in which overt Ds appear as expletive Ds on mass and ab-stract nouns. In Chapter 4, I will argue more elaborately that Nata speakersswitch the augments on semantic basis, and that augments are not seman-tically vacuous. I showed that neither Case nor deixis can condition theaugment. I also argued that the Nata augment is not like English-type Ds asthey do not encode or contrast for definiteness, i.e., augments do not inducea common ground interpretation or presuppose existence or uniqueness. Ihave shown that Nata utilizes OMs and DEMs to derive definite readings.Thus, definiteness and specificity features are not part of the semantics ofthe Nata augment; these features come from elsewhere in the grammar.The Nata augment also cannot be analyzed as a specificity marker as itappears in a variety of non-specific contexts. The position I have taken isthat Nata overt augments are neutral with regard to (in)definiteness and(non)specificity and this explains why they appear in definite/specific aswell as in indefinite/non-specific contexts.I showed that Nata augments are compatible with domain restriction,however, they do not require domain restriction given that augments can beused in contexts that do not contextually restrict the interpretation of DPscontaining them. Definiteness, specificity, deixis, and domain restrictioncome from elsewhere.The exposition of this chapter forms the basis for the theoretical un-derstanding of the syntax and the semantics of augments in the context ofa broad typology of Ds cross-linguistically. In the next chapter I focus onthe syntactic proposal then move to my proposal about the semantics ofaugments.98Chapter 3The Syntax of Nata D3.1 IntroductionOn the basis of data from Nata, I argue in this chapter that the category Dis instantiated by the augment. I provide arguments that support analyzingthe augment as a realization of the functional category D (see similar claimsin Bantu: Dreu 2008; Visser 2008; de Dreu 2008; Giusti 2008; Taraldsen2010; Ndayiragije et al. 2012; Carsten and Mletche 2015). Many of thediagnostics are novel, and the data sets considered cover a wide range ofcontexts, so this study both deepens and broadens our understanding of thesyntax of the augment.The chapter is organized as follows. In §3.2 I discuss the DP internalstructure and present the proposal that augments are Ds. In §3.3 I show thataugments cannot be used in nominal predicates. In §3.4 I give evidence toshow that Ds are obligatorily required in all argument positions. In §3.5 Idiscuss the distribution of polarity Ds and show that they are licensed by anon-factual operator. In §3.6 I talk about areas for future research, and in§3.7, I give a summary and conclusion.993.2 The internal syntax of the Nata DPTraditional approaches to Bantu nouns treat the augment as part of thenoun class prefix (N-prefix), as in (132a) (see Guthrie 1967-71; Meeussen1967; and others). Based on evidence from Nata, I treat a noun suchas (132a) as morphosyntactically complex, that is, as a DP, as shown in(132b).(132) [Nata]a. omo-subheomo-súβeC2-√man‘a/the man’b. o=mo-subheo=mo-súβeD=C2-√man‘a/the man’I first discuss the decomposition of Nata nouns into D=φ-N structure,then I provide arguments that the augment fits to be analyzed as D.3.2.1 The decomposition of the Nata nounI decompose the Nata noun into the lexical part (√N), the N-prefix(number/phi(φ)), here labelled as C(lass)), and the augment (D). I treatthe DP o=mo-subhe a/the man’ in (132b), for instance, as having the lex-ical part/nominal stem -subhe ‘man’, which first merges with φ, and φPmerges with D, (133a). The full DP structure of o=mo-subhe is in (133b).(133) a. DPD φPφ Nb. DPDo=φPφ-mo-N-súβeI treat the augment o= and number/φ -mo- in (133b) as functional ele-ments (cf. Carstens 1991, 2001, 2008; Diercks 2012, 2010; Déchaine et al.2014. See also Borer 2005; Marantz 1997; Wiltschko 2009 and others). I100agree with the Bantuists’ proposal that the internal structure of the Bantunoun projects Number/φ as a functional head of NumP/φP (see Carstens2001; Giusti 2008; Déchaine et al. 2013; Gambarage 2012, 2013; and oth-ers). The rationale for the assumption that number/φ projects as a func-tional head (F0), is based on the fact that number is semantically predictable(cf. Kihm 2005; Carstens 2001, 2005, 2008). Therefore, I choose to usethe label φP/Phi but Num can also be used (Aboh 1998; Morava-Contini2000)1. The choice between these two labels is completely arbitrary. Notealso that in some Bantu accounts, number and class features are analyzedas playing a role in Agree relations, i.e., checking φ-features of agreement[gender, Number] (see Baker 2003; Diercks 2010; Carstens 2001; and oth-ers for discussion). In this thesis I will not investigate these claims, as theywill take us far afield. For an extensive discussion of the descriptive and theevaluative function of N-prefixes in Bantu, see Fortune (1984), Déchaine etal. (2014) and for Nata see Déchaine and Gambarage (2016) and Déchaineet al. (2017). For insights about number see Schwarzschild (2002) and oth-ers. For various lattice approaches to number, see Link (1983), Landman(1991), Chierchia (1998), Rullmann and You (2006), and many others, andfor Bantu see Déchaine et.al. (2014).While (133b) is an example of a DP formed from a non-deverbal noun,I extend the decomposition of the noun into D, φ and N to DPs formedfrom deverbal nouns as well. Deverbal nouns are nouns whose stems arecomposed of a verbal root and a final vowel morpheme (FV)2. For instance,1. There are other proposals about the F0 that projects when the N-prefix merges with thelexical head/L:(i) a. N-prefix projects nP (Ferrari-Bridgers 2008).b. N-prefix projects GenderP (Corbett 1991).c. N-prefix projects NumP (Aboh 1998, Cinque 1995).I do not use the label nP to avoid confusion with the use of this label for other functionalelements, eg., see Déchaine et al (2017) on final vowels. I do not use genderP due tonon-uniform treatments of N-prefixes as gender (see Ferrari-Bridgers (2008) and Carstens(2008).2. I assume with Mudzingwa (2010) that stems and roots are coexistensive in non-deverbalnouns, and that roots and FVs make a stem in deverbal nouns. The root is a radical, it doesnot have internal structure (see Bauer 1983; Good 2005; Gambarage 2011; and others).101the DP o=mokomi ‘a collector’ in (134a) has a verbal root kom- ‘collect’ anda FV i, while the non-deverbal noun in (133b) repeated in (134b) lacks theseformatives.(134) [Nata]a. o=*(mo)-kom-io=*(mo)-kóm-iD=C1-√collect-FV‘a/the collector’b. o=*(mo)-subheo=*(mo)-súβeD=C1-√man‘a/the man’Except for proper names which I discuss in Chapter 6, noun class is oblig-atory for all common nouns, hence both nominal roots and nominal stemsmust be inflected with φ before they take D. Thus, I treat the non-deverbalnoun o=mo-súβe ‘a/the man’ in (135a) as structurally similar to the de-rived noun, (135b)3.(135) a. DPDo=φPφmoNsúβeb. DPDo=φPφmóNkóm-iI use the label N for lexical projections, which may contain FVs. How-ever, for the discussion of types of roots and the contribution of the finalvowel morpheme in Nata see Déchaine et al. (2017)4.The structural position of Nata augments proposed in (135) is consis-tent with the DP hypothesis (Brame 1982; Abney 1987) and its variousextensions (Szabolsci 1987; Déchaine 1993; Longobardi 1994 and others)3. Note that one signature of deverbal nouns is that H-tone is always on the N-prefix; seeDéchaine et al. (2017) and Anghelescu (to appear) for discussion.4. In Déchaine et al. (2017) the deverbal nouns have type-flexible roots (e.g., they areverbal in the context of tense and are nominal in the context of φP, while type-rigid rootscannot be used in the context of tense. Refer to Déchaine et al. (2017) for the implicationsof this claim in relation to Distributed Morphology (Marantz and Halle 1993; Marantz 1997,2013; Embick and Noyer 2001).102in which heads must project to phrases. However, unlike word-level Dsin English and other languages with word-level Ds, the augment has mor-phosyntactic properties of a clitic as I argue below.3.2.2 The augment as a proclitic DOne possible criticism of treating the augment as D may come from the mor-phosyntactic appearance of the augment, i.e., the augment does not appearas a word class determiner like a or the in English. Indeed the augment isnot a word class D. I claim that the augment has the morphosyntactic sta-tus of a proclitic (Meeussen 1967; Van de Velde 2008; van de Velde 2019),a bound element that attaches to a phrase (eg., φP in Nata) (cf. Zwicky1977; Ajiboye 2005, Déchaine 1993). Thus, I make a distinction between aword category and a structural position in a tree that elements with similarstructural status may occupy. I mark the augment with a clitic convention“=” throughout.The dissociation between word classes and functional elements that areF0 is elaborated in Ghomeshi et al. (2009) (see also Lyons 1999):...determiners are assumed to occupy a position fixed by thehierarchy of functional categories, which allows for the disso-ciation between the word class (determiner) and the syntacticposition its members occupy (D). Ghomeshi et al. (2009: 05)The treatment of augments as proclitics, instantiating the category D,may contribute to our understanding of why certain pre-nominal procliticsthat may occupy the D position are in complementary distribution with theaugment. Next, I focus on these distributional properties.1033.2.2.1 The augment does not co-occur with the DEM procliticThe first case that supports the analysis of the augment in Nata as a procliticD comes from the fact that it cannot co-occur with the pre-nominal DEMwhich also attaches to φP as a proclitic D, as in (136a)5:(136) a. u-nọ=mu-kariu-nɔ=mú-kariC1-DEM=C1-womann-aa-ku-yaar-an-aa-ku-yáar-aC-SA1-PROG-run-FV‘The woman is running.’b. *u-nọ=o=mu-kari*u-nɔ=o=mú-kariC1-DEM=D=C1-womann-aa-ku-yaar-an-aa-ku-yáar-aC-SA1-PROG-run-FV’The woman is running.’I propose that the pre-nominal DEM in (136a) sits in D position, (137b).Thus the augment (137a) can only be used if the D slot is not occupied byother D material.(137) a. DPDo=φPmu-kárib. DPDunɔ=φPmú-kariThe one semantic difference between the augment and the pre-nominalDEM in Nata is that the pre-nominal DEM functions as a strong D (i.e.,has anaphoric readings) (cf. Ashton 1944; Carstens 1991, 2008; Van deVelde 2005; 2019; and others) while the augment does not. I will analyzethe augment and the pre-nominal DEM as proclitics that occupy the samestructural position, hence the two cannot co-occur.5. Nata has two demonstratives: the pre- and the post-nominal. Refer to Chapter 6 where Idiscuss the post-nominal DEM which co-occurs with the augment but differs both from thepost-nominal DEM and the augment in phonological and deictic features.1043.2.2.2 The augment does not co-occur with the honorific procliticThe proclitic ɲa=, appears to have the honorific meaning ‘Mr /Ms/Mas-ter/Mistress’ and syntactically, appears to replace the augment. Comparethe augment data in (138) with the examples with the ɲa= honorific in(139):(138) The overt augment positionD=Cl-N Glossa. o=mu-twé (C1) ‘a/the head’b. a=ma-ŋána (C6) ‘(the) words’c. a=ma-βí (C6) ‘(the) poop’(139) The honorific proclitic positionHon=Cl-N Glossa. ɲá=(*o=)mu-twe (C1) ‘Master head’ (big-headed person)b. ɲá=(*a=)ma-ŋana (C6) ‘Master words’ (talkative person)c. ɲá=(*a=)ma-βi (C6) ‘Master poop’ (smelly kid)Déchaine et al. (2014) report on a class of evaluatives in Shona called “hon-orifics” which translate as Mr/Ms, e.g., when mù-kómáná ‘a/the boy’ takesthe honorific prefix va- as in vá-mù-kómáná it renders the meaning ‘Mr.boy’. In their analysis they argue that honorifics in Shona associate to theD position in the syntax. This is consistent with my treatment of the Natahonorific proclitic ɲa as D, which predicts that it will be in complementarydistribution with the augment as in (139). The corresponding structuresfor o=mu-twé ‘a/the head’ and ɲá=ɣu-twe ‘Mr/Ms head’ are in (140):(140) a. DPDo=φPmutwéb. DPDɲá=φPmutwe105The Nata honorific proclitic does not function as an evaluative N-prefixas is the case in Shona, which allows prefix stacking. In Nata, evaluativemeaning is created by substituting an ordinary N-prefix, e.g., o=mu-twé‘head’ (C3), for the evaluative N-prefix as in o=ɣu-twé ‘big bad head’ (C20)(cf. Déchaine and Gambarage 2016). Therefore, both the evaluative andhonorific interpretations are possible at the same time: ɲá=ɣu-twe ‘Master/Mr/Ms big bad head’. I suggest that the honorific prefix and the augmentoccupy the same structural position.3.2.3 Predictions for the proposal that augments are DsThe proposal that the category D in Nata is instantiated by augments makescertain predictions about the syntactic distribution of augments on nom-inals. There is ample evidence in the literature that Ds serve to turn apredicate into an argument (cf. Carlson 1980; Higginbotham 1985; Stow-ell 1989; Longobardi 1994, 2001, 2008; Matthewson 1998; Van de Velde2019). In a series of papers Longobardi provides extensive argumentationin favour of the view that D is required for argumenthood:(141) D and argumenthood (Longobardi 1994: 620, 628)a. A ‘nominal expression’ is an argument only if it is introduced bya category D (p. 620).b. DP can be an argument, NP cannot.I adopt the hypothesis that nominal arguments are DPs (Longobardi1994, 2001, 2008). The Longobardi-style analysis is consistent with thesplit between arguments and predicates in Nata. I show below that pred-icates lack a D, (142a); .and nominal arguments are DPs which may varyin the overt syntax according to whether D is pronounced (overt), (142b)or unpronounced (covert/D∅), (142c). The analysis developed here distin-guishes three types of nominal expressions, as follows.(142) a. Nominal predicates (no augment): φ-Nb. Argument DPs (overt augment): D-φ-N106c. Argument DPs (covert augment): D∅-φ-NI discuss the syntactic distribution of each of these three structures withexamples from Nata starting with predicates such as φ-N, (143a), thenD=φ-N arguments such as (143b), and arguments with a phonologicallynull D such as (143c).(143) The contrast in argument DPs [Nata]a. [φP=predicate]MakuruMakuruMakurum=mw-aanam=mu-anáCOP=C1-child‘Makuru is a child.’b. [DP=argument]MakuruMakuruMakurua-gha-sughut-aa-ɣa-súɣut-aSA-PST-push-FVu=mw-aanau=mu-anáD=C1-child‘Makuru pushed a/the child.’c. [DP=argument]MakuruMakuruMakurut-a-a-sughut-iret-a-a-súɣut-ireNEG-SA-PST-push-PFVmw-aanamu-anáC1-child‘Makuru did not push any child.’On the surface the DP containing the null D in (143c) appears with thesame φ-N structure as a nominal predicate in (143a). I argue that they aredifferent structurally. The seeming φ-N structure in (143c) is an argumentDP containing a null D, which is accessed semantically in LF. This null D re-quires syntactic licensing, i.e., it must be licensed by a non-factual operator.The nominal predicate φ-N structure in (143a), on the other hand, lacks Daltogether. In fact, we will see in Chapter 4 that while (143c) talks aboutentities that the speaker does not believe exist in the universe of discourse,nominal predicates denote some property and never an entity/individual.1073.3 PredicatesCross-linguistically, verbal and nominal predicates have been reported tolack the D shell (see Déchaine 1993; Lyon 2013; Hedberg and Potter 2010;den Dikken, 2006; Witschko 2009; Matthewson 1998 and many others).In many languages, nominal predicates are found with the predicationalcopula where the post-copula nominal denotes a property predicated ofthe subject (Higgins 1973; Lyon 2013; Hedberg and Potter, 2010; denDikken 2006; Mikkelsen 2005; and others). A classical example comesfrom German, where nominal predicates (NPs) which express a propertyappear without a D, (144a) but argument NPs which enter into identityrelation (DP=DP) occur with a D, (144b):(144) Predicate/Argument [German, Wiltschko 2009:26]a. MariaMariáistis[Lehrerin] [NP=predicate][teacher]‘Mary is [a teacher].’b. MariaMariasahsaw[die Lehrerin] [DP=argument][D teacher]‘Mary saw [the teacher].’Another well known case of predicates is adjectival predicates whichdenote a property or quality (Levinson 1978; Partee 1986; Déchaine 1993;Villalba 2009; and others). Examples come from English copula construc-tions which also require a tensed form of the copula be. In (145) the post-copula predicate which is an adjectival predicate lacks a D:(145) Adjectival predicates in Englisha. John is busy.b. Sarah is honest.While in some languages like English nominal predicates may be DPs, asin Sarah is a teacher, and rarely are NPs as as in Lucy is boss (see Zamparelli1081995, Longobardi 1994; Lyon 2013), I show that Nata nominal predicatesconsistently lack a D.3.3.1 Nata nominal predicates are φ-NLongobardi’s generalization that DPs can be arguments will make a weakclaim for Nata in which nominal predicates consistently lack a D. Thus,Nata needs a stronger version of this, something like (146).(146) Generalization for Nata nominal predicatesNata nominal predicates lack a D.I will show that nominal predicates in Nata have the φ-N structure,which means they always agree with the subject of which the property ispredicated (see Chapter 1 for discussion of concordial agreement). I arguethat in Nata, φ qua class prefix diagnoses predicative expressions. I discusssimple nominal predicates, and secondary predicates, where I show all ofthese lack a D-layer.3.3.1.1 Simple nominal predicates lack a DSimple φ-N predicates are formed of number (φ) and the nominal stem(N) and must occur with the overt copula6. Nominal predicates denote aproperty where they lack a D, as the cases in (a) below show. Note thatthe predicate must agree in number with the subject. The caveat fromthe English translation in cases such as (147a) or (148a) is that nominalpredicates in English may appear as DPs (see Higgins 1973; Déchaine 1993;Mikkelsen 2005; Zamparelli 1995 for arguments about DP predicates inEnglish), even if there is no D in the actual Bantu examples.(147) Question: What gender is Bhahati?6. The copula is a homorganic nasal, as it assimilates to the place of articulation of thefollowing consonant; see Johannes (2007) and Anghelescu (2019).109a. BhahatiβahatiBhahatim=mo-subhem=mo-súβeCOP=C1-man‘Bhahati is a man.’b. #Bhahati#βahatiBhahatino=o=mo-subhene=o=mo-súβeCOP=D=C1-man‘Bhahati is the man.’(148) [Context: A friend is telling you about Masato’s behaviour:]a. MasátoMasátoMasatom=mw-iibh-i [Nata]m=mw-iiβ-íCOP=C1-steal-FV‘Masato is a thief.’b. #Masáto#MasátoMasatono=o=mw-iibh-ino=o=mw-iiβ-íCOP=D=C1-steal-FV‘Masato is the thief.’Note that a property can be predicated of a plural subject and predicatesmust agree in number with the subject DP.(149) [Context: What is the gender of Masato and Bhahati?]a. MasatoMasatoMasatonanaandBhahatiβahatiBhahatim=ba-subhe [Nata]m=ba-súβeCOP=C2-man‘Masato and Bhahati are men.’110b. #Masato#MasatoMasatonanaandBhahatiβahatiBhahatin=a=ba-subhen=a=ba-súβeCOP=D=C2-man‘Masato and Bhahati are (the) men.’(150) [Context: B commenting on the behaviour of two men]a. MasatoMasatoMasatonanaandBhahatiβahatiBhahatim=bi-ibh-i [Nata]m=ba-iβ-íCOP=C2-steal-FV‘Masato and Bhahati are thieves.’b. #Masato#MasatoMasatonanaandBhahatiβahatiBhahatin=a=bhi-ibh-in=aa=βa-iβ-íCOP=D=C2-steal-FVIntended: ‘Masato and Bhahati are the thieves.’Since nominal predicates are not entity-denoting DPs, they are not ex-pected to be used as arguments. This is confirmed by (151) which showsthat a φ-N nominal is not licit in argument position.(151) a. *mo-subhe*mo-súβeC1-mana-a-hik-irea-a-hik-ireSA1-PAST-arrive-PFV[Nata]Intended: ‘A/the man has arrived.’b. *bha-subhe*βa-suβeC2-menbha-a-hik-ireβa-a-hík-ireSA2-PAST-arrive-PFVIntended: ‘(The) men have arrived.’1113.3.1.2 D-linked wh-phrases as complex nominal predicatesThe ex-situ wh-questions are formed of a nominal predicate introduced by acopula nasal, followed by a wh-word, forming a cleft structure, as in (152a);or they may be formed of a predicate nominal and wh-word, without a cop-ula nasal, as in (152b). The host NP never takes a D as (152c) shows. Thenon-use of the augment is predicted for predicate nominals if the augmentis a D.(152) a. m=mw-aana=kem=mu-ana=keCOP=C1-child=WHa-ku-rayaar-aa-ku-rajáar-a3s-PROG-run-FVha-yọ?ha-jɔ?C16-there‘[It] is which child running there?’b. mw-aana=kemu-ana=keC1-child=WHa-ku-rayaar-aa-ku-rajáar-a3s-PROG-run-FVha-yọ?ha-jɔ?C16-there‘Which child is running (out) there?’c. *u=mw-aana=ke*u=mu-aná=keD=C1-child=WHa-ku-rayaar-aa-ku-rajáar-a3s-PROG-run-FVha-yọ?ha-jɔ?C16-thereIntended lit: ‘[It] is which child running there?’‘Which child is running (out) there?’I treat the homorganic copula nasal in (152a) as a FOC marker (cf. Rizzi(1997); Allen 2014; Gambarage and Keupdjio 2013), and the φ-N=ke in(152a)-(152b) as a predicate nominal. I treat the invariable wh-element=ke which encliticizes on a nominal predicate as a complementizer (C),introducing the φ-N nominal predicate. As has been widely observed inBantu expletive constructions, Bantu lacks overt expletive pronouns (cf.Simango 2006; Riedel 2009)112(153) FOCP(FOC)NXPφPmw-aana‘child’CPC=ke‘which’IPakurayaara...‘s/he is running...’Since D-linked wh-questions ask about an entity/individual establishedin the discourse context i.e., which individual out of a larger salient set(Den Dikken and Giannakidou 2002; Hirose 2003), the lack of D in theex-situ D-linked wh-phrases can only be explained by syntactic factors.One piece of evidence showing that the cases in (152) involve predica-tion comes from Nata in-situ type questions. When the wh-element and thecopula nasal remain in-situ therefore separated from the nominal predicate,the NP may now take a D to form a DP.(154) a. *mw-aana*mu-anáC1-childu-yọu-yɔC1-RELa-ku-yaar-aa-ku-jáar-a3s-PROG-run-FVha-yọha-jɔC16-therene-we?ne-we?COP-C1.whIntended: ‘The child who is running (out) there is which one?’b. u=mw-aanau=mu-anáD=C1-childu-yọu-jɔC1-RELa-ku-yaar-aa-ku-jáar-a3s-PROG-run-FVha-yọha-jɔC16-therene-we?ne-we?COP-C1.whLit: ‘The child who is running (out) there is which one?’The syntactic status of D in D-linked questions varies across languages(see Pesetsky 1987; Den Dikken and Giannakidou 2002; Hirose 2003). In113a language like Zulu, D-linked questions may or may not co-occur with theD depending on the syntactic position of the wh element (Buell 2011; Pou-los and Msimang 1998; Halpert 2012). For instance, Buell (2011) reportsthat when an agreeing element -phi ‘which’ follows the noun, the D cannotoccur, (155a); however, when the right-dislocated noun occurs with it, theD shows up on the noun, (155b):(155) D-linked wh-words [Zulu, Halpert 2012: 133]a. w-a-bona2s-PST-seemu-ntu1-personmu-phi?1-which‘Which person did you see?’b. w-a-bona2s-PST-seemu-phi1-whichu=mu-ntu?AUG-1-person‘Which person did you see?’I leave open the question whether in the in-situ type, the wh elementoccupies the D position like English D-linked wh-word. Further research isneeded to understand the locus of variation for D-linked questions withinBantu and beyond.3.3.2 Nata adnominal predicates are φ-AI follow the Bantuist tradition of grouping together nouns and adnomi-nal modifiers (i.e. adjectives) as “nominal expressions” (see Wilkins andKimenyi 1975; Givón 1970 and others). The language-internal evidencefor this comes from the concordial agreement: adnominal modifiers are in-flected for number/noun-class and must agree in number with a head noun.That is, adnominal modifiers appear with a φ-A structure (see Chapter 1 fordiscussion).3.3.2.1 Post-copula adjectives lack a DOne category of adnominal predicates is post-copula adjectives which areφ-A. Post-copula adjectives appear with no D and pattern the same as other114clearly predicative nominals as we saw above (cf. Déchaine 1993, 2001;Lyon 2013).(156) a. o=mu-kario=mu-káriD=C1-womanm=mu-rẹm=mu-rɛCOP=C1-tall[Nata]‘A/the woman is tall.’b. e=bhe-bhusee=βe-βúseD=C8-monkeym=be-nyiinim=be-ɲíiniCOP=C8-clever‘Monkeys are clever.’3.3.2.2 Adjectival modifiers lack a DThe second category of adnominal predicates are adjectival modifierswhich also have φ-A structure and they immediately follow the nominalthat they modify, resulting in a surface N-A order, (157).(157) a. o=mu-kario=mu-káriD=C1-womanmu-rẹ [Nata]mu-rɛC1-tall‘a tall woman.’b. e=bhe-bhusee=βe-βúseD=C8-monkeybe-nyiinibe-ɲíiniC8-clever‘clever monkeys.’As predicted under the current theory that all nominal predicates denote aproperty, φ-A adnominal predicates are ruled out in argument positions7.7. To convey the equivalent of degree modification as in very tall, Nata deploys stress on apredicate, which as expected lacks a D (i):115(158) a. *mu-rẹ*mu-rɛC1-talla-a-hik-irea-a-hik-ireSA1-PAST-arrive-PFV[Nata]Intended: ‘A/the tall one has arrived.’b. *bhe-nyiini*βe-ɲíiniC8-cleverbhe-e-hik-ireβe-e-hik-ireSA8-PAST-arrive-PFVIntended: ‘The clever (ones) have arrived.’In §3.4 I show that when the augment is used with φ-A predicates an argu-ment DP is formed where it denotes an entity, and as such the DP can beused in argument positions. Here, I conclude that all φ-Ns and φ-A struc-tures are predicates therefore they lack a D.3.3.3 Nata secondary predicates are φ-XAnother context to illustrate predication is with secondary predicates foundin small clauses. Secondary predicates are expressions that convey infor-mation about the subject or the object but are not the main predicate ofthe clause (Déchaine 1993; Huddleston and Pullum 2002; Ullrich 2018; Ir-imia 2005; Stowell 1981; Schneider-Zioga and Mutaka 2014). The currentanalysis predicts that predicates in these structures will lack a D. The com-mon types of secondary predicates I discuss here are: (i) complement smallclauses (SC) and (ii) adjunct predicates. In a language like English, com-plement SCs appear with no tense marker (159); with an infinitive copulalike to be or the relator as, (160a). These compare to a full CP structure (atensed clause) in (160b):(i) o=mo-teo=mo-téD=C3-treeMU-REmú-rɛvery.C3-tall‘a very tall tree.’116(159) a. I consider [Joe intelligent]. [=descriptive]b. I hammered [the metal flat]. [=resultative]c. The news made [Lucy sad]. [=causative]d. I saw [Lucy leave]. [=bare infinitive](160) a. I consider [Joe (as/to be) intelligent].b. I believe [that Joe is intelligent].On the other hand, adjunct predicates are not complements and maydepict the subject, (161a) or the object, (161b) (see see Déchaine 1993;Irimia 2006; Pylkkänen 2002).(161) a. Lucy ran the race hungry. [subject depictive]b. Lucy prefers her meat well-cooked. [Object depictive]There are various treatments of secondary predicates that I cannot ex-haustively discuss here (see Déchaine 1993; Irimia 2006; Pylkkänen 2002).One popular syntactic analysis for secondary predicates is the one thattreats depictives as adjunct phrases merged at the level of VP if they refer todirect objects, and at the level of vP if they modify the external argument.In contrast, resultatives are treated as complements merged to the V layer(see Irimia 2006; Pylkkänen 2002). I show that Nata secondary predicateslack a D, as we would expect if they were nominal predicates.3.3.3.1 Nata secondary nominal predicates lack a DI argue that secondary predicates in all the small clause structures availablein Nata consistently lack a augment, as one would expect if the augment isD. The evidence comes from complement clauses with either the infinitivecopula as in (162a), or with the relator -anga ‘as’ or ‘like’, (163a), which in-troduce a nominal predicate. Note the verb rọr- here is multiply ambiguousbetween see, consider and find.117(162) What is your consideration about Wasato? [Nata]a. N=ni-haa-rọr-aN=ni-haa-rɔŕ-aFOC=1sg-HAB-see/find-FV[Wasato[Wasato[Wasatom=mu-ghabho]m=mú-ɣaβo]COP=C1-healer]‘I find [Wasato to be a healer].’b. #N=ni-haa-rọr-a#N=ni-haa-rɔŕ-aFOC=1sg-HAB-see/find-FV[Wasato[Wasato[Wasatono=o=mu-ghabho]ne=o=mu-ɣaβo]COP=D=C1-healer]Intended: ‘I find [Wasato to be a healer].’(163) What do you think about Wasato’s manner?a. N=ni-haa-rọr-aN=ni-haa-rɔŕ-aFOC=1sg-HAB-consider-FV[Wasato[Wasato[Wasaton=aangan=aŋgáSAM=asmw-aana]mu-aná]C1-child]‘I see [Wasato as a child].’b. #N=ni-haa-rọr-a#N=ni-haa-rɔŕ-aFOC=1sg-HAB-consider-FV[Wasato[Wasato[Wasaton=aangan=aŋgáSAM=asu=mw-aana]u=mu-aná]D=C1-child]Intended: ‘I see [Wasato as a child].’I confirm that the augment cannot be used in any nominal predicates.This conclusion is also reached with adjectival predicates.3.3.3.2 Nata secondary adjectival predicates lack a DSecondary adjectival predicates also appear with copulas and indicate theproperty is predicated of the subject. As expected, such predicates have a118φ-A structure and do not carry a D. Note that the copula is crucial for thesecases as well:(164) What do you think about Wasato’s look? [Nata]a. N=ni-haa-bhugh-aN=ni-haa-βúɣ-aFOC=1sg-HAB-consider-FV[Wasato[Wasato[Wasatom=mu-chọmu]m=mu-tʃɔḿu]COP=C1-good]‘I consider Wasato beautiful.’b. #N=ni-haa-bhugh-a#N=ni-haa-βúɣ-aFOC=1sg-HAB-consider-FV[Wasato[Wasato[Wasatono=o=mu-chọmu]ne=o=mu-tʃɔḿu]COP=D=C1-good]Intended: ‘I consider Wasato beautiful.’Small clauses formed of predicate adjuncts are also found in Nata andshow that predicate adjuncts lack a D as predicted by the current analysis.The depictive predicates, which are adjectives, may depict a property overthe subject, (165), direct object (166a)8. In any case, the secondary pred-icate cannot appear with a D. Note that the predicates here are adjectivalin nature.(165) Secondary predicates [Subject depictive]a. MariaMariaMariaa-ka-yar-aa-ka-jáar-aSA1-PST-ran-FVmo-rwiire [Nata]mo-rwíireC1-sick‘Maria ran sick.’8. Pylkkänen (2002) argues that in English a depictive cannot modify an indirect argument(ia) or a DP inside PP (ib).(i) a. *I gave Mary the meat hungry.b. *I talked to Sue drunk.Irimia (2006) shows with data from Romanian, Slavic and Albanian that these results arenot cross-linguistically valid. In Nata, modification of an indirect object or of a DP insidePP is also not possible with regular φ-A, but is possible with typical adverbs.119b. *Maria*MariaMariaa-ka-yar-aa-ka-jáar-aSA1-PST-ran-FVo=mo-rwiireo=mo-rwíireD=C1=sickIntended: ‘Maria ran sick.’(166) Secondary predicates [Object depictive]a. MasatoMasatoMasatoa-ka-ria-ka-riSA1-PST-eat-FV[a=n.yama[a=ɲ.ama[D=C9.meatm-bese]m-bése]C9-raw]‘Masato ate [the meat raw].’b. #Masato#MasatoMasatoa-ka-ria-ka-riSA1-[PST-eat-FV[a=n.yama[a=ɲ.amaD=C9.beefa=m-bese]a=m-bése]D=C9-raw]Intended: ‘Masato ate [the meat raw].’The lack of augment in depictive predicates support the argument thatthe augment is D and that predicates cannot take a D.3.3.4 Nata adverbials lack a DThere are two main classes of adverbials in Nata. One class is spatial ad-verbials which behave like the English adverbs down or up, which cannottake an augment/D.(167) Spatial adverbs as non-argumental adverbialsa. ghi-ka-gh-iɣi-ka-ɣ-iC7-PST-go-FV[ighoro/haa-se][iɣóro/haa-sé][up/C16-down]‘(It) went up/down.’120b. *ghi-ka-gh-i*ɣi-ka-ɣ-iC7-PST-go-FV[i=ighoro/a=haa-see][i=iɣóro/a=haa-sé][D=up/D=C16-down]Intended: ‘(It) went up/down.’The other class is temporal adverbials, which are treated as secondary(depictive) predicates (see Déchaine 1993). My analysis correctly pre-dicts that adverbs may require a φ, and adverbial modification will blockthe augment/D. This is correct. Nata speakers reject sentences with theaugment on a temporal adjunct equivalent to “all day/night” in English,(168b/169b). These cases also sound odd to me.(168) Temporal adjuncts as non-arguments [Nata]a. a-gha-tẹm-aa-ɣa-tɛm-a3sg-PST-hit-FVa-m-beerea-m-béerePPF-C9-drum[bho-tiko[βo-tíko[C14-nightbho-ghima]βó-ɣima]C14-whole]‘S/he played a/the drum all night.’b. *a-gha-tẹm-a*a-ɣa-tɛm-a3sg-PST-hit-FVa-m-beerea-m-béerePPF-C9-drum[o=bho-tiko[o=βo-tíko[o=C14-nightbho-ghima]βó-ɣima]C14-whole]Intended: ‘S/he played a/the drum all night.’(169) Temporal adjuncts as non-arguments [Nata]a. a-ka-rer-er-aa-ka-rér-er-a3sg-PST-cry-APPL-FVe-ki-ghẹsọe-kí-ɣɛsɔPPF-C7-knife[mw-ise[mw-isé[C3-daymo-ghima]mó-ɣima]C3-whole]‘S/he cried for a/the knife all day.’b. *a-ka-rer-er-a*a-ka-rér-er-a3sg-PST-cry-APPL-FVe-ki-ghẹsọe-kí-ɣɛsɔPPF-C7-knife[o=mwise[o=mw-ise[o=C3-daymo-ghima]mó-ɣima]C3-whole]‘S/he cried for a/the knife all day.’121I analyze the temporal adjuncts as non-argument adjuncts which adjointo the VP as shown below. This explains why they lack a D.(170) VPVPVtema‘hit’DPa=mbeere‘drum’YPbhotiko bhoghima‘all night’3.3.5 Nata infinitives lack a D-layerIn Bantu, infinitives are formed with the C15 prefix ku-, which Bantu gram-marians have described as having both the properties of a verb and a noun(a.k.a verbo-nominal stems) (see Schadeberg 2006: 80). When ku-stemsare used without a D they must be verbal predicates, (171a)/(172a):(171) a. Ni-kwẹẹnd-aNe-kwɛnd-á1Sg.SM-want-FVkw-eemb-akw-eemb-áC15-sing-FVkẹkɛlikeMasatoMasatoMasato‘I want to sing like Masato.’b. *Ni-kwend-a*Ni-kwend-á1Sg.SM-want-FVu=kw-eemb-au=kw-eemb-áD=C15-sing-FVkẹkɛlikeMasatoMasatoMasatoIntended: ‘I want to sing like Masato.’122(172) a. MasatoMasatoMasatoa-ke-end-aa-kɛ-ɛnd-aSA1-PST-want-FVko-bhor-iko-βor-iINF-ask-FV‘Masato wanted to ask.’b. *Masato*MasatoMasatoa-ke-end-aa-kɛ-ɛnd-aSA1-PST-want-FV[o=ko-bhor-i][o=ko-βor-i][D=INF-ask-FV]Intended: ‘Masato wanted to ask.’Stems attaching to INF or inflected with C15 and taking a D must functionas gerunds rendering an entity denoting reading, as I show in §3.4.The data with predicates above demonstrated that nominal predicatesin Nata lack the augment which is consistent with the analysis of it as D-layer. We saw that predicates do not denote individuals but a property; assuch they cannot be used in an argument position. In what follows, I turnto D=φ-N structures where I show that D is required with all argumentnominals.3.4 Argument nominals are D-φ-NIn the previous section I showed that Nata nominal predicates Nata lack aD. Here, I claim that an augment is required to close off the NP predicateand create a DP, which is consistent with its analysis as D (Stowell 1989;Longobardi 1994; Alexiadou et al. 2007). One piece of evidence for ana-lyzing the augment as a D is that it can turn a predicate into an argument(cf. Longobardi 1994; 2001; Déchaine 1993). The following examples mayprovide evidence for this argument9.9. Partee (1986) argues that property-denoting nominals can be type-shifted to entity-denoting DPs of type e, and type e can raise to a property of type . Similar claimsare found in Longobardi (1994), Heim and Kratzer (2010), but see Chierchia (1998) for adifferent approach, in which he argues that in some languages, plural/mass bare NPs arearguments on their own.123(173) [Context: I know that the police caught a thief but I didn’t knowthe thief’s name is Masato. The next day a friend asks me if I feelsorry for Masato. I ask: Who is Masato? He answers:]a. #Masáto#MasátoMasatom=mw-iibh-i [Nata]m=mw-iiβ-íCOP=C1-steal-FV‘Masato is a thief.’b. MasátoMasátoMasatono=o=mw-iibh-ino=o=mw-iiβ-íCOP=D=C1-steal-FV‘Masato is the thief.’[Context: I know that the police caught two male thieves but I didn’t knowthe thieves’ names are Masato and Bhahati. The next day a friend asks meif I feel sorry for Masato and Bhahati. I ask: Who are Masato and Bhahati?He answers:](174) a. #Masato#MasatoMasatonanaandBhahatiβahatiBhahatim=bi-ibh-i [Nata]m=ba-iβ-íCOP=C2-steal-FV‘Masato and Bhahati are thieves.’b. MasatoMasatoMasatonanaandBhahatiβahatiBhahatin=a==bhi-ibh-in=aa=βa-iβ-íCOP=D=C2-steal-FV‘Masato and Bhahati are the thieves.’The cases in (173)-(174) point to equative and specificational copulas inwhich DP1 denotes the same individual as DP2 (cf. Higgins 1973; Mikkelsen2005; Lyon 2013; Sneider-Zioga and Mutaka 2014). It is worth pointing outthat the definite construal of o=mwiibhi ‘the thief’ in (173) or a=bhiibhi ‘thethieves’ in (174) is not an inherent property of the D; rather the definiteness124effect comes from the context (refer to Chapter 2 for further discussion). Iclaim that the generalization in (175) is robust in all argument positions inNata:(175) The generalization about argument nominalsAll Nata argument nominals must have a DI present a range of Nata data to show that the D is required in all argu-ment positions, consistent with Longobardi’s claim that argument nominalsmust have a D. I focus on the D requirement with simplex Ns in argumentposition (§3.4.1) and with complex Ns in argument position: N-N com-pounds and D spreading contexts, (§3.4.2).3.4.1 D is required in all argument positionsWith the exception of proper names, which I discuss in Chapter 6, the Din Nata is obligatory in all argument positions. I consider in turn the sub-ject (§3.4.1.1), direct object (§3.4.1.2), indirect object (§3.4.1.3), possessor(§3.4.1.4), and prepositional object positions (§3.4.1.5). For each of thesecontexts, I provide examples of both singular and plural DPs and for allnoun types: count, abstract and mass nouns. This is because in many lan-guages, plural DPs behave differently than singular DPs relative to the real-ization of overt D (see Vergnaud and Zubizarreta 1992; Krifka 1999, 2003;Borer 2005; Chierchia 1998; Longobardi 2001; Déchaine et al. 2018).3.4.1.1 D is required in subject positionThe D is obligatory in subject position. The (b) cases show that argumentnominals are consistently ruled out if they do not appear with a D. I startwith count nouns.125(176) Count N Singular, Subject Positiona. e=ghi-kọmbẹe=ɣi-kɔɔ́mbɛD=C7-cupghi-ka-hiringit-aɣi-ka-híriiŋgit-aSA7-PST-roll-FVha-aseha-asédown‘A/the cup rolled down.’b. *ghi-kọmbẹ*ɣi-kɔɔ́mbɛC7-cupghi-ka-hiringit-aɣi-ka-híriiŋgit-aSA1-PST-roll-FVha-aseha-asédownIntended: ‘A/the cup rolled down.’(177) Count N Plural, Subject Positiona. e=bhi-kọmbẹe=βi-kɔɔ́mbɛD=C8-cupbhi-ka-hiringit-aβi-ka-híriiŋgit-aSA8-PST-roll-FVha-aseha-asédown‘(The) cups rolled down.’b. *bhi-kọmbẹ*βi-kɔɔ́mbɛC8-cupbhi-ka-hiringit-aβi-ka-híriiŋgit-aSA8-PST-roll-FVha-aseha-asédownIntended: ‘(The) cups rolled down.’Abstract nouns are argument DPs as they obligatorily occur with the D.Most abstract nouns take class 14 βo which marks abstract entities. Theclass is neutral with respect to the singular-plural contrast10.(178) Abstract N, Subject Position:10. Recall from Chapter 1 that there are exceptions with noun classifications, hence ab-stract nouns may feature with other classes as well: u=rw-ambe (C14), ‘gossip’, a=ma-reghe (C6) ‘commotion’ etc. I will not investigate here what controls the choice of the classmarker/number morphology on these cases.126a. o=bho-bhiihio=βo-βíihiD=C14-liesbhu-gha-suut-aβu-ɣasúut-aSA14-PST-cost-FVMariaMariaMaria‘(The) lies cost Maria.’b. *bho-bhiihi*βo-βíihiC14-liesbhu-gha-suut-aβu-ɣasúut-aSA14-PST-cost-FVMariaMariaMariaIntended: ‘(The) lies cost Maria.’As with count and abstract nouns, a D is obligatory with mass nouns inargument position. Certain mass nouns denoting fluids that clot or entitiesdescribed as ‘some amount of X’ or ‘a count of X’, (??) as opposed to ‘lotsof X or X’, (??) can take singular morphology (refer also to Chapter 2):(179) Mass N, Subject Position:a. ri=i-sahẹrí=í-saahɛD=C5-bloodri-ko-om-ari-ka-om-áSA5-PST-dry-FV‘The/some amount of blood dried up.’b. *i-sahẹ*í-saahɛC5-bloodri-ko-om-ari-ka-om-aSA5-PST-dry-FVIntended: ‘The/some amount of blood dried up.’(180) Mass N, Subject Position:a. a=ma-saahẹa=ma-saahɛD=C6-bloodgha-ghi-itek-aɣa-ɣa-iték-aSA6-PST-spill-FV‘(The) blood spilled.’127b. *ma-saahẹ*ma-saahɛC6-bloodgha-ghi-itek-aɣa-ɣa-iték-aSA6-PST-spill-FVIntended: ‘(The) blood spilled.’Mass nouns that show no singular plural count take plural morphologyalways. A noun such as a=manche ‘water’ in (181) as opposed to fluidsthat clot as in (179) above, takes plural morphology, but still takes the D.This shows that the D is obligatory in all these cases:(181) Mass N, Subject Position:a. a=ma-nchea=ma-ntʃéD=C6-watergha-ghi-itek-aɣa-ɣa-iték-aSA6-PST-spill-FV‘(The) water spilled.’b. *ma-nche*ma-ntʃéC6-watergha-ghi-itek-aɣa-ɣa-iték-aSA6-PST-spill-FVIntended: ‘(The) blood spilled.’Ds are required on all subject arguments, whether they are in the mainclause (182) or in the subordinate clause (183):(182) Subjects in main clausesa. [u=-mw-aana[u=-mu-ana[D=C1-childa-ka-bhugh-a]a-ka-βuɣ-a]SA1-PST-say-FV]angoaŋgothatMariaMariaMarya-gha-sẹk-aa-ɣa-sɛk-aSA1-PST-laugh-FV‘[A/the child said] that Mary laughed.’128b. *[mw-aana*[mu-ana[C1-childa-ka-bhugh-a]a-ka-βuɣ-a]SA1-PST-say-FV]angoaŋgothatMariaMariaMarya-gha-sẹk-aa-ɣa-sɛk-aSA1-PST-laugh-FVIntended: ‘[A/the child said] that Mary laughed.’(183) Subjects in subordinate clausesa. MariaMariaMarya-ka-bhugh-aa-ka-βuɣ-aSA1-PST-say-FV[ango[aŋgo[thatu=-mw-aanau=-mu-anáD=C1-childa-gha-sẹk-a]a-ɣa-sɛk-a]SA1-PST-laugh-FV]‘Mary said [that a/the child laughed].’b. *Maria*MariaMarya-ka-bhugh-aa-ka-βuɣ-aSA1-PST-say-FV[ango[aŋgo[thatmw-aanamu-anáC1-childa-gha-sẹk-a]a-ɣa-sɛk-a]SA1-PST-laugh-FV]Intended: ‘Mary said [that a/the child laughed].’3.4.1.2 D is required in direct object positionsWhen an argument is complement to V (i.e., is a direct object) the D is stillobligatory, as the examples in (184) and (185) show:(184) Count N, Direct Object Singulara. MariaMariaMarya-kẹ-ẹghia-ka-ɛɣ-íSA1-PST-wash-FVe=ghi-kọmbẹe=ɣi-kɔɔ́mbɛD=C7-cup‘Mary washed a/the cup.’129b. *Maria*MariaMarya-kẹ-ẹghia-ka-ɛɣ-íSA1-PST-wash-FVghi-kọmbẹɣi-kɔɔ́mbɛC7-cupIntended: ‘Mary washed a/the cup.’(185) Count N, Direct Object Plurala. MariaMariaMarya-kẹ-ẹghia-ka-ɛɣ-íSA1-PST-wash-FVe=bhi-kọmbẹ [Count, pl.]e=βi-kɔɔ́mbɛD=C8-cup‘Mary washed (the) cups.’b. *Maria*MariaMarya-kẹ-ẹghia-ka-ɛɣ-íSA1-PST-wash-FVbhi-kọmbẹβi-kɔɔ́mbɛC8-cupIntended: ‘Mary washed (the) cups.’Abstract nouns behave the same as count nouns in requiring a D in asthe examples below show:(186) Abstract N, Object positiona. MakuruMakuruMakurua-ka-ghamb-aa-ka-ghamb-aSA1-PST-speak-FVo=bho-bhiihio=βo-βíihiD=C14-lies‘Makuru told lies.’b. *Makuru*MakuruMakurua-ka-ghamb-aa-ka-ghamb-aSA1-PST-speak-FVbho-bhiihiβo-βíihiC14-lies‘Makuru told lies.’There is no exception for mass nouns not to take a D in any syntacticposition. Recall that mass nouns may exhibit a number contrast taking130either singular morphology, (187), or plural morphology, (188); whereasmass nouns that show no such contrast take plural morphology always,(189). As the (b) cases show, the D is obligatory in all these cases:(187) Mass N, Object positiona. a-gha-sangor-ia-ɣa-sáŋgor-i3sg-PST-clean.up-FVri=i-saahẹrí=í-saahɛD=C5-blood‘S/he cleaned up the/some amount of blood.’b. *a-gha-sangor-i*a-ɣa-sáŋgor-i3sg-PST-clean.up-FVi-saahẹí-saahɛC5-bloodIntended: ‘S/he cleaned up the/some amount of blood.’(188) Mass N, Object positiona. a-ka-ruus-ia-ka-rúus-i3sg-PST-draw-CAUSa=ma-saahẹa=má-saahɛD=C6-blood‘S/he drew (the) blood.’b. *a-ka-ru-si*a-ka-rúu-si3sg-PST-draw-CAUSma-saahẹmá-saahɛC6-bloodIntended: ‘S/he drew (the) blood.’(189) Mass N, Object positiona. a-ka-rẹẹt-aa-ka-rɛɛt-a3sg-PST-bring-CAUSa=ma-nchea=má-ntʃéD=C6-water‘S/he brought (the) water.’131b. *a-ka-rẹẹt-a*a-ka-rɛɛt-a3sg-PST-bring-FVma-nchemá-ntʃéC6-waterIntended: ‘S/he brought (the) water.’3.4.1.3 D is required in indirect object positionNot only are Ds required when the argument is a direct object of the mainpredicate, but also when an argument is an indirect object, as in (191):(190) Count N, Indirect Object Singulara. MariaMariaMariaa-ka-haa-ka-h-áSA1-PST-give-FVu=-mw-aanau=mu-anáD=C1-childe=ghi-kọmbẹe=ɣi-kɔɔ́mbɛD=C7-cup‘Maria gave a/the child a/the cup.’b. *Maria*MariaMariaa-ka-haa-ka-h-áSA1-PST-give-FVmw-aanamu-anáC1-childe=ghi-kọmbẹe=ɣi-kɔɔ́mbɛD=C7-cupIntended: ‘Maria gave [a/the child] a/the cup.’(191) Count N, Indirect Object Plurala. MariaMariaMariaa-ka-haa-ka-h-áSA1-PST-give-FVa=bha-anau=βa-anáD=C2-childe=ghi-kọmbẹe=ɣi-kɔɔ́mbɛD=C7-cup‘Maria gave (the) child a/the cup.’b. *Maria*MariaMariaa-ka-haa-ka-h-áSA1-PST-give-FVbha-aanaβa-anáC2-childe=ghi-kọmbẹe=ɣi-kɔɔ́mbɛD=C7-cupIntended: ‘Maria gave a/the children a/the cup.’132The D is required with abstract nouns, (192) and mass nouns, (193) inan indirect object position as well. Note that the verb here carries applica-tive inflection to add an inanimate indirect object argument. The orderof arguments is rigid due to what is believed to be a Focus marking effectwhich I do not discuss here (see Sadlier-Brown 2013 for a discussion ofword order between DO and IO in Nata).(192) Abstract N, Object positiona. A-ka-bhọh-ẹr-aA-ka-βɔh́-ɛr-aSA1-PST-sue-APPL-FVMakuruMakuruMakuruo=bho-bhiihio=βo-βíihiD=C14-lies‘S/he sued Makuru because of lies.’b. *A-ka-bhọh-ẹr-a*A-ka-βɔh́-ɛr-aSA1-PST-sue-APPL-FVMakuruMakuruMakurubho-bhiihiβo-βíihiC14-liesIntended: ‘S/he sued Makuru because of lies.’(193) Mass N, Object positiona. a-gha-tah-er-aa-ɣa-táh-er-aSA1-PST-fetch-APPL-FVa=ma-nchea=má-ntʃéD=C6-watere=ghi-kọmbẹe=ɣi-kɔɔ́mbɛD=C7-cup‘S/he fetched (the) water with a/the cup.’b. *a-gha-tah-er-a*a-ɣa-táh-er-aSA1-PST-fetch-APPL-FVa=ma-nchea=má-ntʃéD=C6-waterghi-kọmbẹɣi-kɔɔ́mbɛC7-cupIntended: ‘S/he fetched (the) water with a/the cup.’Finally, I turn to gerundive nouns which I show require D, unlike in-finitive predicates which lack one all together.1333.4.1.4 D is required with gerundsThe infinitives or ku-stems in Nata cannot be used in argument positionsas the (b) cases show. When the augment/D is present, the infinitive verb(the predicate) is turned into a DP/gerund and it is used as an argument11:(194) a. u=kw-eemb-au=kw-eemb-áD=C15-sing-FVkokoof15MasatoMasátoMasaton=gu-chomuŋ=gu-tʃómuCOP=C15-good‘Masato’s singing is good.’b. *kw-eemb-a*kw-eemb-aC15-sing-FVkokoof15MasatoMasatoMasaton=gu-chomuŋ=gu-tʃómuCOP=C15-goodIntended: ‘Masato’s singing is good.’(195) a. o=ko-mer-ao=ko-mer-aD=C15-swallow-FVghu-ka-mor-er-iɣu-ka-mó-rer-iSA-PST-OM-cry-CAUS‘(The) swallowing made him/her cry.’b. *ko-mer-a*ko-mer-aC15-swallow-FVghu-ka-mor-er-iɣu-ka-mó-rer-iSA-PST-OM-cry-CAUSIntended: ‘(The) swallowing made him/her cry.’11. For Bantu languages with no augments/Ds such as Swahili, the stem with the infinitiveku- can be used in the context of a verbal predicate (infinitive) as in (1a), or as a gerund,(1b) (cf Carstens 1993: 178):(i) a. Ni-na-tak-a1sg-PRES-want-FVku-imbaC15-singkamalikeJuma [Swahili]Juma‘I want to sing like Juma.’b. ku-imbaC15-singkwa15ofJumaJumaku-me-ni-furah-ish-aSA15-PFV-1sg-be.happy-CAUS-FV‘Juma’s singing made me happy.’1343.4.1.5 D is required with temporal argumentsWe saw in §3.3.2 that temporal adjuncts equivalent to “all day/night” inEnglish (196a) cannot take a D as they modify a VP.(196) Temporal adjuncts as non-arguments [Nata]a. a-ka-rer-er-aa-ka-rér-er-a3sg-PST-cry-APPL-FVe-ki-ghẹsọe-kí-ɣɛsɔPPF-C7-knife[mw-ise[mw-isé[C3-daymo-ghima]mó-ɣima]C3-full]‘S/he cried for a/the knife all day.’b. *a-ka-rer-er-a*a-ka-rér-er-a3sg-PST-cry-APPL-FVe-ki-ghẹsọe-kí-ɣɛsɔPPF-C7-knife[o=mwise[o=mw-ise[o=C3-daymo-ghima]mó-ɣima]C3-whole]‘S/he cried for a/the knife all day.’My analysis predicts that when a temporal expression is an argument, itmust be a DP, and so will be licit with the augment/D. This property obtainswith Nata adverbials which parallel adverbials in many languages: for in-stance, temporal adjunct cases found in St’át’imcets, on which Matthewson(1998) remarks,If temporal nouns such as day have an intrinsic temporal com-ponent, the temporal adjuncts may already be licensed (or ‘sit-uated’) without the need for a determiner to perform this func-tion... When a determiner is needed for syntactic reason to cre-ate an argument, even temporal phrases will require a D, (p.81).This seems to be borne out in the following Nata example:(197) Question: How much time do you need to assemble my table?a. Nu-u-h-ẹNe-u-h-ɛ1sg-2sg-give-MOD[o=mwise[o=mwisé[o=C3-daymo-ghima]mó-ɣima]C3-whole]‘You give me a/the whole day.’135b. *Nu-u-h-e*Ne-u-h-ɛ1sg-2sg-give-MOD[mw-ise[mw-isé[C3-daymo-ghima]mó-ɣima]C3-whole]Intended: ‘You give me a/the whole day.’For the cases like (197) I propose the tree in (198), in which φ projectsa φP which merges with the D to create an argument DP.(198) VPVNuuhe‘You give me’DPDo=φPφPmw-ise‘day’φPφmoA-ghima‘whole’3.4.2 D is required with complex nounsThis section shows that compounds and nouns in partitive structures (thosewhich are a result of D spreading) must have an augment to function asarguments, which supports an analysis of the augment as D.3.4.2.1 D is required with N-N compoundsIt is widely known that DPs which are formed by the compounding of twonouns allow one determiner (cf. Ndayiragije et al. 2012; Stegen 2002;Bauer 2003; and others). The English examples in (199) show that inde-pendent DPs have each a D; however, if they form a compound, the twonouns result in one DP which will retain only the leftmost D, as in (200a):136(199) a. A babyb. A sitter(200) a. A baby-sitterb. *baby a sitterc. *A baby a sitterThe table below indicates that both stem (N-prefix plus √N) and root(√N) compounds exist in Nata. Crucially, when an N-N compound isformed, only one D is realized, namely the D of the head (which is theleft-hand member of the N-N compound):(201) Stem-stem compoundD=C7-√N D=C9-√N D=C7-√N–C9-√Na. e=ke-mer-a a=ɲ-tʃɔka e=ke-mer-a–ɲ.tʃɔkaD=C7-swallower D=C9-snake D=C7-swallower-FV–C9.snake‘a swallower’ ‘a snake’ ‘a/the snake-swallower’.b. ″ ″ *ke-mer-a–a=ɲ.tʃɔkaC7-swallower-FV–D=C9.snakeIntended: ‘A/the snake-swallower’.c. ″ ″ *e=ke-mer-a–a=ɲ.tʃɔkae=C7-swallower-FV–D=C9.snakeIntended: ‘A/the snake-swallower’.137(202) Root-root reduplicationD=C1-√N D=C1-√N D=C1-√N–√Na. o=mo-súβe o=mo-súβe o=mo-súβe-suβéD=C1-man D=C1-man D=C1-man-RED‘a man’ ‘a man’ ‘a real man.’b. ″ ″ *mo-súβe-o=mo-suβéC1-man-D=C1-manIntended: ‘a real man.’c. ″ ″ *o=mo-súβe-o=mo-suβéD=C1-man-D=C1-man’Intended: ‘a/the real man.’As is evident from these examples, all nominal compounds in Nata allowone D12.3.4.2.2 D is required with determiner spreadIn some languages, when DPs take certain modifiers (i.e., adjectives, quan-tifiers, etc.), a determiner or a clitic-like particle occurs on the modifiers.This phenomenon is commonly known as D-spreading/doubling (see Buell2011; Giusti 1994 on Hebrew; Delsing 1993 on Scandinavian dialects; Mor-wood 2001; and numerous authors on Classical (CG) and Modern Greek(MG), Alexiadou et al. 2007, Alexiadou 2014)13.(203) D Spreading with adjectivesa. heDETgunewomenheDETsofe [CG, Morwood 2001: 125]wise‘The wise woman.’12. There are different types of compounding discussed in the literature; here I only presentcrucial data for the current proposal.13. Note also that languages like Albanian, LIthunian, and Romanian have D doubling thatinvolves Ds appearing as suffixes or clitic-like-material (see Alexiadou et al. 2007).138b. tothevivliobooktotheoreogood[MG, Alexiadou et al. 2007: 73]‘The good book.’c. enastorbigenakarman[Northern Swedish, Delsing 1993: 143]‘A big man.’This agreement phenomenon does not receive a uniform treatmentacross languages. In some languages D spreading has a semantic contri-bution (e.g., in Greek it is definiteness spreading, and in Northern Swedishdialect it is indefiniteness spreading (Delsing 1993). In other languages likeModern Hebrew, the contribution is debatable; some scholars say D dou-bling correlates with the construct state (Borer 1988; Ritter 1991; Siloni2002; Kremers 2009; Alexiadou et al. 2007; Alexiadou 2014; and others);while others say it is purely a syntactic phenomenon (see Giusti 2002)14.In Nata, augments display an agreement phenomenon similar to D-doubling/spreading. First, a modified nominal expression in argument po-sition permits a D on a head noun but does not require D-doubling on themodifier, (204). These DPs have an “indefinite construal” e.g., are used indiscourse-new contexts.(204) o=mu-kario=mu-káriD=C1-womanmo-nyiinimo-ɲíiniC1-wisea-ka-het-aa-ka-hét-aSA1-PST-pass-FVhanọ [Nata]hánɔhere‘A wise woman passed here.’In D-doubling structures both the nominal expression and the modifierappear with the overt D, (205)-(206). I argue that D-doubling in Nata is notpleonastic. The function of the extra D on adjectives and weak quantifiers14. Androutsopoulou (1996) argues that the double determiners found with DefiniteSpreading structures are “expletive” Ds at least to some Greek speakers. Giusti argues the Dspreading in Hebrew serves a purely syntactic role, particularly that of encoding agreementbetween the DP and its satellites by bearing φ and case features.139is to create DPs that can pick out a subset from the set referred to by thefirst DP, hence rendering a presuppositional/partitive reading (cf. Alexi-adou and Wilder 1998, Kolliakou 2004; Alexiadou et al, 2007; Alexiadou2014)15.(205) D-spreading: Adjectivesa. o=mu-kario=mu-káriD=C1-womano=mo-nyiinio=mo-ɲíiniD=C1-wisea-ka-het-aa-ka-hét-aSA1-PST-pass-FVha-nọ [Nata]há-nɔ16-hereLit: ‘The woman, the wise (one) passed here’‘The wise woman passed here.’b. e=ghi-tabhoe=ɣi-taβoD=C7-booke=ghi-chomue=ɣi-tʃómuD=C7-goodghi-ka-gwaɣi-ka-ɣw-aC7-PST-fall-FVLit: ‘The book, the good (one) fell’‘The good book fell.’(206) D-spreading: weak quantifiersa. a=bha-anau=βa-anáD=C2-childa=bha-ndea=βáa-ndeD=C2-some/otherbha-ka-het-aβa-ka-hét-aSA2-PST-pass-FVha-nọhá-nɔ16-here‘The other kids passed here.’b. a=bha-anaa=βa-anáD=C2-childa=bha-rua=βaa-rúD=C2-manybha-ka-het-aβa-ka-hét-aSA2-PST-pass-FVhanọhánɔhere‘Many of the kids passed here.’15. There is evidence in the literature that D-spreading never applies to demonstratives.This is true in Nata, as we saw in Chapter 1.140Here the presuppositional/partitive reading is that there is one individ-ual or multiple individuals out of a group of individuals being talked about.I analyze the D-doubling structures as appositive structures with D=φ-Nand D=φ-A arguments. I treat the DP internal modification by none D-doubling structures as corresponding to the structure in (207a), and theD-doubling cases as corresponding to the structure in (207b), which is theappositive structure.(207) a. DP2Da=φPφPbha-ana‘kids’φPbhande‘certain’b. DPDP1a=bhaana‘kids’DP2a=bhaande‘the other’D-doubling structures in Nata correlate to the construct state in He-brew (Ritter 1991; Siloni 2002), and to Greek, in which clitic-doubling hasa familiarity effect (see Anagnostopoulou 1994; Alexiadou et al. 20007;Alexiadou 2014). However, as I argued in Chapter 2, the presupposition-al/familiarity reading of the D-doubling structures is not inherently fromDs, but from the contextually salient set. In the next section I will focus onthe distribution of covert Ds.3.5 Polarity-sensitive argument DPs are D∅-φ-NThe Nata overt D in (208a) contrasts with the covert D∅ in (208b).(208) a. N-ka-rọr-aN-ka-rɔŕ-a1sg-PST-see-FVu=mw-aanau=mu-anáD=C1-child‘I saw a/the child.’141b. N-ty-a-a-rooch-eN-tj-a-a-rootʃ-e1sg-NEG-SA1-PST-see-PFVmw-aanam-anáiC1-child‘I didn’t see any child.’The contrasting Nata Ds are presented as in (219), where (219a) is anovert D and (219b) is a covert D:(209) a. DPDu=φPφ-mw-N-aanab. DPD∅φPφ-mw-N-aanaThe phonologically null D is restricted in its distribution, appearing onlyin polarity contexts where it is c-commanded by a non-factual operator; forthis reason I refer to it as a polarity-sensitive D∅, a label I use interchange-ably with the null D∅ (cf. Progovac 1993; Haspelmath 1997; Matthewson1998; Lahiri 1998; Farkas 2002; Giannakidou 2001, 2011; von Fintel 1999;and others). Consistent with the literature on polarity-sensitive elements,I claim that the distribution of the polarity-sensitive D∅ is captured by thefollowing generalization:(210) Polarity licensing condition for null D:The polarity D∅ must fall under the c-command domain/scope of anon-factual operator. (The set of non-factual operators in Nata in-cludes Negation, Modality, Question operators, and Conditionals.)The terms ‘C-command’ and ‘Operator’ need to be defined. For c-command,I adopt the definition of asymmetrical c-command given in (211) (Reinhart1421976; Chomsky 1981). This type of c-command is consistent with the factthat the operator always c-commands the polarity D∅16.(211) α c-commands β iff:a. The first binary-branching node that dominates α also dominates β.b. Neither α nor β dominate each other.c. α c-commands β but β does not c-command α.The term ‘Operator’ is defined as an element which binds a variable(see Heim 1982; Chierchia and McConnell-Ginet 1990; Longobardi 1994;Cinque 1990; and others). For instance Heim (1982) gives the followingexamples of Operators:We take operators to include quantifiers, negation, temporaland modal operators (which are in some sense quantifiers, i.e.,quantifiers over times and possible worlds).[Heim 1982:143]Heim’s examples of operators include both factual and non-factual op-erators. I distinguish a set of operators that licenses polarity Ds from thosewhich do not. I argue that DP operators such as quantifiers, (212a), nullDPs in Spec, CP, (212b) or wh-phrases, (212c) do not license the polarityD∅. This is the case in Nata, (213)-(214) (see also Matthewson 1998 forsimilar observations in Salish). Since I discuss Nata wh-phrases in §3.5.2 Iwill not give any examples here to avoid repetition.(212) DP operatorsa. Every child found a key.b. [The man [Opi that found my keys ti]] was Socrates.c. Who found my keys?16. Evidence for this comes from lexical negation and the negative light verb which cannotlicense a subject DP containing the D∅, as we shall see below.143(213) a. u=mw=aanau=mw=aanáD=C1-childw-ọọsẹw-ɔɔsɛ ́C1-everya-a-tọọr-irea-a-tɔɔr-ireSA1-PST-find-FVo=ro-hungurọo=ro-huŋgurɔ́D=C11-key‘Every child found a key.’b. *u=mw=aana*u=mw=aanáD=C1-childw-ọọsẹw-ɔɔsɛ ́C1-everya-a-tọọr-irea-a-tɔɔr-ireSA1-PST-find-FVro-hungurọro-huŋgurɔ́C11-keyIntended: ‘Every child found a key.’(214) a. o=mo-too=móo-toD=C1-personu-nọu-nɔC1-RELa-a-tọọr-irea-a-tɔɔr-ireSA1-PST-find-FVo=ro-hungurọo=ro-huŋgurɔ́D=C11-keyrw-anẹrw-anɛC11-mym-Makurun-MakuruFOC-Makuru‘A person who found my key was Makuru.’b. *o=mo-to*o=móo-toD=C1-personu-nọu-nɔC1-RELa-a-tọọr-irea-a-tɔɔr-ireSA1-PST-find-FVro-hungurọro-huŋgurɔ́C11-keyrw-anẹrw-anɛC11-mym=Makurun=MakuruFOC-MakuruIntended: ‘A person who found my key was Makuru.’Crucially, all operators sit in A’-position; the DP cases in (212a)/(212b)in English, and the corresponding examples in Nata, have already raisedto an A’-position in LF, while (212c) has done so in the overt syntax (seeLongobardi 1994; Cinque 1990; Matthewson 1998 and others).I will largely focus on ‘non-factual’ operators, operators which licensethe polarity D∅. I show that in some cases the non-factual operators may144take the entire proposition under their scope/c-command. I use an Englishnegation example for illustration:(215) a. John didn’t see any woman.b. It is not the case [that there is a woman that John saw].The DP containing the D∅ must always fall under the scope of a non-factual operator. In Chapter 4 I argue that in this environment Nata DPs donot allow an existential interpretation (cf. Hoeksema 2012; Matthewson1998, 1999; Progovac 2000; Giannakidou 2000, 2011; von Fintel 2009;and others). In (215b), the sentence does not assert existence of a womanthat John saw. This contrast will be discussed extensively in Chapter 4.Crucially, the Nata polarity-sensitive D∅ appears in a variety of ‘polarityenvironments’, consistent with the typology of environments allowing po-larity sensitive elements. These environments include super-strong, strong,weak, and super-weak (see (see Zwarts 1998; Progovac 1998, 2000; Gian-nakidou 1998, 2000, 2006; Hoeksema 2012; Krifka 1995; von der Wouden1997; Gajewski 2011; Lin 2015; and others). Building on polarity classifi-cations, the appropriate term that fits the broad range of environments inwhich the Nata D∅ occurs is non-veridical (cf. Giannakidou 2006; Gajewski2011; Hoeksema 2012; Ladusaw 1980)17. Giannakidou and Mari (2018)define veridicality as followings:(216) a. F is veridical iff Fp entails p i.e., when Fp is true p is also true.b. F is non-veridical iff Fp does not entail p i.e., when Fp is true pmay or may not be true.One other possibility we may discuss regarding licensing of the polarityD is with downward entailing (DE) operators (cf. Ladusaw 1980; Progovac1993; Halpert 2012; Giannakidou 2011). While the Nata polarity D may17. There are many other ways of testing levels of negativity, e.g., anti-additive and anti-morphic, which I do not discuss here. For a discussion of these cases see Hoeksema 2012and others.145be licensed in a variety of downward entailing environments such as nega-tion, conditional, etc., I do not assume that the Nata polarity D can onlybe licensed in DE environments. Typical cases of non-DE operators are Q-morphemes in polar questions, and models which may license the polarityD in Nata but are not DE operators (see Progovac 1993; 2000). Anotherreason is that not all DE operators may license the polarity D in Nata. Forinstance, a quantifier like -suhu ‘few’ is a DE operator but does not inher-ently license the polarity element; I showed in Chapter 2 that all QPs inNata take a range/a DP that contains an overt D (see further discussion inChapter 4). In what follows I present a variety of licensing environmentsto show that polarity Ds consistently appear under the scope/c-commanddomain of a non-factual operator: negation (§3.5.1), question operators(§3.5.2), conditionals (§3.5.3) and modals (§3.5.4).3.5.1 NegationNegation as a non-factual operator licenses the polarity D in Nata. Unlikemany languages that display a strong tendency for polarity elements tofollow their licensor, Nata presents a different parameter for licensing ofpolarity Ds on subject DPs, i.e., the subject DP in (217a) precedes NEG inthe overt syntax18. In Nata, overt c-command is not a licensing requirementas the D∅ in any non-existential construction may precede the licensor as(217a) illustrates. In negative existential constructions, the DPs containinga polarity D∅ may precede the licensor, (217a) or follow it (217b), withno difference in meaning in the two structures.(217) Negative existential constructions in Nataa. mu-karimu-káriC2-womanta-a-nyihọta-a-ɲíhɔNEG-3SG-there-isa-a-rooch-ea-a-rootʃ-é3SG-PST-like-PFVYohanajohanáJohnLit: ‘There is no woman (who) saw John’¬[9x [woman (x) & x saw John]].18. Needless to say, objects DPs follow negation.146b. ta-a-nyihọta-a-ɲíhɔNEG-3SG-there-ismu-karimu-káriC2-womana-a-rooch-ea-a-rootʃ-é3SG-PST-see-PFVYohanajohanáJohnLit: ‘There is no woman (who) saw John’¬[9x [woman (x) & x saw John]].Based on these data I claim that it is the underlying (syntactic) representa-tion, i.e., input structure/the structure before spell-out, that fixes the scoperelation in Nata (see Sportiche et al. 2013 for a related discussion). Theclaim that scope relations are determined by the input structure coincideswith the well-known Bantuist claim that subject DPs are base generated atSpec, vP and are realized at a higher position, Spec, XP after movement(see Koopman and Sportiche 1991; Ngonyani:1998; Carstens 2005, 2001;Halpert 2012; and many others)19. I claim that licensing happens in thecovert syntax at which point the Op c-commands the subject before raisingto Spec, XP.(218) Licensing of subject polarity DsXPD∅-φPsubj NegPNeg vP3D∅-φPsubjLicensed herevPv0 VPV DPobj19. I do not discuss the structural position of various functional elements such as tense andsubject agreement (INFL) in Nata (see Ngonyani 1999 and references therein for discussionin Bantu.)147The landing site of the subject DP is a matter of controversy in Bantu,and I will not discuss this topic here (see Ngonyani 1998, 1999, 2001;Kinyololo 1991; Diercks 2010 for discussion). In Chapter 6, I discuss thec-command parameter in other Bantu languages and show that Nata isconsistent with Giannakidou’s (2001, 2011) observation that not all polar-ity elements must be c-commanded by their licensers in the overt syntax;language-specific conditions may apply.I show that different kinds of negation license the D∅ in Nata. Thecases I discuss here include sentential negation, adverbial negation, ad-versative/negative predicates, and intonational negation. I start with thesentential one.3.5.1.1 Sentential negation licenses D∅Sentential negation licenses the polarity D∅ in all argument positions.Some accounts may predict that licensing of polarity D may be restricted toargument position or with some verb asymmetry. For instance, there maybe some restriction with either unergative verbs for which the subject ofthe verb is the agent argument or with unaccusative intransitive verbs forwhich the subject of the verb is a patient/theme argument (see Perlmutter1978; Burzio 1986; Zeller and Ngoboka 2013)20.20. In Nata, unergative and unaccusative verbs seem to pattern differently when they takea passive morpheme. Unergative verbs allow passivization, (ia), while unaccusative verbsdo not, (ib). (Adding an applicative morpheme here will neutralize the contrast.).(i) a. o=mu-gharukao=mu-ɣárukaD=C1-eldera-gha-sẹk-u [unergative verb]a-ɣa-sɛk-uSA1-PST-laugh-PASS‘An/the elder was laughed at.’b. *o=mu-gharuka*o=mu-ɣárukaD=C1-eldera-ka-hik-u [unaccusative verb]a-ka-hik-uSA1-PST-arrive-PASSIntended: An/the elder was caused to arrive.’I take this as language internal evidence that there is an unergative/unaccusative contrastin Nata. Other tests available in the literature are inapplicable in Nata. For instance, inEnglish unergative subjects pattern differently from unaccusative subjects when used withthere–constructions: *There laughed two men vs There arrived two men. See also Zeller and148(219) a. vPAGENT vPv Vlaughb. vPv VPVarriveNTHEMEI start by showing there is no verb asymmetry or number restriction inthe licensing of D∅. I argue that since NEG occurs above vP a polarity sen-sitive D∅ in the subject position of an unergative intransitive verb, (220);or with an unaccusative intransitive verb, (221) will be licensed:(220) a. mu-gharukamu-ɣárukaC1-elderta-a-sẹk-ire [unergative verb]ta-a-sɛk-iréNEG-PST-laugh-PFV‘No elder laughed.’b. bha-gharukaβa-ɣárukaC2-elderbha-ta-a-sẹk-ireβa-ta-a-sɛk-iréSA2-NEG-PST-laugh-PFV‘No elders laughed.’(221) a. muu-kimuu-kíC1-ladyta-a-hik-ire [unaccusative verb]ta-a-hik-iréNEG-PST-arrive-PFV‘No lady arrived.’b. bhaa-kiβaa-kíC1-ladybha-ta-a-hik-ireβa-ta-a-hik-iréSA2-NEG-PST-arrive-PFV‘No ladies arrived.’Ngoboka 2014; Bresnan and Kanerva 1989; Demuth 1997 who use a locative inversion testwhich also does not work for Nata.149Additionally, sentential negation can license D∅ in in-situ objects; thisholds of both singular (222a) and plural (222b) DPs:(222) a. N-tj-a-ghor-ireN-te-a-ɣór-iré1sg-NEG-PST-buy-PFVki-ghẹsọkí-ɣɛsɔC7-knife‘I didn’t buy any knife.’b. N-tj-a-ghor-ireN-te-a-ɣór-iré1sg-NEG-PST-buy-PFVbhi-ghẹsọβí-ɣɛśɔC8-knife‘I didn’t buy any knives.’Sentential negation can also license the D∅ with a transitive verb onsubject (223a), object (223b) or both subject and object, (223c)21.(223) a. mw-aanamu-anáC1-childt-a-a-rooch-et-a-a-rootʃ-éNEG-SA1-PST-see-FVMakuruMakuruMakuru‘No child saw Makuru.’b. MakuruMakuruMakurut-a-a-rooch-et-a-a-rootʃ-éNEG-SA1-PST-see-FVmw-aanamu-anáC1-child‘Makuru didn’t see any child.’c. mo-subhemo-suβéC1-mant-a-a-rooch-et-a-a-rootʃ-éNEG-SA1-PST-see-FVma-yaanima-jaaníC1-gazelle‘No man saw any gazelle.’21. Since I have already shown that number does not matter in licensing, from now on Iwill present singular-plural pairs only when necessary.150Similarly, for the ditransitive verb ha ‘give’ in (224), NEG can licensenull D with any one of the arguments of the verb: with the subject DPmu-kari ‘woman’, the Goal/indirect object mw-aana ‘child’, and the the-me/direct object ma-βɛɛ́rɛ ‘milk’:(224) a. mu-karimu-káriC1-womant-a-a-h-eret-a-a-h-eréNEG-SA1-PST-give-PFVmw-aanamu-anáC1-childma-bhẹẹrẹma-βɛɛ́rɛC6-milk‘No woman gave any child any milk.’b. ¬[9xyz [woman (x) & milk (z) & child (y) & x gave z to y]].The informal semantics in (224) corresponds to the underlying (syn-tactic) representation in (225) where NEG scopes over the entire proposi-tion, thereby c-commanding all the polarity Ds in the sentence (cf. Uribe-Etxebarria 1994; Bruening 2014; Matthewson 1998).(225) Licensing of polarity Ds in ditransitive verbsNegPNeg vP3D∅-φPsubj vP3D∅-φPINDO vPv0 VPV 3D∅-φPDO151Extensional suffixes such as applicatives (APPLs) take a specifier-argument, hence add an extra argument (see Rugemalira 1993; Pylkkänen2002; Halpert 2012)22. An example of a sentence with a transitive applica-tivized verb in (226) shows that NEG can scope over the subject (226a),any one of the objects (226b-d), all the objects (226e), or the subject andall the objects (226f):(226) a. mo-subhemo-súβeC1-mant-a-a-kw-eegh-er-it-a-a-kw-eeɣ-éer-iNEG-SA1-PST-teach-APPL-FVo=mu-ghenio=mú-ɣeniD=C1-visitoru-nọú-nɔC1-DEMu=mw-anau=mw-anáD=C1-childu-yọu-jɔC1-DEMe=ki-ghambọe=kí-ɣambɔD=C7-languageki-nọki-nɔC7-DEM‘No man is teaching that child that language for this visitor.’b. o=mo-subheo=mo-súβeo=C1-manu-nọú-nɔC1-DEMt-a-a-kw-eegh-er-it-a-a-kw-eeɣ-éer-iNEG-SA1-PST-teach-APPL-FVmu-ghenimú-ɣeniC1-visitoru=mw-aanau=mu-anáD=C1-childu-yọú-jɔC1-DEMe=ki-ghambọe=kí-ɣambɔD=C7-languageki-nọki-nɔC7-DEM‘This man is not teaching that child this language for any visitor.’c. o=mo-subheo=mo-súβeD=C1-manu-nọú-nɔC1-DEMt-a-a-kw-eegh-er-it-a-a-kw-eeɣ-éer-iNEG-SA1-PST-teach-APPL-FVo=mu-ghenio=mú-ɣeniD=C1-visitoru-nọú-nɔC1-DEMmw-aanamu-anáC1-childe=ki-ghambọe=kí-ɣambɔD=C7-languageki-nọki-nɔC7-DEM‘This man is not teaching any child this language for this visitor.’22. The applicative morpheme has the highest valency (i.e., number of arguments), for thatreason I do not give examples of causativized verbs which take fewer arguments.152d. o=mo-subheo=mo-súβeD=C1-manu-nọú-nɔC1-DEMt-a-a-kw-eegh-er-it-a-a-kw-eeɣ-éer-iNEG-SA1-PST-teach-APPL-FVo=mu-ghenio=mú-ɣeniD=C1-visitoru-nọú-nɔC1-DEMu=mw-aanao=mu-anáD=C1-childu-yọu-jɔC1-DEMki-ghambọkí-ɣambɔC7-language‘This man is not teaching that child any language for this visitor.’e. o=mo-subheo=mo-súβeD=C1-manu-nọú-nɔC1-DEMt-a-a-kw-eegh-er-it-a-a-kw-eeɣ-éer-iNEG-SA1-PST-teach-APPL-FVmu-ghenimú-ɣeniC1-visitormw-aanamu-anáC2-childki-ghambọkí-ɣambɔC7-language‘This man is not teaching any children any language for any visitor.’f. mo-subhemo-súβeC1-mant-a-a-kw-eegh-er-it-a-a-kw-eeɣ-éer-iNEG-SA1-PST-teach-APPL-FVmu-ghenimú-ɣeniC1-visitormw-aanamu-anáC2-childki-ghambọkí-ɣambɔC7-language‘No man is teaching any kids any language for any visitor.’Given the evidence that polarity-sensitive D∅ is licensed in any argu-ment position in an applicative sentence, it must be the case that sententialNEG occurs above both Spec, vP and Spec, ApplP in covert syntax:153(227) Licensing of polarity Ds in ditransitive applicative verbsNegPNeg ...vP3D∅-φPsubj vPv0 ApplP3D∅-φPO1 VP3D∅-φPO2 VPV 3D∅-φPDONEG, in a matrix clause, can license the polarity D∅ in the subjectand/or the object of an embedded clause, (228a). It can also take the entireroot clause licensing any argument in its c-command in the covert syntax,(228b)(228) a. MakuruMakuruMakurut-a-a-bhugh-iret-a-a-bhugh-ireNEG-3sg-PST-say-PFV[ango[aŋgo[thatmw-aanamw-aanaC1-childa-ki-ibh-aa-ka-iβ-a3sg-PST-steal-FVma-kuwa]ma-kuwa]C6-sugarcane]‘Makuru didn’t say that any child stole any sugarcane.’154b. mu-ghabhomu-ɣaβoC1-healert-a-a-bhugh-iret-a-a-βuɣ-ireNEG-3sg-PST-say-PFV[ango[aŋgo[thatmw-aanamw-aanaC1-childa-ki-ibh-aa-kaiβ-a3sg-PST-steal-FVma-kuwa]ma-kuwa]C6-sugarcane]‘No healer said that any child stole any sugarcane.’I now consider adversative predicates, which confirm the generalizationthat licensing occurs when NEG appears above the DP containing D∅ incovert syntax.3.5.1.2 Adversative predicates license D∅Adversative predicates, which are verbs that are inherently negative (a.k.alexical negations), can license D∅ in object position but not in subject posi-tion (cf. Progovac 1993, 2000; Giannakidou 1998, 2001). This is consistentwith Yoon’s (2013) observation that lexical negations cross-linguisticallydo not truth-conditionally negate a proposition. For example, adversativepredicates such as ang- ‘refuse’ can license D∅ in object position, (229a),but not in subjects of either an intransitive, (229b), or transitive sentence,(229c).(229) a. A-ka-ang-aA-ka-ang-aSA1-PST-refuse-FVku-ghẹgh-aku-ɣɛɣ́-aINFT-carry-FVmw-aanamu-anáC1-child‘S/he refused to carry any child.’b. * mw-aana* mu-anáC1-childa-ka-ang-aa-ka-aŋg-aSA1-PST-refuse-FVghu-sẹk-aɣu-sɛḱ-aINFT-laugh-FV*‘Any child refused to laugh.’155c. * mu-kári* mu-káriC1-womana-ka-ang-aa-ka-aŋg-aSA1-PST-refuse-FVku-ghẹgh-aku-ɣɛɣ́-aINFT-carry-FVmw-aanamu-anáC1-child*Any woman refused to carry a child.’The data in (229) contrast with cases with overt Ds in the subject po-sition of adversative predicates, (230). These are correctly predicted to beacceptable as overt Ds do not need to be c-commanded by a licenser.(230) a. u=mw-aanau=mu-anáD=C1-childa-ka-ang-aa-ka-aŋg-aSA1-PST-refuse-FVghu-sẹk-aɣu-sɛḱ-aINFT-laugh-FV‘A/the child refused to laugh.’b. o=mu-kário=mu-káriD=C1-womana-ka-ang-aa-ka-aŋg-aSA1-PST-refuse-FVku-ghẹgh-aku-ɣɛɣ́-aINFT-carry-FVmw-aanamu-anáC1-child‘A/the woman refused to carry any child.’The restriction on licensing the polarity D∅ in subject argument DPswith the adversative predicates in (229b,c) can be represented as in (231b).This contrasts with the non-lexical licensing in (231a) in which the lexicalnegation is lower in the tree than NEG.156(231) Syntactic constraints on licensing of D∅a. Sentential negation can license D∅ in subjectsNegPNeg vP3D∅-φPsubj v0verb VPV 3D∅-φPobjb. Lexical NEG cannot license D∅ in subjects*vP7D∅-φPsubj v0neg.verb VPV 3D∅-φPobjLexical negation in Nata can be used as a diagnostic for the positionin which Nata subject DPs with D∅ are licensed. It shows that not onlydo subjects have to be under Spec, vP to be licensed, also the licenser hasto c-command the polarity element, hence the structure in (231b) is notpossible.3.5.1.3 Covert negation/reduced pitch licenses D∅It appears that Nata lacks lexical items corresponding to the English nega-tive predicates ‘doubt’ or ‘deny’ in (232).157(232) a. I doubt that Mary ate anything. [Progovac 1998: 08]b. I deny that Mary ate anything.c. *Mary ate anything.When Nata speakers are casting doubt on or disagreeing about the ex-istence of an entity, they can use a special kind of reduced pitch, renderingsimilar meanings to ‘I doubt’ or ‘I deny’. This may license the polarity D∅on either subject, (233a), object, (233b), or both, (233c). I mark this pitcheffect with the symbol ▽ as in (233):(233) a. ▽ mu-kari▽ mu-káriC1-womana-ka-bhaator-aa-ka-βáátor-a3sg-PST-spank-FVMakuruMakuruMakuru‘(I deny that) any woman spanked Makuru.’b. ▽Makuru▽MakuruMakurua-ka-bhaator-aa-ka-βáátor-a3sg-PST-spank-FVmw-aanamw-anáC1-child‘(I deny that) Makuru spanked any child.’c. ▽ mu-kari▽ mu-káriC1-womana-ka-bhaator-aa-ka-βáátor-a3sg-PST-spank-FVmwa-anamwá-anaC1-child‘(I doubt that) any woman spanked any child.’A voice recording of a Nata male consultant captured differences in theintonation patterns between the regular sentence (Figure 3.1) and the sen-tence containing the covert negation (Figure 3.2). In the regular sentence(Figure 3.1), the blue curve (pitch) in the waveform has higher frequencythan in the sentence marked with covert negation (Figure 3.2), which ap-pears with compressed pitch, thus low frequency:158Figure 3.1: Pitch pattern for regular sentencesFigure 3.2: Pitch pattern for sentences with covert negationTo account for the licensing of D∅ in the cases like (233), I proposethat the reduced pitch is realized as a covert negation (cf. Uribe-Etxebarria1994; Hoeksema 2012; Swart 2009; and others) which licenses the D∅. Thecovert negation unselectively licenses the D∅ in subject, object or both asseen in (233). Since the overt D requires no licensing it is expected that it159will also be used with the covert negation. This prediction is correct as theexample below shows:(234) ▽o=mu-gheni▽o=mú-ɣeniD=C1-guesta-ka-bhaator-aa-ka-βáátor-a3sg-PST-spank-FVu=mwa-anau=mwá-anau=C1-child‘(I doubt that) a/the visitor hit a/the child.’The semantic contrast between the overt and the covert D∅ will bedealt with in Chapter 4. The last case under negation is about the negativeadverbs keeré ‘yet’ and naabha ‘never’, which I now turn to.3.5.1.4 Negative light verbs license D∅Polarity Ds can be licensed by a negative light verb keeré ‘be before’ ‘bewithout’ or ‘be yet’ (cf. Giannakidou 2002; Watanabe 2004; Halpert 2012;and others), (235). This negative light verb cannot license the subject, asthe case in (235b) shows. This restriction also points to a common factabout licensing, namely that the subject DPs containing a D∅ are licensedat Spec, vP, hence they cannot be licensed by the negative light verb theyprecede:(235) a. u=mw-aanau=mw-aanaD=C1-childa-kerea-keré3sg-be.yetku-nyw-aku-ɲw-aINFT-drink-FVke-rongoorike-róŋgooriC7-porridgeLit: ‘A/the child is yet to eat any porridge.’b. * mw-aana* mw-aanaC1-childa-kerea-keré3sg-be.yetku-nywaku-ɲwaINFT-drinkke-rongoorike-róŋgooriC7-porridgeIntended: ‘Any child is yet to eat any porridge.’Since the licensor heads the vP, the subject DP containing the D∅ gen-erated in its Spec cannot be licensed:160(236) The negative light verb cannot license D∅ on subjects*vP7D∅-φPsubj vPv0keereVPV 3D∅-φPobjIf we use a covert negation or other kinds of licensors that can occurwith the minimizer, the subject will be licensed by such a licensor:(237) ▽ mw-aana▽ mw-aanaC1-childa-kerea-keré3sg-yetku-nywaku-ɲwaINFT-drinkke-rongoorike-róŋgooriC7-porridge‘(I doubt that) any child is yet to eat any porridge.’The structure in (238) corresponds to the licensing of the subject by ahigher operator. This shows that licensing of D∅ in subject DPs strictlyobserves the syntactic condition in (210) that the licensor is above Spec,vP, where it scopes over/c-commands the subject in the covert syntax.(238) An Op above vP can license D∅XPOp/▽ ...vP3D∅-φPsubj vPv0keereVPV 3D∅-φPobj161I close this subsection by showing that the speaker can switch the D∅to the overt D in the cases discussed above:(239) u=mw-aanau=mw-aanaD=C1-childa-kerea-keré3sg-yetku-nywaku-ɲwaINFT-drinke=ke-rongoorie=ke-róŋgooriD=C7-porridge‘A/the child is yet to eat (the) porridge.’I will argue in chapter 4 that this switch corresponds to a semantic differ-ence to do with the notion of existence.3.5.2 QuestionsI first discuss polar/yes-no questions which license the polarity D∅ and thenturn to content questions where I show that DP operators do not license thepolarity D∅. For a discussion of D-linked wh-questions refer to predicateclefts under §3.5.4.3.5.2.1 Polar questions license D∅In polar/yes-no questions, D∅ can be licensed under the scope of the ques-tion operator. This is shown in (240):(240) [Context: Speaker is not sure if there was any child]a. angoangoQmw-aanamw-áanaC1-childa-ka-rọr-aa-ka-rɔŕ-a3s-PST-see-FVMaria? [Subject]María?Mary‘Did any child see Mary?’b. angoangoQMariaMaríaMarya-ka-bhọn-aa-ka-βɔń-a3s-PST-find-FVmw-aana? [Object]mw-áana?C1-child‘Did Mary find any child?’162As Gambarage and Keupdjio (2013) demonstrate, polar questions inNata do not have an overt Q(uestion) morpheme. The Q-morpheme inNata derives from a rising intonation (↑) (cf. Cheng 1991; Rooryck 1994;Cheng and Rooryck 2000). Thus, the Q-operator is sitting in C where it c-commands the DP containing the D∅ (see Progovac 2000; Gambarage andKeupdjio 2013):(241) Licensing of D∅ in polar questionsCPC(Qn↑) ...X vPDPsubj v0v VPV DPobjIn this analysis, the Q-operator takes the clause as its complement,hence c-commanding the DP containing the D∅ in the covert syntax. Iclaim that this rising intonation is a realization of C which licenses D∅.As expected, overt Ds are not restricted in their distribution whichmeans they can be used with the polar question. The question about whatforces different D choices in Nata will be answered in Chapter 4 which talksabout the semantics of Nata D.(242) [Context: Speaker is talking about his friend’s child]a. angoangoQu=mw-aanau=mw-áanaD=C1-childa-ka-rọr-aa-ka-rɔŕ-a3s-PST-see-FVMaria? [Subject]María?Mary‘Did a/the child see Mary?’163b. angoangoQMariaMaríaMarya-ka-bhọn-aa-ka-βɔń-a3s-PST-find-FVu=mw-aana? [Object]u=mw-áana?D=C1-child‘Did Mary find a/the child?’In Chapter 4, I will adopt the common analysis that the DP containingthe overt D such as in (242) scopes above the non-factual operator (cf.Matthewson 1998,1999; Giannakidou 2002; Gambarage 2012; and manyothers).3.5.2.2 Wh-questions do not license D∅Wh-phrases, like some other DP operators like quantifiers discussed above,cannot license the polarity D in any object DP argument, hence the overtD must be used, (243). In subordinate clauses, however, some other non-factual operator, e.g, the evidential verb bhugha (discussed below) can li-cense the D∅, (244):(243) a. Ne=weNe=waCOP=WHa-a-rooch-ea-a-róotʃ-é3sg-PST-see-FVe=ke-bhuse?e=ke-βúse?D=C7-monkey‘Who saw a/the monkey?’b. *Ne=we*Ne=weCOP=WHa-a-rooch-ea-a-rootʃ-é3sg-PST-see-FVke-bhuse?ke-βúse?C7-monkeyIntended: ‘Who saw a/the monkey?’[Consultant comment: If you do not have anything in mind,why would you ask?][Context: There are no monkeys in Vancouver. B heard someone claim-ing to have seen one; he thinks it was some other animal. B says withincredulous voice:]164(244) Ne=weNe=weCOP=WHabhughirea-βuɣ-ireSA-say-PFVa-ka-rọr-aa-ka-rɔr-a3sg-PST-see-FVke-bhuse?ke-βúse?C7-monkey‘Who said s/he saw a monkey? (I don’t believe there was any).’In Chapter 4 I argue that this restriction has to do with the presuppo-sitional property of the wh-phrases (cf. Erteshik-shir 1993; Matthewson1998; Gambarage 2013). For the discussion of Nata D-linked questionswhich are concealed predicate clefts in which case they lack a D, see §3.6.I next turn to conditionals where I show that, like negative sentences andpolar questions, conditionals also license D∅.3.5.3 ConditionalsPolarity Ds are licensed in non-factual if-clauses/conditional sentences.[Context: A mother has a sick child and only elderly people know thetraditional cure of the disease. She would be happy if any elder comesbut that’s impossible, because there are no elders in this community]23.[Adapted from Matthewson, 1999: 90]23. When I re-ordered (245) such that the apodosis (consequent) came before the protasis(antecedent), some speakers accepted it marginally and some accepted it fully:(i) %Maria%MariaMarian=a-nga-chọọmiir-un=a-ŋga-tʃɔɔmiir-uSAM=SA1-COND-be.happy-FVmu-gharukamu-ɣarukaC1-eldera-nga-bhọnẹk-irea-ŋga-βɔnɛk-ireSA1-COND-show.up-PFV‘Mary would be happy if any elder showed up.’[Consultant comment: She cannot talk about wellness of a child first if there are noelders.]Based on the consultant’s comment, the issue why this is not straightforwardly good or badmay have to do, I believe, with the reordering effect (the consequent before the antecedent).Further research is needed to pin down what exactly is going on with cases like (i).165(245) mu-gharukamu-ɣárukaC1-eldera-nga-bhọnẹk-irea-ŋga-βɔnɛk-ireSA1-COND-show.up-PFVMariaMariaMarian=a-nga-chọmiir-un=a-ŋga-tʃɔɔmir-uSAM=SA1-COND-be.happy-FV‘If any elder showed up Mary would be happy.’For conditional sentences, I claim that the conditional Operator nga ‘if’sits in C position above the matrix clause, where it c-commands the subjectin the antecedent as the proposed structure in (249) shows (cf. Progovac1993, 2000):(246) Licensing of D∅ in conditionalsXPŋga‘if’IP...vP3D∅=φPsubj vPv0 VPV0 CPIPI ...As expected, the overt D is not restricted in its distribution therefore itmay be switched with the polarity D, (247).166[Context: A mother has a sick child and only elderly people know thetraditional cure of the disease. There is a specific elder who knows themedicine for the disease. A mother says she will be happy if that eldershowed up] [Adapted from Matthewson, 1999: 90](247) o=mu-gharukao=mu-ɣárukao=C1-eldera-nga-bhọnẹk-irea-ŋga-βɔnɛk-ireSA3-COND-show.up-PFVMariaMariaMarian=a-nga-chọmiir-un=a-ŋga-tʃɔɔmir-uSAM=PST-COND-be.happy-FV‘If an/the elder showed up Mary would be happy.’In Chapter 4, I will explain what the semantic difference is between theovert and the covert D. The next section is about modals where I showthat they are also non-factual operators that can license the polarity D likenegation, the Q-morpheme in polar questions, and conditionals.3.5.4 ModalsTo conclude this section, I consider epistemic modals, which express thestrength of a speaker’s commitment to the truth of a proposition. In Nata,they include possibility modals, the subjunctive mood, and hearsay eviden-tials. My assumption behind treating mood and evidentials as subclasses ofmodality in Nata is that these two also express uncertainties or hypothet-ical situations (see Givón 2018; De Haan 1999; Matthewson et al. 2007;and others)24.3.5.4.1 Epistemic modals license D∅The weak epistemic modal, the adverbial hamwe ‘maybe’ or ‘perhaps’,can license the polarity D∅, (248). When this adverbial modal is used it24. See Aikhenvald (2014, 2015) for the view that in some languages evidentiality is not asubcategory of modality; also see Déchaine et al. (2017) for arguments that some, but notall, evidentials have modal force.167means that the speaker is casting doubt on the validity of the propositionunder the modal (cf. Heim 1992; Giannakidou 1998; Condoravdi (2002);Matthewson 1998, Matthewson et al. 2007; Giannakidou and Mari 2018;and others).[Context: It’s started raining and it’s getting dark. A girl comes home cry-ing; someone pushed her at the playground. She tells Mom and Grandmathat almost everyone has left by now. Mom goes to the playground to askwho did that. Grandma is wondering about Mom’s delay:](248) a. HamweHamweMaybemw-anamu-anáC1-childa-a-mo-bhereki-irea-a-mó-βereke(r)-ireSA1-PST-OM1-call-PFV‘Maybe a kid called her.’b. HamweHamweMaybea-bhwin-ea-βwíin-e2s-find-PFVmw-anamw-aanáC2-childha-yọhá-jɔC16-therea-a-bhor-iria-a-βór-iriSA1-PST-ask-PFV‘Maybe she found a kid there and asked.’The syntactic licensing of the D∅ must be that the modal operatorhamwe, which adjoins to IP as an adjunct, c-commands the DP argumentwith D∅. Given the examples in (248) the Operator must occur higher thanthe subject or the object, c-commanding the polarity element:168(249) Licensing of D∅ in ModalsXPYPHamwe‘maybe’IP...vP(3D∅=φPsubj) vPv0 VPV0 (3D∅=φPObj)We have seen very consistently that an overt D may be used with anyof the operators since it is not syntactically restricted. Indeed, the overt Dcan be used with the modal25.[Context: It is a sunny day and it’s around 3PM and lots of kids are atthe playground. A girl comes home crying; someone pushed her at theplayground. Mom goes to ask who did that but she is delayed comingback. Grandma is wondering about Mom’s delay]:(250) a. HamweHamweMaybeu=mwa-anaa=mw-anáD=C1-childu-yọu-jɔC1-DEMn-a-yar-iren-áá-jar-ireSAM-SA1-run-PFV‘Maybe that kid ran away.’25. When the copula nasal which functions as a strong assertion marker (SAM) or Focusmarker is used, the polarity D∅ cannot be used.(i) HamweHamweMaybe(*n-)a-a-bhwin-e(*n-)a-a-βwíin-eSAM-SA1-PST-find-PFVmw-anamw-aanáC2-childha-yọhá-jɔC16-therea-a-bhor-iria-a-βór-iriSA1-PST-ask-PFV%Intended: ‘Maybe she did find a kid there and asked.’A combination of a strong assertion marker and the polarity D∅ may be ruled out for se-mantic reasons. I do not investigate the restrictions involving this element on this occasion.169b. HamweHamweMayben-a-ku-ghamban-an-a-ku-ɣamban-aSAM-SA1-PROG-talk-FVnu-u=mw-aanana-u=mu-anáwith-D=C1-childu-yọu-jɔC1-DEM‘Maybe she is talking to that kid.’The semantic contrast between the overt D and the polarity D∅ will bediscussed in Chapter 4.3.5.4.2 Subjunctive mood licenses D∅Subjunctive clauses are often analyzed as being non-factual. The preciseformalization differs widely, with proposals treating them as non-assertive(Bolinger 1972; Hooper 1975), as irrealis (Givón 2018; Palmer 1968), asnonveridical (Giannakidou 2009), or as modal (Giorgi and Pianesi (1998);Portner 1998). Whatever the precise mechanism, relevant to the presentanalysis is the fact that, in Nata, the subjunctive patterns together withother non-factual operators in licensing D∅. Example (251) illustrates thelicensing of D∅ in the context of a subjunctive clause, where the subjunctiveclause type is marked by the subjunctive mood marker -ɛ. The (a) exampleillustrates D∅ in subject position, and the (b) example illustrates D∅ inobject position:(251) [Context: An athlete wondering if there are any rope-hurdles.]a. ro-siriro-siriC11-roperoo-n-dich-ẹro-n-ritʃ-ɛ ́SA11-1sg-trip-SUBJVn-te-kw-imeerer-an-te-kw-iméérer-a1sg-NEG-PROG-stop-FV‘Should a rope trip me, I am not stopping.’b. m-bọn-ẹm-bɔn-ɛ ́1sg-find-SUBJVro-siriro-síriC11-ropeenieniIn-ko-bhururuk-an-ko-βúúruk-a1sg-PROG-jump-FVtutujust‘Should I encounter a rope, I am just jumping over.’170Here the subjunctive mood conveys the speaker’s attitude that they arenot committing to the truth of the embedded proposition, which is whatGiannakidou (2009) means by non-veridicality. Since the overt D is notrestricted in its distribution it can be use in subjunctive sentences as (251)shows:(252) [Context: An athlete talking about rope-hurdles on the jumps.]a. o=ro-sirio=ro-sirio=C11-roperoo-n-dich-ẹro-n-ritʃ-ɛ ́SA11-1sg-trip-SUBJVn-te-kw-imeerer-an-te-kw-iméérer-a1sg-NEG-PROG-stop-FV‘Should a/the rope trip me, I am not stopping.’b. n-dọr-ẹn-dɔɾ-ɛ ́1sg-see-SUBJVo=ro-siri,o=ro-síri,o=C11-rope,enieniIn-ko-bhururuk-an-ko-βúúruk-a1sg-PROG-jump-FVtutujust‘Should I see a/the rope, I am just jumping over.’In Chapter 4, I show that there is a clear semantic contrast which forcesNata D choice in all these environments. The last case is about evidentials.3.5.4.3 Evidentials license D∅Evidentiality is a grammatical way of encoding the source of information,e.g., the speaker has firsthand information (i.e., using sensory evidence)or non-firsthand information (i.e., s/he heard about the event or was justtold about it) (Jakobson 1957; Aikhenvald 2004 and many others). Lan-guages differ in the way they code evidentials, with some languages codingthe source of information morphologically (Nuu-chah-nulth), others syn-tactically (Plains Cree) and yet others lexically (English) (see Waldie 2012;Matthewson et al. 2007; Déchaine et al. 2016; and others). I discuss lexi-cal evidentials in Nata, which have to do with the verb of perception -oogu‘hear’ and the verb of report -bhugha ‘say’. The use of these evidential verbsis based on participants presenting the meaning of a proposition p without171committing to the truth of p, hence p may not be asserted. I show thatevidentials, as non-factual operators, may license the D∅.3.5.4.3.1 Reportative evidential verbsWhen the verb -oogu ‘hear’ is used, the speaker may be casting doubt onthe reliability of the information source. In such contexts, the D∅ is used:[Context: B hears a rumour about a death of a child. The person who saidit is not a reliable source and also no wailing was heard. B reports:](253) a. ny-og-uɲ-oog-u1sg-hear-PASSmw-aanamu-anáC1-childa-a-ku-rea-a-ku-ire3sg-PST-die-PFV‘I heard a child died (I doubt it).’b. ny-og-uɲ-oog-u1sg-hear-PASSMariaMariaMariaa-kw-er.ir-ua-kw-éér.ir-uSA1-die-APPL2-PASSmw-aanamu-anáC1-child‘I heard Maria lost a child (I doubt it).’The overt D which is not restricted in its syntactic distribution is alsopossible with this operator.[Context: A/the child had no hope to recover from leukaemia. There waswailing last night. B reports the bad news](254) a. ny-og-uɲ-oog-u1sg-hear-PASSu=mw-aanau=mu-anáD=C1-childa-a-ku-rea-a-ku-ire3sg-PST-die-PFV‘I heard a/the child died (and it’s confirmed).’172b. ny-og-uɲ-oog-u1sg-hear-PASSMariaMariaMariaa-kw-er.ir-ua-kw-éér.ir-uSA1-die-APPL2-PASSu=mw-aanau=mu-anáu=C1-child‘I heard Maria lost a/the child (and it’s confirmed).’If the child is believed to exist but it did not die this will also leadto the use of the overt D. In Chapter 4, we will see that the semantics ofthe Nata Ds aligns with the fact that when interlocutors cast doubt on theinformation source about the existence of the DP referent, D∅ is licensed,and when they trust the information source and believe that the DP referentexists, the overt D is used.3.5.4.3.2 The quotative evidential marker -bhughaLike the evidential verb -oogu ‘hear’, the verb -bhugha ‘say’ is used as aquotative evidential and can license the D∅ as shown below:[Context: Person A and B are walking, and C stops B to inform her of thedeath of a child. C is not a reliable source and also no wailing was heard.B tells A:](255) a. a-bugh-aa-βuɣ-a3sg-say-FVmw-aanamu-anáC1-childa-a-ku-rea-a-ku-ire3sg-PST-die-PFV‘She says (that) a child died.’b. a-bugh-aa-βuɣ-a3sg-say-FVMariaMariaMariaa-kw-er.ir-ua-kw-éér.ir-uSA1-die-APPL2-PASSmw-aanamu-anáC1-child‘She says that Maria lost a child.’173The overt D can also be used with the quotative evidential verb asshown in (256) below26.[Context: Person A and B are walking, and C stops B to inform her of adeath of a child who has been sick for long time. C is a trustworthy person.Before A asks B what was C saying, B says:](256) a. a-bugh-aa-βuɣ-a3sg-say-FVu=mw-aanau=mu-anáD=C1-childa-a-ku-rea-a-ku-ire3sg-PST-die-PFV‘She says (that) a/the child died.’b. a-bhugh-aa-βuɣ-a3sg-say-FVMariaMariaMariaa-kw-er.ir-ua-kw-éér.ir-uSA1-die-APPL2-PASSu=mw-aanau=mu-anáD=C1-child‘She says that Maria lost a child.’As I show in Chapter 4, If the speaker believes that the child existsbut does not believe that it died, the overt D should be used. We haveseen that the evidentials are non-factual operators that can license the D∅.Thus evidentials are consistent with all the other non-factual operators wediscussed above, which govern the meaning of the polarity D∅ appearingin their c-command. The question about what forces D choice with theseoperators will be answered in Chapter 4.26. There is a (dis)confirmational particle ango (here marked as a complementizer (C) thatcan be used with the polarity D∅ as in the example below:(i) AngoAŋgoCa-bhugh-aa-βuɣ-a3sg-say-FVmw-aanamu-anáC1-childa-a-ku-rea-a-ku-ire3sg-PST-die-PFV‘She says a child died (I disconfirm it).’I do not discuss this discourse particle here.1743.6 Summary and conclusionIn this chapter, based on various distributional tests for Nata Ds, I havereached the conclusion that the Nata augment is syntactically a D. I havedistinguished nominal expressions that are predicate nominals from thosewhich are argument nominals. Nominal predicates predictably lack a D,thus appear with the φ-N structure. Such nominals do not denote an indi-vidual, rather they denote a property. On the other hand, argument nomi-nals have the D-layer, either overtly or covertly.The DP containing the null D appear with the same φ-N structure as anominal predicate. However, I have shown that the syntactic distributionof each of these structures vary. Predicate nominals need no licensing,while the polarity-sensitive D∅ is syntactically restricted and it must be c-commanded by a non-factual operator. The operators that license D∅ aresummarized in Table 3.1.Table 3.1: Operators that license the polarity-sensitive D∅Environments Op licensing D∅ Subj Obj Diagnostics fromNegation Sentential Neg 3 3 Klima 1964Intonational Neg 3 3 Swart 2009Lexical Neg 7 3 Yoon 2013Negative light v 7 3 Halpert 2012Interrogatives Yes/No Qns 3 3 Cheng 1991WH-Qns 7 7 Pesetsky 1987Conditionals If-clauses 3 3 Matthewson 1999Modals Modals 3 3 Matthewson 1998Subjunctive mood 3 3 Giannakidou 2002Hearsay evidentials 3 3 Matthewson 1998I argued that subject DPs with D∅ are generated in Spec, vP and arelicensed under the c-command domain of Negation. This correctly rulesout licensing of D∅ in subjects by lexical negation and negative light verbswhich do not c-command the subject DP with a D∅. I have shown thatunaugmented NPs referred to as ”bare NPs” in much Bantu literature (see175Progovac 1993; Hyman and Katamba 1993; Visser 2008; Halpert 2012; andothers), in Nata, are in fact DPs containing a polarity D∅. Nata argumentDPs cannot be bare NPs/φPs as demonstrated in this chapter; only nominalpredicates can be bare NPs/φPs.On the overt Ds, I have shown that the overt Ds appear as a syntacticdefault; as such, they appear in a variety of syntactic environments andthey need no licensing. Having established that the Nata augment is a D,in the next chapter I present and then argue for the semantic proposal thatunderlies the two Ds presented below:(257) a. Argument DPs: Db. Polarity-sensitive Argument DPs: D∅I will claim that D choice in (257) is forced by the semantic notion of ‘ex-istence’.176Chapter 4The semantics of Nata D4.1 IntroductionBefore we delve into my proposal about the semantic function of aug-ments/Ds in Nata, I wish to make a quick recap of what we have learnedso far about Nata augments. In the previous chapter, I established thatthe augment is the realization of the functional category D. In chapter 2,I examined various potential D contrasts, and concluded that the general-izations in (258) are upheld in Nata:(258) a. Augments/Ds do not encode (in)definiteness.b. Augments/Ds do not encode (non)specificity.c. Augments/Ds do not encode deictic features.d. Augments/Ds do not encode Case.e. Augments/Ds do not encode mass-count distinction.f. Augment/Ds are not domain restriction elements.In this chapter, I present and then argue for a proposal about the se-mantic function of the Nata D. I seek to provide a definitive answer to the177question of what semantic principle underlies the contrast between the twoNata DP types in (259):(259) a. Argument DPs with an overt D.b. Argument DPs with a covert D/D∅.I claim that the contrast between Nata DPs with an overt D in (259a) andthe DPs with a covert D/D∅ in (259b) involves the notion of ‘existence’ (seeGivón 1978; Matthewson 1998; Van de Velde 2005; Gambarage 2012). Iwill show that the Nata D system is strikingly similar to the D system ofSt’át’imcets (Lillooet Salish), hence the two represent an emerging typologyof ‘speaker-oriented existence Ds.’ Within a formal account, I claim thatNata Ds require an analysis involving choice functions (Reinhart 1997). Iprovide an existence-based choice function account following Matthewson(1999). However, since languages differ in the way they encode existence,i.e., the Nata belief-of-existence Ds differ from the assertion-of-existence Dsin St’át’imcets, I will discuss the implication of Nata Ds for this approach.This chapter is organized as follows. In §4.2 I introduce the system ofexistence Ds. In §4.3 I present my proposal for Nata Ds and discuss theimpetus for treating augments as speaker-oriented existence Ds. In §4.4 Idiscuss the main properties of speaker-oriented existence Ds. In §4.5, I dis-cuss other correlated properties of existence Ds where I argue that they fol-low from the main properties of speaker-oriented existence Ds. Thus, §4.4and §4.5 highlight the many parallels found between Nata and St’át’imcetsD systems. In §4.6 I show that languages encode semantic notions of ex-istence slightly differently. Following this, I discuss various factors for pa-rameterizing existence Ds in Bantu and Salish. The choice function accountis presented in §4.7. In §4.8 I present further predictions and the theoreticalimplications. In §4.9 I conclude.4.2 Defining Existence DsMatthewson (1998) provides an analysis of indefinite Ds in St’át’imcetswhich encode ‘assertion of existence.’ According to Matthewson, the overt178Ds X...-a, (where X is a placeholder for different elements encoding deicticdistinctions), as in (260a,b), consistently encode assertion of existence; onthe other hand, ku (260c) encodes non-assertion of existence, though notby denying it:(260) St’át’imcets, Matthewson 1998: 55-56Encoding Existencea. ʔá ’z-ən-ašaz’-en-asbuy-TR-3ERG[ti[ti[DETš ’cúqʷa ’z-a]sts’úqwaz-a]fish-DET]kʷ-škw-sDET-NMLZRSophieSophieSophie‘Sofie bought [a fish].’=9x, x a fish, Sofie bought xb. xʷʔazcw7aozNEGk-škw-sDET-NMLZRʔá ’z-ən-ašá ’z-en-asbuy-TR-3ERG[ti[ti[DETš ’cúqʷa ’z-a]sts’úqwaz’-a]fish-DET]kʷ-škw-sDET-NMLZRSophieSophieSophie‘Sofie didn’t buy [a fish].’=9x, x a fish, ¬Sofie bought xc. xʷʔazcw7aozNEGkʷ-škw-sDET-NMLZRʔá ’z-ən-ašáz’-en-asbuy-TR-3ERG[kʷ-u[ku[DETš ’cúqʷa ’z]sts’-úqwaz’]fish]kw-škw-sDET-NMLZRSophieSophieSophie‘Sofie didn’t buy [a/any fish].’=¬9x, x a fish, Sofie bought xWithin Bantu, Givón illustrates using Bemba data that DPs with an overtD refer to entities that ‘exist’ in the world of discourse. However, Givónopts to use the term ‘referentiality’, which I will not adopt here because of179the confusion that may arise given that some scholars use that term syn-onymously with ‘specificity.’ The Bemba D distinction is consistent withMatthewson’s characterization of the assertion of existence Ds in St’át’imcets.In (261a,b) for instance, Givón shows that nominals with overt Ds (thosewith an initial vowel and a class prefix (VCV nominals, his term) denoteexistence/referentiality, while their counterparts (those without an overtD) always fall under the scope of a non-factual operator (negation, modals,conditional, etc.) where they denote non-existence/non-referential mean-ing, (261c). Note that (261d) is bad because there is no licensor.(261) D contrast in Bemba [Adapted from Givón 1978: 301]a. u=mu-anaD=C1-childa-a-someneSM1-PST-readi=ci-taboD=C7-book‘A/the child read a/the book.’b. u=mu-anaD=C1-childt-a-a-someneNEG-SM-PST-readi=ci-taboD=C7-book‘A/the child did not read a/the book.’c. u=mu-anaD=C1-childt-a-a-someneNEG-SM-PST-readci-taboC7-book‘A/the child did not read any book.’d. *u=mu-ana*D=C1-childa-a-someneSM1-PST-readci-taboC7-book‘*A/the child read any book.’Based on this characterization, both Givón (1978) and Matthewson(1998) define existence in the following terms:(262) Informal definition of existenceIt involves, roughly, the speaker’s intent to ‘refer to’ or ‘mean’ anominal expression to have non-empty references– i.e. to ‘exist’ –180within a particular universe of discourse (i.e., not necessarily withinthe real world) Givón (1978: 293-294).Below I present my full proposal where I claim that the D contrast inNata is also based on the notion of existence. Given that not all languagesencode exactly the same notion of existence, the type of existence encodedin Nata Ds is extensively discussed in this chapter.4.3 The proposal: Nata Ds encode ‘existence’The existence D distinction found in St’át’imcets and Bemba Ds also de-rives the D choice in Nata. As the examples in (263a,b) show, when Nataspeakers intend to commit to the existence of a referent contained in theproposition, the overt D is used. When speakers do not wish to commit toexistence, the only option available is to use the null D/polarity sensitiveD, which must fall under the c-command domain of a non-factual operator,e.g., negation, (263c). For this reason, speakers will never use the polarityD in affirmative/positive declarative sentences as they have no licensor,(263d).(263) D contrast in Nataa. MakuruMakurúMakurua-ka-ghor-aa-ka-ɣór-aSA1-PST-buy-PFVe=ghi-tabhoe=ɣí-taβoD=C7-book‘Makuru bought a/the book.’=9x [book(x) & [Makuru bought x]]b. MakuruMakurúMakurut-a-a-ghor-iret-a-a-ɣor-iréNEG-SA1-PST-buy-PFVe=ghi-tabhoe=ɣí-taβoD=C7-book‘Makuru did not buy (a/the) book.’=9x [book(x) & ¬ [Makuru bought x]]181c. MakuruMakurúMakurut-a-a-ghor-iret-a-a-ɣor-iréNEG-SA1-PST-buy-PFVghi-tabhoɣí-taβoC7-book‘Makuru did not buy any book.’¬[9x [book(x) & Makuru bought x]]d. *Makuru*MakurúMakurua-ka-ghor-aa-ka-ɣór-aSA1-PST-buy-PFVghi-tabhoɣí-taβoC7-bookIntended: ‘Makuru bought a/the book.’The semantic core of the D distinction in St’át’imcets, Bemba and Natais the notion of ‘existence’ consistently. Based on Givón’s characterizationof Bemba Ds, we may simply present the contrast expressed in the Nata Dsystem for the examples in (263) as (264):(264) Determiner choice in Nata (Preliminary)a. Overt D: conveys the speaker’s commitment to existence of anentity for the noun phrase.b. Covert D: conveys a lack of speaker commitment to existence.While (264) seems at first glance to be a fair characterization of Nata Ds,within the emerging typology of speaker-oriented existence Ds discussed inthis thesis, (264) cannot be a sufficient generalization to capture the dif-ferent existence D distinctions. As Matthewson illustrates, in St’át’imcets,speakers fail ‘positively’ to assert existence in examples similar to (265), inwhich a referent has not materialized yet. I show below that this is not thecase with Nata.(265) a. I will marry the next chief of Fountain (whoever it is).b. I will donate a chair for the new school.182I claim that both the Nata and the St’át’imcets D systems indeed have ex-istence as their semantic core. The core difference between the two systemsrelates to cases where the speaker believes in the existence of a referent ofa noun phrase, but does not have positive evidence for its existence. To bemore precise, in St’át’imcets existence is asserted, while in Nata existenceis believed. Thus, the two speaker-oriented existence Ds behave differentlywith respect to the requirement for asserting existence. The Nata overt Dbehaves as a weaker version of St’át’imcets assertion-of-existence Ds, as aspeaker’s personal evidence for the referent is not a requirement for encod-ing existence in Nata. To account for the general behaviour of Nata Ds, Iclaim that the generalization in (266) is crucial:(266) Informal definitionNata Ds encode ‘speaker’s belief of existence’: the speaker believesthat a nominal expression has a non-empty reference– i.e., a refer-ent ‘exists’ within a particular universe of discourse (not necessarilywithin the actual world).The Nata D contrast proposed in (264) has to be revised as in (267):(267) Determiner choice in Nata (Final)a. Overt D: conveys the speaker’s commitment to a belief of exis-tence of an entity for the noun phrase.b. Covert D: conveys a lack of speaker commitment to a belief ofexistence for the noun phrase.The interpretive contrast given in (267) is consistently supportedthroughout the Nata data1. I discuss data from both Nata and St’át’imcetsand claim that the parameter of variation between the two D systems is aspresented in Table 4.1:1. Note that ‘belief of existence’ is a cover term for the description of Nata Ds. One mayalso choose to call Nata Ds presumptive, assumptive, suppositional, etc. Thanks to MichaelRochemont for this observation.183Table 4.1: Requirements for use of the existence DsSt’át’imcets NataLocus of parametric variationExistence of an entity is only believed 7 3Existence of an entity is always asserted 3 7I will argue that the Nata belief-of-existence Ds do not have a require-ment for asserting existence, as is the case in St’át’imcets. Based on thisfact, I propose a split in speaker-oriented-existence Ds as in (268):(268) Speaker-oriented Existence DsAssertion of existenceeg., St’át’imcetsX...-a kuBelief of existenceeg., NataOvert D Covert D∅While the contrast in the St’át’imcets Ds is expressed overtly, the lackof phonological content for the polarity sensitive D∅/the covert D in Natais also explained within the syntactic-semantic mapping. The covert D isinterpretable at LF as marking DPs with non-existential interpretation. Thismapping for Nata is presented in (269) below:(269) Syntactic-Semantic mappingSemantic function Phonology Overt syntaxCommitment to existence Pronounced D Argument DP: DLack of commitment to existence Unpronounced D Argument DP: D∅Property denoting N/A Predicate: NPIn (269), the D contrast is expressed overtly in the syntax where it in-volves the deployment of the phonological content of D. My claim in Chap-ter 3 was that property-denoting nominals differ from the DPs containing184a D∅ in that predicate nominals do not have a D shell, but argument nom-inals in Nata are DPs (cf. Stowell 1989; Longobardi 1994; Déchaine andTremblay 2011; and others)2. The motivation for analyzing Nata overt Dsas existence Ds is discussed below.A major motivation for the proposal that Nata Ds encode the notionof existence comes from the many parallels that Nata Ds have with the‘speaker-oriented existence Ds’ in languages such as Bemba (Bantu) (Givón1978), Luganda (Bantu) (Van De Velde 2005) and St’át’imcets (Salish)(Matthewson 1998; 1999).Givón argues clearly that the core notion that forces Bemba D choiceis that an entity ‘exists’ in the world of discourse. Givón’s definition ofexistence is repeated below:(270) Givón’s definition of existenceIt involves, roughly, the speaker’s intent to ‘refer to’ or ‘mean’ anominal expression to have non-empty references– i.e. to ‘exist’ –within a particular universe of discourse (i.e not necessarily withinthe real world)’ Givón (1978: 293-294).More explicitly, Matthewson acknowledges that the Bantu D system is sim-ilar in many respects to the St’át’imcets system, as she remarks:Givón makes the cross-linguistic claim that nominals fallingunder the scope of a possible modality or negative modalitymay receive either a referential (i.e. assertion of existence) ornon-referential interpretation; otherwise all nominals are inter-preted referentially (1978: 294). This is the case in Bembaas well as in Salish. [...]Givón’s definition is based on Bemba(Bantu), whose determiner system shows similarity with Salishsystems. [Matthewson 1998: 55, 69]2. As I argued previously, I will maintain unaugmented NPs (referred to as “bare NPs” inmuch Bantu literature (cf. Progovac 1993; Hyman and Katamba 1993; Visser 2008; Halpert2012)), in Nata are DPs containing a covert D.185The parallels between Bantu Ds and the Salish ‘assertion of existence’Ds have also been observed by Van de Velde (2005) in Luganda:[...]the recurrent observation that objects of negative verbs andnouns modified by a question word do not have an augmentis reminiscent of the role of the determiner in Salish languages.According to Matthewson (1998), Salish determiners encode as-sertion of existence. In the Salish languages there are determin-ers that assert the existence of a referent and there are deter-miners that do not assert existence (without, however, denyingit). The non-assertion of existence marker in St’át’imcets is ku.It is restricted in its syntactic distribution... As I see it, the de-terminer ku in St’át’imcets corresponds to a certain extent to theabsence of the augment in Ganda (J15), whereas the assertionof existence determiners correspond to the augment.[Van de Velde 2005: 16]While I agree also that the two language families (Bantu and Salish)have ‘existence’ as the core semantics of their Ds, I argue below (and inChapter 5) that Nata, like Luganda, encodes the notion of belief of existencewhich is slightly different from the notion of assertion-of-existence foundin St’át’imcets and Bemba systems. The parallels between Nata belief-of-existence Ds and St’át’imcets assertion-of-existence Ds provide further evi-dence that speaker-oriented existence is robustly available as a determinerdistinction (see Gambarage and Matthewson 2019). Below I will discussthe two different notions of existence which introduce the locus of varia-tion within the speaker-oriented existence Ds. I first discuss core propertiesof the speaker-oriented existence Ds.4.4 Properties of speaker-oriented existence DsThe Nata belief-of-existence system shares some core properties with otherexistence Ds, e.g., assertion-of-existence Ds in St’át’imcets. I propose that(271) provides the defining core properties of existence Ds.186(271) a. Existence Ds are speaker-oriented systems.b. Existence Ds encode existence (i.e., by assertion or belief).I discuss these properties with data from both Nata and St’át’imcetsstarting with the speaker-based property.4.4.1 Speaker-oriented existence DsSpeaker-oriented existence Ds, unlike the speaker-hearer English systemwhich involves common ground knowledge, have a negative setting ofthe common ground parameter, i.e., do not access the hearer’s knowledge(Matthewson 1998). Both in Salish and Bantu (Nata) the speaker is the solearbiter. This does not mean that assertion-of-existence or belief-of-existenceDs cannot feature in definite/familiar discourse contexts. Matthewson, forinstance, illustrates that the D X...-a Ds in St’át’imcets only encode assertionof existence and not a novel-familiar distinction. The same is true in Nata.Belief-of-existence Nata Ds are neutral with respect to the novel-familiardistinction; as a result, they are used both in novel and in familiar contexts(see Van de Velde 2019; and Chapter 2).4.4.2 Ds encode a core notion of existenceIn both systems, the D choice is forced by the notion of ‘existence’, and notdefiniteness or specificity. I show that existence Ds may encode the exis-tence of either actual things or non-actual things in the world of discourse.4.4.2.1 Existence with actual referentsMatthewson argues explicitly that D choice in St’át’imcets is based on as-sertion of existence. She illustrates that the DP ti sts’úqwaz’a with the Dti…a in (272) is interpreted with existential force whereby it asserts theexistence of a fish, as the informal semantics show. On the other hand, theDP with the polarity D ku is interpreted under the scope of a non-factualoperator (negation, conditionals, modals, and question-morphemes) where187it is associated with a non-existential interpretation. In (273) ku is licensedby negation.(272) Existential Force [St’át’imcets, Matthewson 1998: 55]a. ʔá ’z-ən-ašaz’-en-asbuy-TR-3ERG[ti[ti[DETš ’cúqʷa ’z-a]sts’úqwaz-a]fish-DET]kʷ-škw-sDET-NMLZRSophieSophieSophie‘Sofie bought [a fish].’=9x, x a fish, Sofie bought xb. xʷʔazcw7aozNEGk-škw-sDET-NMLZRʔá ’z-ən-ašá ’z-en-asbuy-TR-3ERG[ti[ti[DETš ’cúqʷa ’z-a]sts’-úqwaz’-a]fish-DET]kʷ-škw-sDET-NMLZRSophieSophieSophie‘Sofie didn’t buy [a fish].’=9x, x a fish, ¬Sofie bought x(273) No-existential Force [St’át’imcets, Matthewson 1998: 56]a. xʷʔazcw7aozNEGkʷ-škw-sDET-NMLZRʔá ’z-ən-ašáz’-en-asbuy-TR-3ERG[kʷ-u[ku[DETš ’cúqʷa ’z]sts’-úqwaz’]fish]kw-škw-sDET-NMLZRSophieSophieSophie‘Sofie didn’t buy [a/any fish].’=¬9x, x a fish, Sofie bought xThe Nata belief-of-existence D/the overt D can be used in contextswhich assert existence, (274). Here, the DP e=ghi-tabho ‘a/the book’ is in-terpreted with existential force paralleling the use of the St’át’imcets overtD ti…a in (272) above:188(274) Existential Interpretation [Nata]a. MakuruMakurúMakurua-ka-ghor-aa-ka-ɣor-aSA1-PST-buy-PFVe=ghi-tabhoe=ɣí-taβoD=C7-book‘Makuru bought a/the book.’=9x [book(x) & [Makuru bought x]]b. MakuruMakurúMakurut-a-a-ghor-iret-a-a-ɣor-ireNEG-SA1-PST-buy-PFVe=ghi-tabhoe=ɣí-taβoD=C7-book‘Makuru did not buy (a/the) book.’=9x [book(x) & ¬ [Makuru bought x]]Likewise, the non-belief-of-existence D must be interpreted under thescope of a non-factual operator such as negation to render the non-existential interpretation:(275) Non-existential Interpretation [Nata]MakuruMakurúMakurut-a-a-ghor-iret-a-a-ɣor-ireNEG-SA1-PST-buy-PFVghi-tabhoɣí-taβoC7-book‘Makuru did not buy any book.’¬[9x [book(x) & Makuru bought x]]The polarity/non-belief-of-existence D in Nata parallels the determinerku in St’át’imcets, which fails to assert existence; hence, both Ds are notassociated with existential interpretation. Note crucially that while DPsinterpreted under the scope of negation may be argued to deny existenceof a referent denoted by the NP, speakers do not always deny existence.When a polarity DP is interpreted under the scope of a modal or a questionmorpheme, the speaker is not denying existence, but rather conveying alack of commitment to existence (see Matthewson 1998, 1999).1894.4.2.2 Existence with non-actual referentsBoth Givón (1978) and Matthewson (1998) show that in certain contexts,existence Ds can be used to talk about referents that exist only in the mindof the speaker, i.e., the Ds can be used in non-actual worlds such as vi-sions and dreams. Thus, the use of the assertion-of-existence D ti...a in(276) is consistent with the existential interpretation of the policemen inthe speaker’s dream (i.e., the policeman is a participant that exists only inthe mind of the speaker).(276) Existence Ds in dreams [St’át’imcets, Matthewson 1998: 132]a. kʷʔíkʷlaxʷ-kankw7íkwlacw-kandream-1SG-SUBkʷ-a-škw-a-sDET-PROG-NMLZRtúp-u ’n-aštúp-un’-aspunch-TR-3ERGš-Johns-JohnNMLZR-John[ti[tiDETplíšmən-a]plísmen-a]policeman-EXIS‘I dreamed that John hit a policeman.’b. *kʷʔíkʷlaxʷ-kan*kw7íkwlacw-kandream-1SG-SUBkʷ-a-škw-a-sDET-PROG-NMLZRtúp-u ’n-aštúp-un’-aspunch-TR-3ERGš-Johns-JohnNMLZR-John[ku[kuNON-EXISplíšmən]plísmen]policemanIntended: ‘I dreamed that John hit a policeman.’The same is true in Nata: the belief-of-existence Ds can also be used indescribing dreams or visions. Example (1) talks about non-actual worlds.The speaker is only describing mythical creatures such as ogres that hedreamt about3.3. One may wonder if it is possible to convey existence in non-actual world using a thirdperson, for instance, Maria dreamed about ogres fighting with each other. Unlike St’át’imcetswhere lack of speaker personal knowledge is linked with the use of an non-assertion of190(277) Description of dream [Nata]a. n-ka-rọọt-an-ka-rɔɔt-aSAM-1sg-PST-drean-FVa=amanania=ma-naniD=C6-ogresgha-ra-rwaan-aɣa-ra-ruan-aSA-PROG-fight-FV‘I dreamed about ogres fighting with each other.’b. *n-ka-rọọt-a*n-ka-rɔɔt-aSAM-1sg-PST-drean-FVma-nanima-naniC6-ogresgha-ra-rwaan-aɣa-ra-ruan-aSA-PROG-fight-FVIntended: ‘I dreamed about ogres fighting with each other.’Below I show that all other correlated properties of existence Ds followfrom the core properties of speaker-oriented existence Ds.4.5 Correlated properties of speaker-orientedexistence DsI discuss the correlated diagnostics of speaker-oriented existence Ds andclaim that they follow from the core properties of the speaker-oriented ex-istence Ds discussed in §4.4 above. These are summarized below:existence D, in Nata it is possible for the speaker to use the overt D in such contexts if hehas reason to believe that the subject of the sentence.If the speaker has a reason to notcommit to a belief that Maria dreamt s/he will embed the proposition under a reportativeor quortative verb which will license the polarity D, and render the non-belief of existenceinterpretation of ogres; something like:(i) Nyoghw-aɲ-oɣw-a1sg-hear-FVM. a-ka-rọọt-aM. a-ka-rɔɔt-aM. SAM-1sg-PST-drean-FVamananima-naniC6-ogresgha-ra-rwaan-aɣa-ra-ruan-aSA-PROG-fight-FV‘I hear (that) M. I dreamed about ogres fighting with each other.’For the discussion about evidential lexical verbs see Chapter 3.191Table 4.2: Summary of correlated diagnosticsCorrelated diagnostics of Ds in the two languages St’át’imcets NataEncoding definiteness 7 7Encoding specificity 7 7Polarity sensitivity 3 3Some of the Nata data discussed here are repeated from Chapters 2 and3 for the purpose of comparing the Nata belief-of-existence system and theSt’át’imcets assertion-of-existence system.4.5.1 Speaker-oriented Ds do not encode definitenessThe lack of a definiteness distinction in St’át’imcets and Nata follows frommy analysis that Ds in these languages are speaker-based; they do not accessthe hearer’s knowledge. Matthewson argues explicitly that Ds in Salish donot make any familiar-novel distinction, as the English system does. TheSt’át’imcets data in (278) show that the assertion-of-existence D ti...a is usedwhen the individual is novel and when it is familiar:(278) No familiar-novel contrast [St’át’imcets, Matthewson 1998:34]a. hú ’y-ɬkanhuy’-lhkangoing.to-1SG.SUBptakʷɬ,ptakwlh,tell.storyptakʷɬ-minptakwlh-mintell.story-APPLlčʔalts7ahere[ti š ’mə́ ’mɬač-a]…[ti smém’lhats-a] …[DET woman(RED)-DET]…‘I am going to tell a legend, a legend about [a girl]i.’ (novel)b. waʔwa7PROGkʷuʔku7QUOTʔílalílalcrylátiʔláti7DEIC[ti šmə́ ’mɬač-a][ti smém’lhats-a][DET woman(RED)-DET]‘[The girl]i was crying there.’ (familiar)192The novelty-familiarity distinction is also irrelevant in Nata. The sameD is used with novel and in familiar referents, as shown below:(279) a. hayọhayotherekarẹkárɛlong.ago[o=mu-tẹmi][o=mu-tɛḿi][D=C1-chief]a-gha-simok-aa-ɣa-símok-aSA1-PST-rise-FV‘Long ago a chief was enthroned.’ (novel)b. o-ro-sikoo-ro-síkoD=C11-dayru-mwe,rú-mweC11-one[o=mu-tẹmi][o=mu-tɛḿi][D=C1-chief]a-gha-kom-aa-ɣa-kóm-aSA1-PST-gather-FVa=bha-toa=βáa-toD=C2-peoplebha-acheβá-atʃeC2-his‘One day the chief gathered his people.’ (familiar)The St’át’imcets assertion-of-existence Ds X...-a and the Nata belief-of-existence D do not encode definiteness. What the interpretations of DPslike ti šmə́ ’mɬač-a ‘a/the girl/woman’ and o=mu-tẹmi ‘a/the chief’ have incommon here is that they are associated with an existential interpretation.Next I consider specificity.4.5.2 Speaker-oriented Ds do not encode specificityThe data in both languages show that existence Ds do not contrast for speci-ficity. The Ds can be used both in specific and non-specific contexts. InSt’át’imcets the same assertion-of-existence D can be used to refer to a spe-cific referent or a non-specific referent4:(280) [Context: The speaker has just heard on the telephone that ateacher she knows named Leo is coming. She reports this infor-4. Note that in these examples the enclitic =a portion of the determiner is phonologicallydeleted following the auxiliary wa7.193mation to a colleague] [St’át’imcets, Matthewson 1998: 42)]:xʷuz’cuz’going.tokʷuʔku7QUOTčʔašts7ascome[ti waʔ čuná ’m-xal][ti wa7 tsunám’ -xal][DET PROG teach-INTR][Specific]‘A teacher is coming.’(281) [Context: The speaker has just heard on the telephone that ateacher is coming (she does not know who). She reports this infor-mation to a colleague] [St’át’imcets, Matthewson 1998: 42)]:xʷuz’cuz’going.tokʷuʔku7QUOTčʔašts7ascome[ti waʔ čuná ’m-xal][ti wa7 tsunám’-xal][DET PROG teach-INTR][Non-specific]‘A teacher is coming.’Similarly in Nata, Ds do not encode specificity. The same D is used torefer both to specific, (282) and non-specific referents, (283).(282) [Context: The speaker has just heard on the telephone that a farmershe knows is coming. She reports this information to a colleague:][Context adapted from Matthewson 1998]a. o=mo-remio=mó-rem-iD=C1-farm-FVn=aa-ku-cha [Specific]n=aa-ku-tʃáSAM-3sg-FUT-FV‘A farmer is coming.’b. * mo-remi* mó-rem-iC1-farm-FVn=aa-ku-chan=aa-ku-tʃáSAM-3sg-FUT-FVIntended: ‘A farmer is coming.’194(283) [Context: The speaker has just heard on the telephone that a farmeris coming (she does not know who). She reports this informationto a colleague:] [Context adapted from Matthewson 1998]a. o=mo-remio=mó-rem-iD=C1-farm-FVn=aa-ku-cha [Non-specific]n=aa-ku-tʃáSAM-3sg-FUT-FV‘A farmer is coming.’b. * mo-remi* mó-rem-iC1-farm-FVn=aa-ku-chan=aa-ku-tʃáSAM-3sg-FUT-FVIntended: ‘A farmer is coming.’These examples show that when the St’át’imcets assertion-of-existenceDs and the Nata belief-of-existence Ds are used, they encode existence ofa referent of a noun phrase. They do not care about whether the referentdenoted by a noun phrase is specific or non-specific, hence the D distinctionin these languages is not based on the notion of specificity.4.5.3 Polarity Ds must be licensedAs we saw in Chapter 3, polarity Ds are syntactically restricted in that theymust be licensed by a non-factual operator. Both non-assertion-of-existenceand non-belief-of-existence Ds are polarity sensitive Ds, hence must be in-terpreted under the scope of a non-factual operator, where they cannot beassociated with an existential interpretation. It is important to note thatlanguages whose Ds do not contrast for the notion of existence may dis-guise this contrast. Consider for instance, a system like English, (284),or Okanagan, (285), whose D systems do not contrast for existence. Theindefinite Ds can be used both in assertion-of-existence and non-assertion-of-existence contexts. The Okanagan data are repeated from Chapter 2.195(284) a. I didn’t talk to a man who yelled at me. [existential]b. I didn’t talk to a man/any man. [non-existential](285) Ds do not contrast for existence [Okanagan, Lyon 2011: 26)]a. iʔDetsqəltmíxʷmanlutaʔNEGkaʔkíc-ísfind.(DIR)-3SG.ERGiʔDetsənkl ’caʔsqáxaʔhorse‘The man didn’t find the horses.’b. iʔDetsqəltmíxʷmanlutaʔNEGkaʔkíc-ísfind.(DIR)-3SG.ERGiʔDetsənkl ’caʔsqáxaʔhorse‘The man didn’t find any horses.’Unlike in English and Okanagan, D choice is forced by existential in-terpretation in both St’át’imcets and Nata. As we have seen from the out-set, the Ds used in contexts where a referent is either asserted or believedto exist must be morphologically distinct from the ones used in contextsthat render a non-existential interpretation. I discuss such contexts first inSt’át’imcets, then in Nata.4.5.3.1 Licensing in St’át’imcetsMatthewson demonstrates that the polarity D ku in St’át’imcets is syntacti-cally restricted and must fall under the c-command domain of a non-factualoperator such as negation (286), a question morpheme (287), a modal(288), or a conditional operator, (289). In such environments, the speakerdoes not intend an existential interpretation.(286) Negation licenses ku [St’át’imcets; Matthewson, 1999: 88]cw7aozNEGkw-sDET-NMLZRáts’x-en-assee-TR-ERG[kuNON.EXIS.DETsqaycw]man‘She didn’t see any men.’ ( ̸= ‘She didn’t see the men.’)196(287) Yes/no Q licenses ku [St’át’imcets; Matthewson, 1999: 88]áts’x-en-lhkácwsee-TR-2SG.SUBJhaYNQ[ku sqaycw][DET man]‘Did you see a man/any man?’.(288) The modal kelh licenses ku [St’át’imcets; Matthewson, 1998: 54]təxʷp-mín-ɬkantəcwp-mín-lhkanbuy-APPL-1SG.SUBJkɬkelhmight[kʷu[ku[DETpukʷ]pukw]book]natxʷnatcwtomorrow‘I might buy [a book] tomorrow’.[Context: Mary will be happy if any elders come, but that’s impossible,because there are no elders in this community].(289) Conditional licenses ku [St’át’imcets, Matthewson 1999: 90]cuz’going.totsa7cwhappykw-sDET-NMLZRMaryMarylh-t’íq-asHYP-arrive-3CONJ[ku[DETqelhmémen’]old.person(DIMIN)]‘Mary will be happy if any elder comes.’Matthewson states that when a speaker has an entity in mind thatmatches the NP description, the assertion-of-existence D must be used. Inthis case, the DPs containing a determiner ending with a take wide scopewith respect to these operators, where they receive an existential interpre-tation. Compare the D data in the modal contexts in (288) and in condi-tionals in (289) and the data below, in which the assertion-of-existence Dis used.197(290) Modal environment [St’át’imcets, Matthewson 1998: 54]təxʷp-mín-ɬkantəcwp-mín-lhkanbuy-APPL-1SG.SUBkɬkelhmight[ti[ti[DETpúkʷ-a]púkw-a]book]natxʷnatcwtomorrow‘I might buy a/the book tomorrow’.[Context: There are a bunch of elders in this community. Mary dislikesmost of these elders and doesn’t want them to come. There is just one elderwho she wants to come].(291) Conditional environment [St’át’imcets, Matthewson 1999: 90]cuz’going.totsa7cwhappykw-sDET-NOMMaryMarylh-t’íq-asHYP-arrive-3CONJ[ti[DETqelhmémen’-a]old.person(DIMIN)-EXIS]‘Mary will be happy if an elder comes.’4.5.3.2 Licensing in NataWe saw in St’át’imcets that the polarity sensitive D must be licensed; thesame is true for Nata D∅. As we saw in Chapter 3, the D∅ may be licensedby negation (292), the Q-morpheme (293), the modal operator (294), andthe conditional morpheme (295). With all of these operators, when speak-ers are not conveying belief of existence of the DP referent, they alwaysswitch to using the D∅.(292) Negation licenses D∅ [Nata]a. ghi-tabhoɣí-taβoC7-bookghi-ta-a-hun-ireɣi-ta-a-hun-ireC7-NEG-PST-hit-PFVMakuruMakuruMakuru‘No book hit Makuru.’198b. MakuruMakurúMakurut-a-a-ghor-iret-a-a-ɣor-ireNEG-SA1-PST-buy-PFVghi-tabhoɣí-taβoC7-book‘Makuru did not buy any book.’(293) [Context: Speaker is not sure if there was any child.]a. angoangoQmw-aanamu-ánaC1-childa-ka-rọr-aa-ka-rɔŕ-a3s-PST-see-FVMaria?María?Maria‘Did any child see Maria?’b. angoangoQMariaMaríaMariaa-ka-bhọn-aa-ka-βɔń-a3s-PST-find-FVmw-aana?mu-ána?C1-child‘Did Maria find any child?’(294) [Context: B is not sure if there were children at the playground.]a. hamwehamwemaybemw-anamw-anáC1-childn-a-areFOC-SA1-PST-aréFOC-SA1-PST-bea-ra-bharaanaa-ra-βaraanaSA1-PROG-playha-yọhá-jɔthere‘Maybe a kid was playing there’.b. hamwehamwemaybeMakuruMakuruMakurua-ka-bhọn-aa-ka-βɔń-a3sg-find-PFVmw-anamw-anáC1-childha-yọhá-jɔC16-there‘Maybe Makuru found a kid there’.(295) [Context: Maria has a sick child and she would be happy if anyelder comes and shows her a cure but that’s impossible, becausethere are no elders in this community] [Adapted from Matthewson1991999:90].mu-gharukamu-ɣárukaC1-eldera-ngi-i-ch-irea-nga-i-tʃ-íreSA3-COND-REFL-come-PFVMariaMariaMarian=a-nga-chọmir-un=a-ŋga-tʃɔmer-uSAM=PST-COND-be.happy-PASS‘If any elder came Maria would be happy.’As we saw in St’át’imcets, if a Nata speaker intends to commit to thebelief of existence of an entity, s/he has to switch and use the belief-of-existence D. As can be seen below, the overt D can be used with negation(296), with the Q-morphemes (297), with the modal (298), and with theconditional (299). In all these cases the DPs with the overt D are associatedwith an existential interpretation.(296) a. e=ghi-tabhoe=ɣí-taβoD=C7-bookghi-ta-a-hun-ireɣi-ta-a-hun-ireC7-NEG-PST-hit-PFVMakuruMakuruMakuru‘A/the book did not hit Makuru.’b. MakuruMakurúMakurut-a-a-ghor-iret-a-a-ɣor-ireNEG-SA1-PST-buy-PFVe=ghi-tabhoe=ɣí-taβoD=C7-book‘Makuru did not buy a/the book.’(297) [Context: Speaker is inquiring about some child]a. angoangoQu=mw-aanau=mw-áanaD=C1-childa-ka-rọr-aa-ka-rɔŕ-a3s-PST-see-FVMaria?María?Maria‘Did a/the child see Maria?’200b. angoangoQMariaMaríaMariaa-ka-bhọn-aa-ka-βɔń-a3s-PST-find-FVu=mw-aana?u=mw-áana?D=C1-child‘Did Maria find a/the child?’[Context: It is a sunny day and lots of kids’ noises are coming from theplayground. Makuru went to the playground. Mom is wondering why heis not back for lunch].(298) a. hamwehamwemaybea=bha-anaa=βa-anáD=C2-childrenm=bha-ku-bharaan-am=bá-ku-βáraan-aFOC=SA2-IMPFV-play-FVna-wẹna-wɛ́with-3sg‘Maybe the/some children are playing with him’.b. hamwehamwemayben=a-a-ku-bharaan-an=á-á-ku-βáraan-aFOC=SA1-IMPF-play-FVnanawitha=bha-anaa=βa-anáD=C2-children‘Maybe he is playing with (the) children’.(299) [Context: Maria has a sick child and only elderly people know thetraditional cure of the disease. There is a specific elder who knowsthe medicine for the disease. Maria says she would be happy if thatelder showed up] [Adapted from Matthewson 1999: 90].o=mu-gharukao=mu-ɣárukaD=C1-eldera-ngi-i-ch-irea-nga-i-tʃ-íreSA3-COND-REFL-come-PFVMariaMariaMarian=a-ŋga-chọmiir-un=a-ŋga-tʃɔmer-uSAM=PST-COND-heal-PFV‘If an/the elder came Maria would be happy.’201I adopt the common analysis for these cases that the DP containing theovert D is interpreted semantically as scoping above the non-factual oper-ator (see Matthewson 1998, 1999; Giannakidou 1998; Gambarage 2012;and others).The many parallels between the Nata augment/D system and theSt’át’imcets assertion-of-existence D system are based on the fact that thecore semantics of both systems is a speaker-oriented commitment to exis-tence of a referent. While this seems to be the case, the two systems alsoseem to slice their semantic pie slightly differently. I discuss these differ-ences below.4.6 The Locus of Parametric VariationThe upshot of the semantic/pragmatic factors that form the basis of theparametric variation for St’át’imcets and Nata is given below.Table 4.3: Locus of variation between St’át’imcets and NataLocus of variation among Ds St’át’imcets NataSpeaker’s personal knowledge is required 3 7Speaker conveys existence by surmising 7 3Speaker conveys existence of non-materialized entities 7 3Speaker conveys existence in possible worlds (7) 3Deictic features in D 3 7To understand how Nata belief-of-existence Ds and St’át’imcetsassertion-of-existence Ds slice up their semantic pie differently in termsof the points in Table 4.3, I consider the generalization in (300):(300) The relation between Nata and St’át’imcets DsAssertion of existence asymmetrically entails belief of existence.Nata overt Ds are more permissive than St’át’imcets Ds ending with-a. The entailment relation in (300) predicts that in all the contexts202where a St’át’imcets assertion-of-existence D X...-a is used, the Nata belief-of-existence D will also be used, but the reverse implication does nothold. For example, the Nata D can be used in surmising contexts but theSt’át’imcets assertion-of-existence Ds cannot. Another way to say this isthat the existence-related Ds are in a subset-superset relationship (i.e., theSt’át’imcets system is a subset of the Nata system). I discuss the relevantcontexts below.4.6.1 Requirement for speaker’s personal knowledgeThe requirement for speaker’s personal knowledge is one of the conditionsfor asserting existence which seems to be missing in Nata. Matthewson(1998) shows that in St’át’imcets, if the speaker did not personally witnessthe event or does not have personal knowledge of a referent, the speakerwill only use the non-assertion-of existence ku. As Matthewson demon-strates, in (301), the speaker was only told by someone that a chief cameto visit her (e.g., she was sick in bed at the time and did not witness anychief visiting), hence the use of the non-assertion-of existence ku. Notethe usage of the quotative particle ku7, which is an evidential marker in-dicating speaker direct knowledge is missing (i.e., s/he did not witness theevent) (Matthewson et al. 2007; Matthewson 1998; 2011):(301) The determiner ku [St’át’imcets; Matthewson 1998: 179]’ƛakt’akgokʷuʔku7QUOTʔá ’cx-ən-č-ašáts’x-en-ts-assee-TR-1SG.OB-3ERG[kʷu[kuNON.EXIS.DETkʷúkʷupiʔ]kukwpi7]chiefʔiiwhen.PASTnátxʷ-ašnatcw-asday-3SG.CONJ‘A chief came to see me yesterday (I was told).’However, when a Stʼátʼimcets speaker is a witness, i.e., has full knowl-edge of the individual, the assertion-of-existence enclitic -a is used. Herethe QUOT particle disappears because the speaker has witnessed the event:203(302) X...a determiners [St’át’imcets; Matthewson 1998: 179]’ƛakt’akgoʔá ’cx-ən-č-ašáts’x-en-ts-assee-TR-1SG.OB-3ERG[kʷu[kuDETkʷúkʷupiʔ-a]kukwpi7-a]chief-EXISʔiiwhen.PASTnátxʷ-ašnatcw-asday-3SG.CONJ‘A chief came to see me yesterday (I saw him).’In Nata, speaker’s personal knowledge/first hand evidence is not a re-quirement for belief of existence. As I showed in Chapter 3, the Nata lexicondoes not have evidential particles/clitics like St’át’imcets, but the languagedoes have verbs that can function as evidentials–for instance, the quotativeverb nyoogwa ‘I heard’ as in (303). Unlike the St’át’imcets example (301),Nata belief-of-existence Ds will be used even in contexts where the speakerwas only told about a chief’s visit and never met him, (303). Recall fromChapter 3 that the quotative verb nyoogwa can in fact license the polarity Din contexts where the speaker does not trust the source of information/doesnot believe that a chief exists, which will parallel the St’át’imcets D ku.(303) ny-oghw-aɲ-oɣhw-a1sg-hear-FVo=mu-tẹmio=mu-tɛḿiD=C1-chiefa-ka-chaa-kaa-tʃáSA1-PST-comeku-n-dọr-aku-ne-rɔŕ-aC15-1sg-see-FVichọitʃɔ́yesterday‘I heard a chief came to see me yesterday.’The use or non-use of the overt Ds does not depend on speaker’s personalknowledge. Whether the speaker lacks personal knowledge as in (303), orthe speaker has personal knowledge as in (304), the overt D is used.204(304) o=mu-tẹmio=mu-tɛḿiD=C1-chiefa-ka-chaa-kaa-tʃáSA1-PST-comeku-n-dọr-aku-ne-rɔŕ-aINFT-1O.sg-see-FVichọitʃɔ́yesterday‘A chief came to see me yesterday (I saw him).’We see that the St’át’imcets assertion-of-existence Ds require speaker’spersonal knowledge (i.e., the speaker to be a witness) in order to be used;however, we see that the Nata D can be both used in similar contexts andbeyond, which indicates that assertion of existence entails belief of exis-tence. In Nata speakers commit to the belief of existence when they havea reason to do so; but obviously speaker’s personal knowledge is not a re-quirement for a belief of existence system.4.6.2 Surmising contextsMatthewson (1998) illustrates that in surmising contexts in St’át’imcets––contexts in which the speaker supposes that entities exist without havingevidence to confirm their existence––the assertion of existence Ds cannotbe used. If the speaker did not witness the event, extra morphologicalmarking by special particles/clitics is required to indicate that the speakerlacks personal knowledge of the event. In (305), for instance, the parti-cle k’a ‘surmise’ (also analyzed and glossed as an inferential evidential inMatthewson et al. 2007) is a non-factual operator which licenses the non-assertion-of-existence D:(305) No speaker knowledge [St’át’imcets; Matthewson 1998: 160/2]a. šámaʔsáma7white.personk’ak’asurmise[kʷu[ku[NON.EXIS.DETšqwal’-ən-táli]sqwal’-en-táli]tell-TR-ERG.EXTR]‘It must have been a white man who told her.’205b. qá ’m’t-š-ašqam’t-s-ashit-CAUS-3ERGk’ak’asurmise[kʷu[ku[NON.EXIS.DETʔúxʷalmixʷ]...ucwalmicw]...person]...‘A person might have been hit...’If no particle or non-factual operator is used, the default interpretationis that the speaker has personal knowledge of the event, and therefore of theindividual involved in the event, hence the assertion-of-existence D mustbe used:(306) Speaker knowledge [St’át’imcets, Matthewson 1998: 160-1]a. túp-u ’n-aštup-un’-aspunch-TR-3ERGš-Johns-JohnNOM-John[ti[ti[DETplíšmən-a]plismen-a]policeman-EXIS]‘John hit a policeman.’(Speaker witnessed the event [so knows the individual]).b. *túp-un-aš*túp-un’-aspunch-TR-3ERGš-Johns-JohnNOM-John[kʷu[ku[NON.EXIS.DETplíšmən]plísmen]policeman]‘John hit a policeman.’(= I saw John hit a policeman whom I have never seen).Furthermore, in cases where the speaker believes in the existence ofan entity due to cultural assumptions, but has not directly witnessed theentity, the non-assertion-of-existence D is used5. Note that the future tensehere licenses the polarity D:(307) [Context: Suppose that there is a belief in this community that ifyou see a trail of ants, you’ll eat meat tonight. You see a trail of5. Thanks to Henry Davis for eliciting these data for me and thanks to Lisa Matthewson forhelping to gloss them.206ants]:#ts’aqw-an’-émeat-TR-1PL.ERGkelhFUTta ts’í7-aDET meat-EXISlhkúnsatodayku-sgapNON.EXIS.DET-evening‘We will eat meat tonight.’Consultant: corrected ta...a to ku [St’át’imcets]In this example, the speaker asks the elicitor to switch the assertion ofexistence D ta...a to the non-assertion of existence D ku, indicating that thespeaker does not agree to assert the existence of meat. This shows furtherthat in St’át’imcets, when an assertion-of-existence D is used, existence isnot merely believed, but rather asserted based on the speaker’s personalevidence of the referent.Nata patterns differently in two ways. The first is that Nata does notmark speaker evidence morphologically; the second is that there is no re-quirement for assertion of existence, i.e., the speaker is not required to haveknowledge of the referent to use the overt D. Since assertion of existenceentails belief of existence, we predict that the Nata belief-of-existence Dwill be used both in contexts where existence of a referent is asserted aswell as in surmising contexts, and this is correct. In (308), the speakerhas personal evidence about the existence of the individual, and belief ofexistence follows from assertion of existence:(308) a. YohanaYohanaJohna-gha-tẹm-aa-ɣa-tɛm-aSA1-PST-hit-FVo=moo-sirikareo=moo-sirikaréD=C1-policeman‘John hit a policeman.’(Speaker witnessed the individual).b. *Yohana*YohanaJohna-gha-tẹm-aa-ɣa-tɛm-aSA1-PST-hit-FVmoo-sirikaremoo-sirikaréC1-policemanIntended: ‘John hit a policeman.’207Belief-of-existence Ds can also be used in contexts where the speaker did notwitness the referent but is only surmising that an entity exists. That is, thespeaker believes in the existence of an entity due to cultural assumptions,as shown in (309)-(311).(309) [There is a sun-shower outside. B says...]a. a=n-gweáa=ŋ-gweD=C9-leopardye-ku-bhaje-ku-βáSA9-PROG-bee-rii-bhor-ae-rii-βór-aC9-SMLT-give-birth-FV‘A leopard will be giving birth.’b. * n-gwe* ŋ-gweC9-leopardye-ku-bhaje-ku-βáSA9-PROG-bee-rii-bhor-ae-rii-βór-aC9-SMLT-give-birth-FVIntended: ‘A leopard will be giving birth.’(310) [Context: Z is chewing and she bites her lip. She says:]a. o=mo-too=móo-toD=C1-personn-aa-ku-n-gaamb-an-a-ku-ŋ-gaamb-aSAM-SA1-PROG-1SG-talkbhwahẹẹnẹβwahɛɛ́nɛwell/good‘Some person is speaking well of me.’b. * mo-to* móo-toC1-personn-aa-ku-n-gaamb-an-a-ku-ŋ-gaamb-aSAM-SA1-PROG-1SG-talkbhwahẹẹnẹβwahɛɛ́nɛwell/goodIntended: ‘Some person is speaking well of me.’(311) [Context: It’s 3p.m. B sees a trail of ants carrying their food. Bsays:]208a. mu-ghorọọbhamu-ɣorọọβaC3-eveningn-to-ko-reer-an-to-ko-réer-aSAM-2sg-FUT-eat-FVa=n.yamaa=ɲ.ámaD=C9-beef‘We will have beef (for dinner) tonight.’b. *mu-ghorọọbha*mu-ɣorọọβaC3-eveningn-to-ko-reer-an-to-ko-reer-aSAM-2sg-FUT-eat-FVn.yamaɲ.amaC9-beefIntended: ‘We will have beef (for dinner) tonight.’These data show clearly that Nata speakers are not asserting the existenceof these referents; rather, they are only surmising, anchoring their beliefsof existence to cultural knowledge. In (311), for instance, the speaker mayhave no idea where beef will come from, but still the belief-of-existence Dmust be used to convey the belief in the existence of beef. Note furtherthat while future tense can license the polarity D in St’át’imcets, as we sawin example (308) above and as we shall see also below, strikingly in Nata,even though there is a licensor, the speaker can only use the overt D becausethey believe there will be beef.Non-belief-of-existence Ds can be used in the opposite contexts basedon cultural assumptions. Suppose speaker B sees a trail of ants carrying nofood or the speaker gets home and finds out that there is no beef for dinner.In such contexts, B has to switch the D and use the non-belief-of-existenceD with the non-factual operator, which is indicative of failure of belief ofexistence:(312) a. mu-ghorọọbhamu-ɣɔrɔɔβaC3-eveningtu-taa-ko-reer-atu-ta-ko-reer-a2sg-NEG-FUT-eat-FVn.yamaɲ.amaC9-beef‘Tonight, we will not have any beef (for dinner).’209b. #mu-ghorọọbha#mu-ɣɔrɔɔβaC3-eveningtu-taa-ko-reer-atu-ta-ko-reer-a2sg-NEG-FUT-eat-FVa=n.yamaa=ɲ.amaa=C9-beefIntended: ‘Tonight, we will not have any beef (for dinner).’Note that if the speaker has reasons to commit to the existence of beef,e.g., s/he saw mom bringing some beef home, (312b) would be good ina context where the speaker saw that there is beef, but asserts that theywill not eat it. The use of the belief-of-existence D for cultural assumptionsin Nata provides further support for augments as belief-of-existence Ds.That is, the speaker only surmises, based on their belief system/culturalknowledge, that a referent exists in some possible world, and yet the belief-of-existence D is used.4.6.3 Non-materialized referentsMatthewson gives clear evidence that St’át’imcets speakers fail to assertexistence in utterances containing referents that do not exist yet. Matthew-son shows that the non-assertion-of-existence D ku must be used in thesecontexts:(313) Non-assertion-of-existence interpretation[Stʼátʼimcets, Matthewson 1998: 57]xʷú ’z-ɬkancuz’-lhkangoing.to-1SG.SUBməlyí-šmelyí-smarry-CAUS[kʷu[ku[NON.EXIS.DETxʷu ’zcuz’going.tokʷúkʷpiʔkúkwpi7chiefɬakʷuʔláku7DEICFountain.]Fountain]Fountain]‘I will marry the next chief of Fountain.’ (whoever it is)In contrast, the belief-of-existence Ds in Nata are freely used in future pos-sibilities where referents have not manifested yet, (314):210(314) [Context: The chief is old; we don’t know who will be the next chief]a. N=ne-gho-kwir-uN=ne-ɣo-kwir-uSAM-1SG-FUT-marry-PASSnanawitho=mu-temi u-nọ a-kuu-ch-ao=mu-temi u-nɔ a-kuu-tʃ-aD=C1-chief‘I will marry the next chief.’b. *N=ne-gho-kwir-u*N=ne-ɣo-kwir-uSAM-1SG-FUT-marry-PASSnanawithmu-temi u-nọ a-kuu-ch-amu-temi u-nɔ a-kuu-tʃ-aC1-chief‘I will marry the next chief.’(315) Question: What will you donate for the new school?a. N=ne-ghu-kọr-aN=ne-ɣu-kɔŕ-aSAM-1SG-FUT-make-FVe=ghe-tuumbee=ɣe-túumbeD=C7-three.legged.chair‘I will make a three-legged chair.’b. *N=ne-ɣu-kɔr-a*N=ne-ɣu-kɔŕ-aSAM-1sg-FUT-make-FVɣe-tuumbeɣe-túumbeC7-three.legged.chair‘I will make a three-legged chair.’In (314a) the coming chief is a future possibility and speakers believethat when that time comes they will marry the chief. Similarly, in (315a)the envisioned chair is a future possibility and they believe that it will man-ifest at a future time t. Obviously, in these cases the speaker is not makingan assertion of existence. Since the speaker has not witnessed the nextchief/chair, it makes sense that Stʼátʼimcets assertion-of-existence Ds can-not be used here; this supports the argument that the assertion-of-existencedenotation entails the belief-of-existence denotation.2114.6.4 Possible worlds: attitude verbsHeim (1992), following Hintikka (1969), proposes that attitude verbs quan-tify over worlds w’ that are doxastically accessible to a belief holder:(316) Accessibility assignmentWorld w′ is doxastically accessible for person x to world w iff w′ iscompatible with the beliefs that x holds in w [Heim 1992: 187].In the same vein, Giannakidou and Mari (2018) also demonstrate thatattitude verbs can be analyzed based on the speaker’s epistemic state inrelation to a set of worlds compatible with the speaker’s beliefs. A definitionof epistemic state is in order. In (317), M is an evaluation model describingthe belief states of individual anchors (i.e., attitude bearers).(317) Epistemic state of an individual anchor iAn epistemic state M(i) is a set of worlds associated with an individ-ual i representing worlds compatible with what i knows or believes[Giannakidou and Mari, 2018: 7].Clauses embedded under attitude verbs can introduce referents that ex-ist in other possible worlds. In Nata, belief-of-existence Ds may appear un-der attitude verbs where they convey the speaker’s commitment to a beliefof existence of a referent. While attitude verbs do not inherently licensethe polarity D in Nata, Matthewson (1998) illustrates that attitude verbstranslated as ‘want’ or ‘look for’ are non-factual operators which licensethe polarity D ku in Stʼátʼimcets6. Below the attitude verb xát’min’ ‘want’licenses ku.(318) Intensional verbs license ku[St’át’imcets; Matthewson 1998: 193/5]6. Matthewson (1998) discusses a subject-object asymmetry showing that subjects of tran-sitive attitude verbs cannot be licensed with ku in St’át’imcets, unless they are in a subordi-nate clause where the licensor can take the entire clause in its c-command. In Nata attitudeverbs do not license the polarity D, hence there is no difference between transitive subjectsand objects. See Chapter 3 for additional data.212a. xá ’ƛ-rni ’n-ɬkanxát’-mín’-lhkanhard-APPL-1SG.SUB[kʷu[ku[NON.EXIS.DET’cúqʷa ’z]ts’úqwaz’]fish]‘I want some fish.’ [transitive object]b. xá ’ƛ-rni ’n-ɬkanxát’-mín’-lhkanhard-APPL-1SG.SUBkʷ-škw-sDET-NMLZRʔá ’z-ən-ašáz’-en-asbyu-TR-3ERG[kʷu[kuNON.EXIS.DETšmúɬač]smúlhats]woman[kʷu[kuNON.EXIS.DETqmut]qmut]hat‘I want a woman to buy a hat.’ [transitive subject]Note that if the speaker has a fish or a woman in mind (i.e., has personalevidence about the existence of such things), assertion-of-existence Ds maybe used with xát’min’ ‘want’.Unlike St’át’imcets, which allows attitude verbs to license the polarityD ku in contexts where the speaker does not wish to assert existence, inNata DPs inside clauses embedded under attitude verbs must contain theovert D; it conveys the existence of a referent denoted by the noun phrase.(319) Intensional verbs do not license D∅ [Nata]a. Ni-kwend-aNi-kwend-a1SG-want-FVa=∅-swe [transitive object]a=∅-swéD=C9-fish‘I want some fish.’b. *Ni-kwend-a*Ni-kwend-a1SG-want-FV∅-swe∅-swéC9-fishIntended: ‘I want some fish.’213(320) a. Ni-kwend-aNi-kwend-a1SG-want-FVo=mu-kario=mu-kario=C1-womana-ghọr-ẹa-ghɔr-ɛSA1-buy-SUBVa=n-goohiiraa=n-goohiiráa=C9-hat‘I want a woman to buy a hat.’b. *Ni-kwend-a*Ni-kwend-a1SG-want-FVmu-karimu-kariC1-womana-ghọr-ẹa-ghɔr-ɛSA1-buy-SUBVn-goohiiran-goohiiráC9-hat‘I want a woman to buy a hat.’As with the other Nata speakers I worked with, my intuition is thatreferents used with attitude verbs, e.g., a=swe ‘a fish’ (319) or o=mukari‘a woman’ in (320), are believed to exist in the world of discourse. Thisis independently supported by comments made by the Nata consultantssaying that one can only say ‘I want X’ or ‘I am looking for X’ if one believesthat such things exist. If speakers do not wish to commit to the belief ofexistence of an entity they would use other syntactic devices such as placingthe covert D under an evidential verb or a conditional.7.7. Hotze Rullmann (p.c) and Florian Schwarz (p.c) on different occasions have asked if it ispossible to use the overt augment when referring to mythical referents as in Maria is lookingfor an ogre. The point is that here the speaker does not believe in the existence of an ogre.This question has two sides. First, it is a concern whether ogres exist in the actual world,and if not, then why would the speaker use the overt D? Speaker-oriented-existence Ds canbe used independently of the truth or falsity of the sentence (Givón 2018; Moltmann 2013and others). The speaker may choose to use the belief-of-existence D conveying existenceof ogres in the speaker’s world, e.g., in referring to ogres in Nata tales, which are crucialcharacters in Nata discourse. This relies on both discourse structures and cultural assump-tions. The second part of the question, is whether the speaker may actually use the verb“look for” with such entities. My intuition is that if I use the first person I cannot uttersuch a sentence. I can only use these intensional verbs with entities that I believe exist inthe universe of discourse, e.g., fish, mangoes etc. Nata speakers consistently avoid utteringsuch sentences in the first person. They can however, use some licensers e.g., evidentialverbs to embed such propositions where the polarity D can be used to render the non-beliefof existence of an ogre:(i) Nyoghw-aɲ-oɣw-a1sg-hear-FVM. n-a-ku-mọh-iM. n-a-ku-mɔh-iM. SAM-1sg-IMPFV-look.for-FVi-nanii-naniC9-ogre‘I hear (that) M. is looking for an ogre.’214I argue that the use of belief-of-existence Ds under attitude verbs inNata is consistent with the semantics of the Nata Ds. I claim that the dif-ferences between the St’át’imcets system and the Nata system cannot belocated in the semantics of intensional verbs, but rather the semantics ofDs themselves. In possible worlds, entities are believed to exist in Nata butare not asserted as in an assertion-of-existence system. This seems to bethe case given that St’át’imcets can in fact use the assertion-of-existence Dsin similar contexts if the speaker has personal evidence, e.g., s/he wants toeat some fish s/he bought yesterday (see Matthewson 1998). Conversely, ifsimilar verbs are negated in Nata, the D∅ is licensed without any problem.4.6.5 Deictic features in DExcept for Okanagan (Lyon 2011, 2013), Ds in Salish are known for en-coding deictic features in the sense of spatial-temporal restrictions (Demir-dache 1996; Demirdache and Matthewson 1995; Matthewson 1998; Gillon2009, 2006; and others). Matthewson (1998: 352) states explicitly that“determiners in Salish are always deictic (i.e. always locate the discoursereferent(s) in time and space).” One implication of deictic features in theSt’át’imcets D system is that the features clash with true generic read-ings. Matthewson (1998, 1999; 2001) notes for instance that generic read-ings for the statements in (321) are missing in the language. Thus, whenSt’át’imcets speakers utter a translation of (321a) for instance, they alwaysassert the existence of some bears they know.(321) Generic readingsa. Bears like honey.b. Owls hunt mice.Matthewson argues further that because of the deictic features of Ds,true generic readings are lacking with universal quantifiers. Matthewson(1998: 352), for instance, remarks that “there is no way of quantifying overa group which is not contextually specified”. She concludes that Stʼátʼimcets215lacks ‘real’ generics; as a result of this fact, quantified DPs such as all thewomen, (322) always involve the universal quantification reading over acontextually specified set8:(322) Quantification over a specific set[St’át’imcets, Matthewson 1998: 46]qʷəlá ’w-əmq’weláw’-empick.berries-INTR[tákəm[tákem[allʔiiPL.DETšyáqčʔ-a]syáqts7-a]woman-EXIS]‘All the women picked berries.’In this case, the quantifier has to range over a contextually definedset of women which is consistent with the deictic features of i...a. Alongwith other Salishanists (see Davis 2010; Gillon 2006; Jelinek 1995; Demir-dache 1996), Matthewson (1998: 352) submits that “the absence of genericuniversal quantifiers in Salish can only be predicted if it is independentlyderived that determiners in Salish are always deictic (i.e. always locate thediscourse referent(s) in time and space).”Nata differs from St’át’imcets in completely lacking spatial-temporaldistinctions in its D system. As I argued in Chapter 2, deictic features inNata are always fixed by the demonstratives, not Ds. One prediction basedon the lack of deictic Ds which locate referents in space and time in Natais that ‘real’ generics may be found. This prediction is correct. As (323)shows, such readings are fine in Nata:(323) Generics in Nata [Nata]8. Matthewson (1998) argues that the absence of the quantifiers ’every’ and ’most’ inSt’át’imcets is due to: (i) a syntactic requirement in the language that every quantifiermust take a full DP containing a D, rather than joining with an NP; (ii) the deictic featuresof the Ds. Nata allows all quantifiers to take a DP as their range, and since there is no deicticrestriction in the language, both the generic readings of plural DPs and of quantifiers like-ọse ‘every’ are available. I am not sure at this moment though if Nata has quantifiers like‘most’ and ‘some’, and if not what could have prevented them. This is an area for futureresearch.216a. u=mw=aanau=mw=aanáD=C1-childw-ọọsẹw-ɔɔsɛ ́C1-everyn-aa-segh-iren-áá-seeɣ-ireFOC-SA1-like-PFVku-bharaan-aku-βá-raan-aINF-play-FV‘Every child likes to play.’b. i=rj-oobhai=rj-ooβáD=C5-sunn-di-segh-iren-rí-seɣ-ireSAM-SA1-love-PFV[a=bhaa-to[a=βáa-to[D=C2-peoplebhọ-ọsẹ]βa-ɔsɛ]́C2-all]‘God loves all people.’c. [e=bhe-bhuse][e=βe-βúse][D=C8-monkey]m=bi-haa-r-im=bi-haa-r-íSAM=HAB-eat-FVe=bhi-tọọkẹe=βí-tɔɔkɛD=C8-banana‘Monkeys eat bananas.’Consistent with the entailment relation between Stʼátʼimcets and NataDs, we see that, in Nata, quantification may yield a pure generic reading asin (323) or may yield a contextually specified set as in (324)9. Note that theuniversal quantifier parallel to all is marked with the plural morphology PL-ọsẹ in Nata, while the universal quantifier parallel to every is marked withthe singular morphology SG-ọsẹ.(324) a. [u=mw-aana[u=mw-aaná[D=C1-childw-ọọsẹ]w-ɔɔsɛ]́C1-all]a-gha-sọh-aa-ɣa-sɔh́-aSA1-PST-enter-FVmo=o=nyumbamo=o=ɲumbáLOC18=D=house‘Every child entered the house.’b. [e=bhi-nyonyi[e=βi-ɲoɲi[D=C8-birdbhj-ọọsẹ]βj-ɔsɛ]́C8-all]m=bi-ike-erem=bi-ka-ireSAM=SA8-sit-FVmo=o=motemo=o=motéLOC18=D=tree‘All the birds are sitting in a tree.’9. Pure generic readings are found mainly with subjects, with the habitual marker, or withsome special meanings with objects, e.g., God loves all people.217The lack of deictic features in D also predicts that Nata belief-of-existence Ds can be used in free choice contexts. Osa-Gómez (2016) demon-strates that the Nata item -o(se)-ose ‘any’, which she analyzes as a domainwidener, corresponds to the English ‘any’ when used in positive sentences.(325) FC environmenta. u=mw-aanau=mw-aanáD=C1-childwọ(sẹ)-wọseuɔ(sɛ)-u-ɔsɛRED-C1-alln-a-gho-tiin-an-a-ɣo-tiin-aSAM-SA1-FUT-be.afraid-FV‘Any child will be afraid.’b. * mw-aana* mw-aanaC1-childwọ(sẹ)-wọseuɔ(sɛ)-u-ɔsɛRED-C1-alln-a-gho-tiin-an-a-gho-tiin-aSAM-SA1-FUT-be.afraid-FVIntended: ‘Any child will be afraid.’(326) FC environment [Osa-Gómez 2016: 02]a. Ghégh-aɣɛɣ́atake-FVo=mu-tẹrẹbhio=mu-tɛrɛβiD=C3-ladlewọ(sẹ)-wọsewɔ(sɛ)-wɔsɛRED-all‘Take any ladle.’b. *Ghégh-a*ɣɛɣ́atake-FVmu-tẹrẹbhimu-tɛrɛβiC3-ladlewọ(sẹ)-wọsewɔ(sɛ)-wɔsɛRED-allIntended: ‘Take any ladle.’Obviously, the DPs in these examples do not correspond to a non-belief-of-existence interpretation, given that the FCI denotes freedom of choicebetween existing entities, i.e., any entity picked will be fine with thespeaker10, (see Kadmon and Landman 1993; Giannakidou 2001; Kratzer10. Matthewson (p.c) asked about what happens if a Nata speaker wants to say somethinglike ‘Greet any 100-year-old person who comes by’ - where you don’t know if any exists.218and Shimoyama 2002; Chierchia 2006, 2013; Menéndez-Benito 2010; Osa-Gómez 2016). It is also clear that such DPs do not refer to any specificentity. These examples accord with the notion of belief of existence; i.e.,the use of the overt D in the DPs u=mwana and o=mu-tẹrẹbhi above con-veys that the speaker believes that such referents exist.While I have demonstrated here that Nata and St’át’imcets Ds differ ondeixis, i.e., Nata lacks spatial-temporal distinctions in its D system, I donot take the lack of deixis as a (direct) diagnostic for a belief-of-existenceD system. One case that supports this line of thinking is a system likeOkanagan (see Lyon 2011; 2013). In Okanagan, like in Nata, Ds do notencode deixis (i.e., locate the referent in space and time) hence true genericinterpretations are possible, (327). However, Okanagan Ds do not encodethe notion of existence (see Lyon 2011; 2013 for discussion)11.(327) Generic readings [Okanagan, Lyon 2013: 151]iʔDETkəkwápdogtáʔliʔveryxm̌ínk-slike-[CAUS].3SG.ERGiʔDETsnaxʷtrun(ANIMAL)‘Dogs really like to run.’In summary, I have illustrated that in both Nata and St’át’imcets, Dsencode a speaker-oriented distinction based on existence of a referent. InSt’át’imcets, the assertion-of-existence Ds X...a are used when the speakerTruly, in Nata such a clause will be embedded under a non-factual operator which willlicense the interpretation of a non-existent referent. Here the conditional operator aribha‘if’ may be used where it will license the polarity D in this context:(i) Aribhaaribhaifmo-kungumo-kúnguC1-old.ladywọ(sẹ)-w-ọsewɔ(sɛ)-w-ɔɔ́sɛRED-C1-allghekongoɣékoŋgotoothlessa-raa-ch-ẹ,a-ráa-tʃ-ɛ,SA1-PROG-come-SUBJVmu-kẹẹr-imú-kɛɛr-iOM1-greet-CAUSLit: If any toothless old lady (very old woman) comes, make her greeted’.Here the speaker does not commit to the belief that such a lady may come. If the speakerbelives that such an individual exists then they will switch and use the overt D.11. Is it possible then to have a language that encodes the notion of assertion-of-existencebut does not encode deixis in its D system? In Chapter 5, I show that Bemba is a case inpoint, hence deixis may be motivated on language-internal grounds.219has strong grounds to assert that a referent exists. In Nata, overt Ds areused when the speaker has some reason to believe that a referent exists. Inwhat follows I relate the Nata data to Matthewson’s (1999) choice functionanalysis in order to show how belief-of-existence Nata Ds would fare in anassertion-of-existence analysis.4.7 Choice function analysisWe saw already that belief-of-existence Ds and assertion-of-existence Dshave many parallels, but also have some obvious differences. The goalof this section is to employ Matthewson’s (1999) choice function analy-sis in order to show how the belief-of-existence Nata Ds compare withother speaker-oriented existence Ds theoretically. This well-known anal-ysis for assertion-of-existence Ds will provide a theoretical space not onlyfor showing how the two D systems are related but also for asking empiri-cal questions about the nature of variation between the two systems. Cru-cially, I will show that Matthewson’s (1999) choice function approach forSt’át’imcets Ds accounts for Nata cases where existence is asserted. How-ever, the analysis created for St’át’imcets will be too restrictive if applieddirectly to Nata. For instance, it will predict that overt/non-polarity Ds areallowed only when the speaker is willing to assert that there is a referent,which is not always the case. I will present the cases that are accounted forstraightforwardly by Matthewson’s (1999) analysis and offer some optionsfor accounting for some cases involving speaker’s belief of existence Ds,along the lines of Gambarage and Matthewson (2019). A complete formalaccount for the belief-of-existence Nata Ds is a goal for future research.4.7.1 Defining choice functionsIn the spirit of previous works on speaker-oriented existence Ds in Sal-ish (Matthewson 1999), I claim that Nata Ds require an analysis involvingchoice functions. A definition of the choice function is in order:220(328) Choice function definition:A function f is a choice function (CH(f)) if it applies to any non-empty set and yields a member of that set.[Reinhart 1997: 372].The evidence for needing choice functions comes from the ability of thechoice function variable to take obligatory widest scope over everythingelse. Thus, the DPs in question scope outside places where they should notbe able to, under a standard analysis of quantifiers and Quantifier Raising(cf. Reinhart 1997; May 1985 and others). Under a choice function analy-sis, DPs can appear inside islands but semantically they can scope outsidethem (see Matthewson 1999; Ebert 2019 for review).While there are many different approaches to choice functions in theliterature (see Ebert (2019) for a helpful summary), I utilize Matthewson’swidest-scope existential closure over a choice function variable in orderto derive the wide-scope effect of DPs used in contexts where existence isasserted. Matthewson (1999) argues that the choice function variable mustbe existentially closed at the highest level to account for the wide-scopeeffect of St’át’imcets indefinites. I discuss the Nata cases that are accountedfor by this approach, but also discuss the cases with DPs used in contextswhere existence is conveyed rather weakly (i.e., by mere beliefs), whichseem to require extra explanation (Gambarage and Matthewson 2019).In this thesis I do not examine quantificational strategies, however,where necessary, I do show that Nata quantifier phrases are amenable toa similar semantic treatment to those in St’át’imcets, hence are accountedfor by the current choice function approach. The one exception I discussis the quantifier SG-ọsẹ ‘every’, which allows both wide-scope and narrowscope readings. Following Matthewson (1999, 2001), my null hypothesesfor the treatment of Nata Ds with respect to the current analysis are that:(329) a. All overt/belief-of-existence Ds are obligatorily interpretedas choice function variables, existentially closed with widestscope.221b. All polarity/non-belief-of-existence Ds are not interpreted aschoice function variables.The discussion will centre around the issue of whether all overt Dsin Nata are interpreted via choice functions which take obligatory widestscope over everything else. I will start with contexts which allow asser-tion of existence, which confirm (329a), then turn to problematic casesin which existence is merely believed rather than asserted. I show that(329b) consistently allows DPs with the covert D to be interpreted as exis-tential quantifiers which take narrow scope under a non-factual operator,the treatment Matthewson (1999) assumes for ku DPs in St’át’imcets.4.7.2 Interpretation of Ds in assertion-of-existence contextsGiven that assertion of existence entails belief of existence, the analysisdeveloped for St’át’imcets Ds will be limited to Nata overt/non-polarity Dsappearing in contexts where a Nata speaker asserts that an entity exists,hence believes in the existence of a referent.Under the assertion-of-existence account, Matthewson illustrates thatthe X...a DPs are interpreted via choice functions which take obligatorywidest scope over everything else. The choice function analysis runs asfollows:(330) Existential Force [St’át’imcets, Matthewson (1999)]a. az’-ən-asbuy-TR-3ERG[ti[DETsts’úqwaz’-a]fish-DET]kw-sDET-NMLZRSophieSophie‘Sofie bought [a fish].’9f [CH(f) & [Sophie bought f(fish)]]=“There exists a fish which Sophie bought.”222b. cw7aosNEGkw-sDET-NMLZRáz’-en-asbuy-TR-3ERG[ti[DETsts’-úqwaz’-a]fish-DET]kw-sDET-NMLZRSophieSophie‘Sophie didn’t buy [a fish].’9f [CH(f) & ¬[Sophie bought (f(fish))]]=“There exists a fish which Sophie didn’t buy.”In the affirmative declarative sentence in (330a) and in the cases in-volving a non-factual operator such as negation, (330b), the non-polarityD is used and it is existentially closed at the highest level. Specifically,the logical form in (330a) says that there is a choice function f and Sophiebought the fish which f picks out from the set of fish; and the logical formin (330b) says that there is a choice function f and Sophie didn’t buy thefish chosen by f from the set of fish. Note that these truth conditions al-low Sophie to have bought one or more fish, which Matthewson shows isconsistent with the St’át’imcets facts.The non-assertion-of-existence D ku is not interpreted as a choice func-tion variable. Matthewson proposes that the ku in examples such as (331)receives a standard indefinite interpretation as an existential quantifierwhich scopes under the non-factual operator. Thus (331) is only true ifSophie bought no fish at all.(331) Existential Force [St’át’imcets, Matthewson (1999)]a. xʷʔazcw7aozNEGkʷ-škw-sDET-NMLZRʔáz’-ən-ašáz’-en-asbuy-TR-3ERG[kʷ-u[ku[DETšc’úqʷaz’]sts’-úqwaz’]fish]kw-škw-sDET-NMLZRSophieSophieSophie‘Sofie didn’t buy [a/any fish].’¬ 9x [fish (x) & Sophie bought x)]=“There does not exist a fish which Sophie bought”.223The non-factual operator has wide scope over the polarity D renderingthe meaning that it is not the case that there is a fish which Sophie bought.Given that assertion of existence entails belief of existence, Nataovert/non-polarity Ds may be used in all assertion of existence contexts.Thus, the analysis developed for St’át’imcets Ds is too restrictive for Nata;it works for the subset of contexts where a speaker asserts that an entityexists, hence believes in the existence of a referent. I start with declaratives.4.7.2.1 Interpretation of overt Ds in declarativesThe overt D introduces a choice function variable that is bound by the ex-istential quantifier to render the existential interpretation intended by thespeaker. Consider (332a) and the corresponding semantic form, symboli-cally and in prose in (332b).(332) a. MakuruMakurúMakurua-ka-ghoor-aa-ka-ɣór-aSA1-PST-buy-FVe=ghi-tabhoe=ɣí-taβoD=C7-book‘Makuru bought a book.’b. 9f [CH(f) & [Makuru bought f(book)]]There is a choice function f and Makuru bought the book whichf picks out from the set of books.In the current analysis, the speaker asserts that there is a choice functionwith that property, which is compatible with the speaker knowing a specificobject, but it does not semantically ensure it to be specific.4.7.2.2 Interpretation of overt Ds with/under negationWe saw that if the speaker intends an existential interpretation, the overtD must be used in negative contexts as in (333a). According to the currentanalysis, the choice function variable must be existentially closed at thehighest level, i.e., must take wide scope over negation. Therefore, (333a)can only be represented as in (333b).224(333) a. MakuruMakuruMakuruta-a-ghor-ireta-a-ɣor-iréNEG-PST-buy-PERFe=ghi-tabhoe=ɣí-taβoD=C7-book‘Makuru did not buy a/the book.’b. 9f [CH(f) & ¬[Makuru bought (f(book))]]There is a choice function f and it is not the case the Makurubought the book that f picks out from the set of books.When the speaker uses the polarity D as in (334a), the polarity D is notconsidered as a variable that ranges over a choice function. The polarityD is interpreted as an existential quantifier scoping under the non-factualoperator. This means the speaker does not commit to the existence of abook.(334) a. MakuruMakuruMakuruta-a-ghor-ireta-a-ɣor-iréNEG-PST-buy-PERFghi-tabhoɣí-taβoC7-book‘Makuru did not buy any book.’b. ¬ 9x [book (x) & Makuru bought x)]“There does not exist a book which Makuru bought.”(334a) will be true if there were two books and Makuru bought noneof them; it will still be true if there were no books at all. In interrogatives,modals and conditionals the polarity D is uniformly treated as not being avariable that ranges over choice functions.4.7.2.3 Interpretation of overt Ds in interrogativesIn both wh-questions and in polar questions, the choice function analysiscan also derive obligatory wide-scope readings for overt Ds, which accordswith the assertion/belief-of-existence interpretation:225(335) [Context: Speaker is inquiring about his friend’s child]a. angoangoQMariaMaríaMariaa-ka-bhọn-aa-ka-βɔń-a3s-PST-find-FVu=mw-aana?u=mu-áana?D=C1-child‘Did Maria find a/the child?’b. 9f [CH(f) & Q[Maria found (f(child))]]There is a choice function f and the speaker wants to knowwhether Maria found the child picked out by f.In the next section I show that this analysis accounts fine for DPs thatescape the scope of modals and conditionals, environments in which exis-tence is asserted/believed.4.7.2.4 Interpretation of overt Ds in modalsThe choice function analysis can also derive obligatory wide-scope read-ings for overt Ds under modals, as shown in (336).[Context: I saw a handicapped child at the playground. After a few minutesa researcher shows up asking if there is any child with a disability in theneighbourhood that she can interview. I tell her to go to the playground.The researcher leaves and promises to be back immediately if she doesn’tfind anyone there. I am wondering why she is not back yet:].(336) a. hamwehamwemayben=a-a-ku-ghamban-an=á-á-ku-ɣámban-aFOC=SA1-IMPF-play-FVnu=u=mw-anana=u=mw-anáwith=D=C1-child‘Maybe she is speaking with a child’.b. 9f [CH(f) & MOD[she is speaking with the (f(child))]].There is a choice function f and the researcher may be speakingwith that child who f picks out from a set of children.Here, MOD is the possibility modal. This analysis extends also to con-ditionals, as I show below.2264.7.2.5 Interpretation of overt Ds in conditionalsIn if-clauses we see that if the speaker intends an existential interpretation,the choice function variable is existentially closed at the highest level, asin (337b).[Context: A mother has a sick child and only elderly people know thetraditional cure for the disease. There is a specific elder who knows themedicine for the disease. A mother says she would be happy if that eldershowed up] [Adapted from Matthewson, 1999: 90](337) a. o=mu-gharukao=mu-ɣárukaD=C1-eldera-ngi-i-ch-irea-nga-i-tʃ-íreSA3-COND-REFL-come-PFVu=mw-aanau=mw-aanaD=C1-childn=a-nga-họọr-(ir)en=aa-ŋga-hɔɔr-ɛSAM=PST-COND-heal-PFV‘If an elder came the child would be healed.’b. 9f [CH(f) & 9g CH(g) & [come (f(elder)) → be-healed(g(child))]]There is a choice function f and a choice function g and f picksout an elder from a set of elders, and if that elder comes, thechild picked by g from the set of children would be healed.While I conclude that all DPs escaping the scope of non-factual oper-ators can be accounted for by wide-scope existential closure over choicefunctions, I show that this approach is too restrictive and does not give usthe intended results when it comes to DPs used in case of speakers’ beliefs.4.7.3 Interpretation of overt Ds in belief-of-existencecontexts: surmising, possible worlds...If we apply the analysis created for St’át’imcets directly to Nata it will betoo restrictive as it will predict that overt/non-polarity Ds are allowed only227when the speaker is willing to assert that there is a referent. However,belief-of-existence Ds need not be used in assertion-of-existence contexts.Consider, for instance, overt Ds used in surmising contexts, (338), or torefer to non-materialized referents, (339) that we saw in §4.6.2. With thesedata the speaker is saying that s/he believes an entity exists but does notassert the existence of a referent.(338) [Context: It’s 3 pm and B sees a trail of ants carrying their food.]a. mu-ghorọọbhamu-ɣorọọβaC3-eveningn-to-ko-reer-an-to-ko-réer-aSAM-2sg-FUT-eat-FVa=n.yamaa=ɲ.ámaD=C9-beef‘We will have beef (for dinner) tonight.’b. *mu-ghorọọbha*mu-ɣorọọβaC3-eveningn-to-ko-reer-a n.yaman-to-ko-réer-aSAM-2sg-FUT-eat-FVɲ.ámaC9-beef‘We will have beef (for dinner) tonight.’(339) [Context: I haven’t seen one yet, but I believe I will buy one to-day...]a. Ni-kwend-aNi-kwend-a1SG-want-FVa=∅-swe’a=∅-swéa=C9-fish‘I want some fish.’b. *Ni-kwend-a*Ni-kwend-a1SG-want-FV∅-swe’∅-swéC9-fish‘I want some fish.’While Nata does not allow future tenses or attitude verbs to licensethe polarity D, St’át’imcets has the option of having the non-assertion-of-228existence ku interpreted under ‘will’ or ’want’, which results in a differentmeaning from Nata Ds (refer to (340) below and to §4.6).(340) Intensional verb licenses ku[St’át’imcets, Matthewson 1998: 193]xá ’ƛ-mi ’n-ɬkanxát’-mín’-lhkanhard-APPL-1SG.SUB[kʷu[ku[NON.EXIS.DET’cúqʷa ’z]ts’úqwaz’]fish]‘I want some fish.’In St’át’imcets, the speaker is not committing themselves to the exis-tence of a referent, hence the D is not a widest-scope choice function whichpicks out an entity which is a fish. In Nata the overt D is used in similarcontexts, indicating that the speaker is committed to the belief that there isan object, but is not asserting the existence of such a referent. This is truefor DPs used with attitude verbs and in non-materialized referent contexts.While a unified formal account for belief-of-existence Ds lies in futureresearch, some initial observations on how such an analysis can be devel-oped have been put forth by Gambarage and Matthewson (2019).The first alternative Gambarage and Matthewson give is to incorporate afull belief-semantics into sentences containing overt Ds. This would meanthat any time there is an overt D, it literally adds to the meaning of thesentence an ‘I believe that’ scoping over everything else. It’s not a veryplausible option, but it might work.The second alternative would be to tie the differences to independentdifferences in evidential systems. We saw that at a propositional level, Natalacks obligatory evidential marking. We also know that some languagesencode evidential information directly on determiners (e.g., Nivaclé, Gui-tiérrez and Matthewson 2012). We could therefore pursue the idea thatSt’át’imcets X…a Ds convey direct evidential semantics (the speaker musthave personally witnessed the referent), while Nata overt Ds do not.The third possibility is to consider variation across languages in the fe-licity conditions for assertions on cultural assumptions. Specifically, we229could propose that Nata speakers are freer than St’át’imcets speakers intheir willingness to assert. Thus, for a Nata speaker but not for a St’át’imcetsspeaker, a cultural belief is sufficient for assertion. This will preserveMatthewson’s (1999) choice function analysis but incorporate speaker’spersonal knowledge and cultural assumptions as assertions with differentstrengths. In addition, the cases under future tense and intensional verbslike ‘want’ will have to have different ‘assertion standards’.4.7.4 Interpretation of overt Ds in quantifier phrasesIn this section, I present some data involving universal quantifiers toshow that the use of the overt D is consistent with the assertion/belief-of-existence interpretation. Matthewson (1999) argues that a subset ofSt’át’imcets Ds, including all Ds which combine with quantifiers, necessar-ily introduce variables over choice functions. While I agree with Matthew-son (1999) that existential closure over the choice function is needed toderive the denotation of existence Ds for all DPs involving assertion of ex-istence, I show that some Nata data involving the quantifier SG-ọsẹ ‘every’seem to allow both wide scope and narrow scope readings, which is a bitpuzzling for a wide scope existential closure interpretation. The Nata datawith SG-ọsẹ ‘every’ seem to suggest that a subset of quantifier data mayneed a separate explanation as I show below.4.7.4.1 Universal quantifiersTypically quantifiers in Stʼátʼimcets and Nata co-occur with theassertion/belief-of-existence Ds. Matthewson (1998, 1999) illustrates thatnarrow scope readings for assertion-of-existence DPs are unavailable withthe quantifiers in Stʼátʼimcets. For instance, one of the examples involvesthe quantifier tákem ‘all’ which ranges over a contextually salient set givenby the DP it combines with:230(341) Object DP cannot scope under a quantifier[Matthewson 1999: 96].wa7PROGxwey-s-twítaslove-CAUS-3PL.ERG[ta[DETsmúlhats-a]woman-DET][tákem[alliDET.PLsqáyqeycw-a]man(PL)-DET]‘All (the) men love a woman.’Rejected in context: Each man loves a different woman.[Consultant’s comment: “There’s just one lady. Can’t mean a different oneeach. It sounds like you’re talking about that one lady.”]As Matthewson argues, the obligatory wide scope reading of the object DPis because the object DP contains an assertion-of-existence D, whose choicefunction is existentially closed with widest scope: there is a CH(f) such thatall the men love the woman picked out by f from the set of women.Nata appears to have exactly the same restriction with the universalquantifier PL-ọsẹ ‘all.’ As (342a) shows, the subject wide-scope reading isunavailable with PL-ọsẹ ‘all.’ Employing Matthewson’s (1999) existentialclosure at the highest level correctly derives the object wide-scope readingin such cases, (342b).(342) a. [e=bhi-nyonyi bhj-ọọsẹ][e=βi-ɲoɲi βj-ɔsɛ]́[D=C8-bird C8-all]m=bi-ike-erem=bi-ka-ireSAM=SA8-sit-FVmo=o=motemo=o=motéLOC18=D=tree3‘All the birds are sitting in a tree.’7‘All the birds are sitting in different trees.’b. 9f [CH(f) & 8x [bird(x) → x is sitting in f(tree)]]c. ̸=8x [bird(x) → 9f [CH(f) & x is sitting in f(tree)]]There is no distributive reading of the universally quantified subject DPsuch that the choice of trees varies with birds. This is why (342c) is un-acceptable. The only reading available is that the choice function f, which231is existentially closed off at the highest level, picks out a tree that a set ofbirds are sitting in.Numeral quantifiers also cannot scope under the subject quantifiedphrase/PL-ọsẹ ‘all’ DPs, (343):(343) a. [a=bha-subhe[a=βa-suβe[D=C2-menbhọ-ọsẹ]βa-ɔɔsɛ]́C2-all]bha-ka-ras-aβa-ka=ras-aSA2-PST-shoot-FV[e=bhe-weere[e=βe-weere[D=C8-wildebeestbhi=tato]βi=tato]C8-three]3‘All (the) men shot three wildebeests.’b. 9f [CH(f) & 8x [men(x) → x shot f(three-wildebeests)]]Here the choice function variable applies to the predicate three wilde-beests denoting the set of plural entities with three singular wildebeests.The choice function D picks out the three wildebeests shot collectively byall the men12. Note that there could be more than three wildebeests buthere only 3 were shot.So far I have shown that the above quantifier data support Matthewson’swide-scope existential closure over choice function analysis. In addition, Ishowed above that all DPs escaping the scope of non-factual operators canbe accounted for by wide-scope existential closure over choice functions.I turn to show that this approach is restrictive and does not give us theintended results when it comes to DPs used in cases of speakers’ beliefs.Below I will also discuss some problems of the current analysis for datawith the Nata version of the quantifier ‘every.’4.7.4.2 The problem with the SG-ọsẹ quantifierUnlike the St’át’imcets D system (Matthewson 1998; 1999; 2001; Davis2010), Nata does not rule out narrow scope existentially quantified choice12. It is possible to have another reading that each man shot one wildebeest, and collec-tively they shot three in total (see Davis 2010 for similar readings in St’át’imcets).232function variables. Recall that in Stʼátʼimcets there is no universal quanti-fier equivalent to ‘every’, unlike in English and Nata13.The Nata universal quantifier SG-ọsẹ ‘every’ is a distributive quantifierwhich differs from PL-ọsẹ ‘all’ in (343), which cannot scope over a DP. In(1), we see that SG-ọsẹ ‘every’ has two readings, both of which yield anon-empty set interpretation consistent with the current analysis of belief-of-existence Ds. The wide-scope existential analysis demonstrated here cor-rectly derives a wide-scope reading of the choice function variable, (344b),but not the subject distributive reading, (344c):(344) a. u=mw-aanau=mw-aanáD=C1-childw-ọọsẹw-ɔɔsɛ ́C1-everya-ka-ria-ka-ríSA1-PST-eata=swea=sweD=C9.fish‘Every child ate a fish.’b. 9f [CH(f) & 8x [child (x) → x ate f(fish)]]There is a choice function f that picks out a fish such that everychild ate that fish (wide-scope for the object).c. 8x [CHILD(x) → 9f [CH(f) & x ate f(fish)]]For every child x, there is a potentially different choice functionf such that x ate the fish f picks out from the set of fish (narrow-scope for the object).The overt D in a=swe ‘a fish’ introduces a choice function variable thatis existentially bound at the highest level when the speaker means that all13. Matthewson (p.c) believes that Nata does not have a real ‘every’ quantifier. Her ob-servation may be right given that the ‘every’ quantifier is phasing out in Nata as in mostcases speakers replace it with the Swahili quantifier kila, pronounced as kira in Nata, whichreplaces the augment (i.e., is a quantificational D).(i) kirakiraevery(*u=)mw-aana(*u=)mw-aaná(D=)C1-childa-ka-ria-ka-ríSA1-PST-eata=swea=sweD=C9.fish‘Every child ate a fish.’When the Nata version is used the quantifier range must be a DP but when the Swahiliversion is used the quantifier range is an NP. The Swahili QP is used in all object positionswhile the Nata version is banned there. Refer to Chapter 2 for further discussion.233the children ate the fish picked by f. The same choice function D scopeslow when every child ate a potentially different fish picked by f. We haveseen consistently that nowhere else are narrow scope readings allowed. Thequestion then is how do we rule out the narrow scope reading of belief-of-existence Ds everywhere else but allow it here? The current tool does notprovide an answer for this question.One potential alternative–which however will not work–would be to trythe type of free choice function variable analysis proposed in Matthewson(2000, 2001), which is utilized in domain restriction contexts such as quan-tification over a specific group; or Kratzer’s (1998) choice function whichpicks out specific referents. For both Kratzer and Matthewson’s later analy-sis, f is not existentially closed and its value is supplied by the context. Forinstance, in (345b) the speaker is referring to a specific book.(345) a. e=ghi-tabhoe=ɣí-taβoD=C7-bookghi-ka-gw-aɣi-ka-ɣw-áC7-PST-FV‘A book fell.’b. fell (f(book))The book that is chosen from a set of books by the contextuallysalient choice function f fell.In (345b), f is a variable over choice functions/book choices, and ityields a contextually defined member of the NP set it applies to. Kratzer’sfree variable over f mimics wide scope (‘pseudo-scope’), therefore a freevariable analysis of choice functions will not be a solution because it doesnot allow narrow scope interpretations in the usual way (i.e., unless thereare bound variables in the noun phrase). Note also that in Kratzer’s accountall choice function variables are specific, which is not the case in Nata.Another alternative would be to keep the existential closure over choicefunctions but adopt an analysis in which choice function variables can beexistentially closed at other logical levels. Reinhart (1997) illustrates that,234in English, a sentence such as (346) can have both the wide scope readingand the quantificational/narrow scope reading involving a choice function:(346) Every lady read some book.a. 9f [CH(f) & 8z [lady (z) → z read f(book)]](Reinhart 1997: 372)b. 8z [lady (z) → 9f [CH(f) & z read f(book)]](Matthewson 1999: 83, presenting Reinhart’s analysis)The respective prose translations for each are given below:(347) a. There is some choice function f, such that every lady read thebook which f picks out from the set of books.b. For every lady z, there is a (potentially different) choice func-tion f such that z read the book which f picks out.In (346a), the indefinite D some introduces a choice function variable whichis existentially closed off at the highest level (wide scope). With the wide-scope reading, f picks out one book that every lady read. In (346b), thechoice function variable is existentially closed with narrow scope, in whichcase each lady reads a potentially different book. While this will work forthe SG-ọsẹ ’every’ cases, it will incorrectly rule in narrow scope in the casesabove where we saw it isn’t allowed. The exact analysis that would workfor the Nata cases like (344) remains an area for future research.4.7.4.3 Simple genericsDPs used in generic contexts take belief-of-existence Ds. It is not obvioushow to implement the wide-scope existential closure over choice functionsfor these cases.(348) Generics and free choice contexts [Nata]235a. e=bhe-bhusee=βe-βúseD=C8-monkeym=bi-haa-r-ím=bi-haa-r-iSAM=HAB-climb-FVe=bhi-tọọkẹe=βi-tɔɔkɛD=C8-banana‘Monkeys eat bananas.’To account for these cases one may need some operator, possibly GEN.However, I will leave this case and the quantifier case discussed above forfuture research.4.8 ConclusionIn this chapter, I have presented empirical evidence to show that D choicein Nata is forced by the speaker’s belief of existence. I have claimed thatlanguages can encode existence differently; nevertheless, the different typesof existence Ds form a class cross-linguistically. The similarities betweenNata and Stʼátʼimcets – two languages coming from two unrelated families –suggest that speaker-oriented existence is robustly available as a determinerdistinction.I have claimed that the lack of definiteness in languages with existenceDs follows from these systems being speaker-oriented. A further conse-quence of these findings is that definiteness should not be regarded asa ‘norm’ (Matthewson 1998, 1999; Gillon 2006; Gambarage 2012; Lyon2013; Van de Velde 2019 and many others).In logical terms, I have compared Nata and Stʼátʼimcets where I arguedthat Matthewson’s wide-scope choice function analysis can derive the NataDPs used in contexts where existence is asserted. However, data with SG-ọsẹ ‘every’, generics and belief-of-existence Ds in contexts such as surmisingsituations presented puzzles that can only be resolved in future research.While speaker-oriented existence Ds in Nata and Stʼátʼimcets are treatedhere as forming a class typologically, the similarities and differences in theapplication of choice function between the two languages are listed below:236Semantic differences Stʼátʼimcets Ds Nata DsCH(f) applies in specific contexts 3 3CH(f) applies in non-specific contexts 3 3CH(f) applies in familiar contexts 3 3CH(f) applies in novel contexts 3 3CH(f) existentially closed at the highest level 3 (3)CH(f) optional under intensional verbs 3 7CH(f) existentially closed with low scope 7 (3)CH(f) applies with non-materialized referents 7 ?CH(f) applies in surmising contexts 7 ?CH(f) applies in generics — ?The different logical possibilities for choice functions have been pre-sented here as introducing a parametric variation in these languages. Iconclude that assertion of existence always entails belief of existence. Inthe next chapter I argue that a D distinction based on the notion of existenceis pertinent to other Bantu languages as well.237Chapter 5Existence in other Bantulanguages5.1 IntroductionIn this chapter I extend the current proposal to other Bantu languages; Iclaim that a D contrast based on a core notion of ‘existence’ is pertinentto other Bantu languages with augments. I demonstrate that within Bantu,languages vary in the exact notion of existence they encode. My analy-sis makes several predictions that can be summarized in terms of micro-parametric variation in existence D systems:(349) Variation in Bantu augments.a. D systems which encode speaker personal knowledge behavelike St’át’imcets assertion-of-existence (AOE) Ds.b. D systems in which existence is only believed pattern like Natabelief-of-existence (BOE) Ds.c. It is possible to have augment languages that do not encodeexistence at all.238I examine augments in the following languages: Runyankore-Rukiga(henceforth, R/R or R/Rukiga) (JE13/JE14), Haya (Hya) (E22), Luganda(Gan) (JE15), Kinande (Nan) (JE42), Xhosa (Xho) (S41), Zulu (Zul) (S42),Bemba (Bem) (M42), and Dzamba (Dza) (C322). The choice of languagesdepends on availability of data. The data I present reveal that, exceptDzamba, all these languages encode existence: R/Rukiga, Haya, Luganda,Kinande, Xhosa and Zulu behave as belief-of-existence D systems. The Be-mba D system seems to behave like an assertion-of-existence D system, i.e.,like St’át’imcets. Dzamba, a language spoken in the Northeastern Demo-cratic Republic of Congo, appears to be the only exception in that its Ddistinction is based on novelty–familiarity (definiteness), which confirmsthe hypothesis in (349c) (cf. Bokamba 1971; Givón 1978, 2018). The dif-ferent Bantu augment/D contrasts I discuss here can be presented visuallyas in (350):(350) Bantu D contrastsSpeaker-oriented existence DsAssertion of existenceeg., BembaOvert D Covert D∅Belief of existenceeg., NataOvert D Covert D∅Other systemsDefiniteness, ...My findings challenge the previous claims made for Zulu and Haya thataugments/Ds in some of these languages have no semantic function (con-tra Halpert 2012, 2015, Riedel 2011, respectively). While I show that thenotion of existence is relevant to most of these languages, I do not claimthat augments across Bantu encode existence or have a semantic function.The languages I report on here fit the category of languages whose aug-ments have meaning. De Blois (1970) and Carter (1963) indicate that in239some languages augments may not have a semantic function, eg., in Tonga(M64). Since most previous studies only looked at the semantic featuresof definiteness and specificity, it remains a question for future researchwhether there are D systems which have been misanalysed, but it is reallyspeaker-oriented existence.I discuss the different D distinctions using the semantic diagnostics pre-sented in Chapters 2 and 41. While it is possible to apply a choice functionanalysis to the speaker-oriented existence Ds in these languages, I set asidethe formal treatment in this chapter and hence talk informally about scope.Overall, the following summary chart will guide our discussion. Belief ofexistence is marked as BOE and Assertion of existence is marked as AOE.Table 5.1: Different D distinctions among augment languagesD encodes/Languages R/R Hya Gan Nan Xho Zul Bem DzaDefiniteness 7 7 7 7 7 7 7 3Specificity 7 7 7 7 7 7 7 7AOE 7 7 7 7 7 7 3 7BOE 3 3 3 3 3 3 7 7I claim with Matthewson (1998, 2001) that D systems that encode existencehave a negative setting of the common ground parameter; as a result they do notencode definiteness. This means that Dzamba has a positive setting of the Com-mon Ground Parameter, as it encodes a novel-familiar/(in)definiteness distinction.I show that overt Ds in the speaker-oriented D systems can be used both in specificand non-specific contexts, as a result none of the Ds in these languages encodespecificity. The D system in Dzamba also does not encode specificity: the covert Dis used in both specific and non-specific contexts; the overt D which encodes def-1. There may be more than one diagnostic to test a particular semantic notion. For in-stance, on definiteness, I could also show whether or not augments in these languagesencode uniqueness under either Schwarz (2009; 2012)’s strong definite or weak definiteanalysis (where, very roughly speaking, strong definites are anaphoric and weak definitesare unique). However, in this chapter, I will use the novelty-familiarity test as the crite-rion for definiteness and assume that augments/Ds in these languages also do not encodeuniqueness. For more diagnostics and findings see Chapter 2 on Nata augments.240initeness is also specific hence there is no specificity contrast. I relate the currentproposal to previous proposals, arguing that, except Dzamba, speaker-orientedexistence Ds in the remaining languages do not encode (in)definiteness (contraDewees 1971; Mould 1974) or (non)specificity (contra Visser 2008; Allen 2014).Before we explore these systems, let me recapitulate the properties of the twospeaker-oriented existence D systems we looked at the previous chapter. One ofthe two ways in which existence is encoded is by assertion of existence, where thespeaker has personal knowledge of the individual. The second way is by belief ofexistence, where the speaker conveys a belief of existence in an entity correspond-ing to the NP description without necessarily having personal knowledge. Thus,the signature of existence Ds is that the D contrast is solely based on the notionof existence, whether it is by assertion or belief. Speaker-oriented existence Ds donot access the hearer’s knowledge or common ground knowledge (cf. Matthewson1998, 1999; Gambarage 2012; see also Givón 1978).The availability of data which show that overt Ds can be used in surmising,possible world contexts, or with non-materialized referents will help us to decipherwhich direction of existence the languages under discussion are leaning towards.In this thesis I have modelled assertion of existence as entailing belief of existence,i.e., in every context where existence is asserted, belief of existence also holds butnot vice versa. I focus on the semantic difference between arguments with theovert D, as in (351a), and those with the polarity sensitive D∅, (351b):(351) a. Argument DP: Db. Argument DP: D∅A wide range of data considered from these languages provides support forthe claim that augments/Ds have a semantic function and can be analyzed in aprincipled manner. The remainder of the chapter is organized as follows. In therest of §5.1 we look back at the semantic features encoded in Nata, then highlightthe properties of speaker-oriented existence Ds. In §5.2 I extend my proposal toother Bantu languages, where I discuss belief-of-existence D systems. In §5.3 Idiscuss the assertion-of-existence D systems. In §5.4 I turn to Dzamba, where Ishow that, like the English D system, the Dzamba D system encodes a novelty-familiarity contrast, but not a speaker-oriented existence D contrast. §5.5 is forsummary, final remarks and conclusion.2415.2 Belief of existence D systemsIn this section, I show that Runyankore-Rukiga, Haya, Luganda, Kinande, Xhosaand Zulu Ds pattern like Nata Ds which encode the speaker’s belief of existence.5.2.1 Existence Ds in Runyankore-RukigaThe augment/D system of Runyankore-Rukiga (R/Rukiga)–a Bantu language spo-ken in Uganda–has all the properties of a belief-of-existence D system. The onlyavailable work on augments in this language is by Allen 2014, a Runyankore-Rukiga speaker-linguist. Thus, the data presented here are from her work andfrom the elicitation I conducted with her2. I discuss the different R/Rukiga Dproperties (summarized in the table below) relative to other existence D systems.Table 5.2: R/Rukiga Ds and their correlation with other D systemsDiagnostics AOE BOE R/RukigaD encodes definiteness 7 7 7D encodes specificity 7 7 7Speaker’s personal knowledge required 3 7 7Used in cultural assumptions 7 3 3Used in possible worlds contexts 7 3 3Used with non-materialized referents 7 3 3Before I consider existence D data I first show that definiteness and specificityare not encoded in the R/Rukiga D system.5.2.1.1 R/Rukiga Ds do not encode definitenessData in R/Rukiga reveal that the overt D does not care about definiteness. Thesame D is used in novel contexts, as in (352), as well as in familiar contexts, (353).2. I thank Asiimwe Allen for her time spent in sharing her language with me and for heruseful comments during the discussion of augment properties in R/Rukiga.242(352) [Novel/indefinite context: A girl telling her friends about what hap-pened to her today:] [R/Rukiga, A.A.]NoomanyaNi-o-many-aPRES-2SG-know-FVngunguthato=mu-shaijao=mu-shaijaD=C1-mane=ri-zoobae=rii-zoobaD=C5-todayyaanyemereza?y-aa-ny-emer-ez-a1-PASTim-1SG-stop-APPL-FV‘Do you know that a man stopped me today?’(353) [Familiar/definite context: A day after, one of her friends follows up.][R/Rukiga, A.A.]o=mu-shaijao=mu-shaijao=C1-mano-kw-emere-izeo-kw-emere-izeSM1.REL-2sg-stop-PASTnyomwebazyonyomwebazyoyesterdayo-mu-reeb-ire?o-mu-reeb-ire2sg-OM-see-PAST‘Did you see the man who stopped you yesterday?’These data are accounted for nicely under the current proposal that the overtD here encodes the notion of existence and not definiteness. We will see that in allsituations where the speaker believes in the existence of an object that correspondsto the NP description, the overt D must be used.5.2.1.2 R/Rukiga Ds do not encode specificityOvert Ds in R/Rukiga are neutral with respect to specificity. The overt D can beused in specific contexts, as in (354), and the same D can also be used in non-specific contexts, as in (355).(354) [Specific context: B: I wish I had a spoon to eat with. My hands are dirty. You:][R/Rukiga, A.A].o=muo=muD=C18nshahon-shahoC9-bago-mw-oo-mu-oDEMrt-C18-MEDIALharimuha-ri-muC16-COP-18.ENCe=ki-giiko.e=ki-giikoD=C7-spoonLit: ‘In that bag there is a spoon.’243(355) [Non-specific context: There are several spoons in the cupboard. B asksa child...] [R/Rukiga, A.A].a. NdengyezaN-rengy-ez-a1sg-pass-APPL-FVe=ki-giikoe=ki-giikoD=C7-spoon‘Pass me a spoon.’b. *Ndengyeza*N-rengy-ez-a1sg-pass-APPL-FVki-giikoki-giikoC7-spoonIntended: ‘Pass me a spoon.’Since overt Ds are used both in specific and non-specific contexts, specificityis neutralized. This challenges Allen’s (2014) proposal that augments behave asmarkers of specificity in R/Rukiga. These data support the current proposal thatthe overt D in (non)specific contexts indicates that the speaker is committing tothe belief that the noun phrase’s referent exists. In the next section I show that Dchoice in R/Rukiga is forced by the notion of existence.5.2.1.3 D distinction in R/RukigaIn R/Rukiga, the overt D encodes the speaker’s belief of existence, and the polar-ity D∅ encodes the speaker’s non-belief of existence of a referent. In (356a) thespeaker wishes to commit to the existence of a/the book; hence, the overt D isused. This is also the case with the negative sentence in (356b), in which the DPtakes wide scope with respect to the non-factual operator/NEG.(356) [Adapted from Allen 2014: 139-140].D Choice in R/Rukigaa. o=mw-ishikio=mu-ishikiD=C1-girlna-a-shom-ani-a-shom-aPROG-3SG-read-FVe=ki-taboe=ki-taboD=C7-book‘A/the girl is reading a/the book.’b. [DP >¬]o=mw-ishikio=mu-ishikiD=C1-girlti-a-ri-ku-shom-ati-a-ri-ku-shom-aNEG-3SG-PROG-INF-read-FVe=kitaboe=ki-taboD=C7-book‘A/the girl is not reading a/the book.’244If the speaker does not wish to commit to the belief of existence of an entity, theD∅ must be used, as in (357a). The D∅ must be interpreted under the c-commanddomain of the non-factual operator, which is the syntactic environment that allowsthe non-existential interpretation. The polarity D∅ will be ungrammatical whenused in syntactic environments that induce existential interpretation, e.g., in adeclarative sentence as in (357b), or where the OM is used, (357c):(357) [Adapted from Allen 2014: 139-140].D Choice in R/Rukigaa. [¬ >DP]o=mw-ishikio=mu-ishikiD=C1-girlta-a-ri-ku-shom-ati-a-ri-ku-shom-aNEG-3SG-PROG-INF-read-FVkitaboki-taboC7-book‘The girl is not reading any book.’b. *o=mw-ishiki*o=mu-ishikiD=C1-girlna-a-shom-ani-a-shom-aPROG-3SG-read-FVki-taboki-taboC7-bookIntended: ‘A/the girl is reading a/the book.’c. *o=mw-ishiki*o=mu-ishikiD=C1-girlna-a-ki-shom-ani-a-ki-shom-aPROG-3SG-OM7-read-FVkitaboki-taboC7-bookIntended: ‘The girl is not reading (it) the book.’The D contrast based on the notion of existence also obtains with subject DPs3.In (358a), the speaker has an entity in mind, and hence uses the belief-of-existenceD. When the speaker does not wish to commit to a belief of existence, s/he mustswitch the D and use the polarity sensitive D∅, (358b).(358) [R/Rukiga, A.A].D choice with subject DPsa. o=mu-shaijao=mu-shaijaD=C1-mano-ri-kukund-ao-ri-kukund-aSM1-IMPF-love-FVMariaMariaMaria‘A man likes Maria.’3. Allen (2014) does not investigate subject DPs with polarity Ds. In my elicitation, I notedthat R/Rukiga has a syntactic constraint that only the highest argument c-commanded bythe Op is licensed. For instance, a null D on a book in John didn’t read any child a bookwouldn’t be licensed. See Chapter 6 for independent constraints on licensing, locality andc-command.245b. ti-haineti-haineNEG-there.ismu-shaijamu-shaijaC1-mano-ri-kukund-ao-ri-kukund-aSM1-IMPF-love-FVMariaMariaMaria‘There is no man (who) likes Maria.’Following Visser (2008), Allen claims that R/Rukiga Ds are specific. She alsoproposes that there are morphosyntactic elements in the sentence that can rein-force specificity/definiteness, e.g., object markers, demonstratives, etc4. The dataI collected and presented above challenge the view that Ds in these languagesare specific. The view that overt Ds can co-occur with morphosyntactic elementsintroducing definiteness/specificity is consistent with the fact that Ds in these lan-guages are neutral with respect to definiteness/specificity, which follows from myanalysis of them as speaker-oriented existence Ds.5.2.1.4 Overt Ds as belief of existence DsAll the diagnostics for belief-of-existence Ds are met in R/Rukiga. First, Allenshows that generics in R/Rukiga do not refer to any particular/specific referent;they can be referring to a kind/genus or describing a state of affairs or a habit, butDPs must appear with the overt D.(359) [Runyankore-Rukiga, Allen 2014: 167]Genericsa. a=ba-anaa=ba-anaD=C2-childØ-ba-kund-aØ-ba-kund-aHAB-SM2-like-FVku-zaan-aku-zaan-aINF-play-FV‘Children like to play.’b. * ba-ana* ba-anaC2-childØ-ba-kund-aØ-ba-kund-aHAB-SM2-like-FVku-zaan-aku-zaan-aINF-play-FV‘Children like to play.’4. Allen shows that there is a definite proclitic element which cannot co-occur with theaugment, e.g., wa (*o=)mwaana ‘the other child.’ This may indicate that the element sitsin D position. The element seems to be the partitive article used in partitive structures. Thisis different from Nata, in which partitive structures are introduced by D-doubling (refer toChapter 3).246In R/Rukiga the overt D is used in in surmising contexts, as in (360a), or withDPs referring to non-materialized future referents, (361a).(360) [R/Rukiga, A.A.]Surmising context: There is a sun-shower. B says:a. e=m-pitsie=n-hitsiD=C9-leopardni-e-hingir-ani-e-hingir-aPRES-9-send.off-FV‘A leopard is sending off (its daughter).’b. * m-pitsi* n-hitsiC9-leopardni-e-hingir-ani-e-hingir-aPRES-9-send.off-FVIntended: ‘A leopard is sending off (its daughter).’(361) [Future referent context: B is considering donating a chair to a newschool. He believes he can find one to buy.] [R/Rukiga, A.A.]a. Ni-nz-aNi-nz-a1SG-come-FVku-reet-aku-reet-aINF-bring-FVe=n-tebee=n-tebeD=C7-chairLit: ‘I will come to bring a chair.’b. *Ni-nz-a*Ni-nz-a1SG-come-FVku-reet-aku-reet-aINF-bring-FVn-teben-tebeC7-chairIntended: ‘I will come to bring a chair.’Overt Ds are freely used when referring to entities in a possible world. Considerthe use of the attitude predicate ronda ‘look for’ (362a):(362) [R/Rukiga, A.A]BOE in attitude verbsa. MariaMariaMariana-a-rond-ani-a-rond-aPRES-3SG.SUBJ-look.for-FVe=mi-yembee=mi-yembeD=C4-mangoes‘Maria is looking for mangoes.’247b. *Maria*MariaMariana-a-rond-ani-a-rond-aPRES-3SG.SUBJ-look.for-FVmi-yembemi-yembeC4-mangoesIntended: ‘Maria is looking for mangoes.’These contexts define R/Rukiga as a belief-of-existence system. Overt Ds here donot assert existence, i.e., personal speaker knowledge is not required.5.2.2 Existence Ds in HayaI show that the underlying semantic principle that drives D choice in Nata andR/Rukiga, namely that D encodes the speaker’s belief of existence, also works forHaya augments. Data for Haya come from my field notes as well as from De Blois(1970), Chagas (1977) and Riedel (2011) and references therein5.Riedel (2011) asks whether Haya augments have any structural propertiesthat may favour their analysis as Ds. She argues that Haya augments do notencode definiteness or specificity. As a consequence of this, she analyses theHaya augment as a strong agreement marker, or as a variant of a class prefix withno definable semantic function. My proposal for Haya augments as Ds builds onthe determiner-like role of the augment found in all argument DPs. Augmentsare required with argument nominals (individual/entity-denoting DPs), andobligatorily absent in predicate nominals (property-denoting nominals), as (363)shows.[Context: In (363a) a student is disclosing that he is the one who stole anotherstudent’s pen. In (363b) the students are talking about John’s behaviour].(363) [Haya, A.K.]Argument–predicate contrasta. Argument=DPInye1sgndiCOP*(o=)mu-shumaD=C1-thief‘I am the thief.’b. Predicate=φPJohnJohn(*o=)mu-shuma(D=)C1-thief‘John is a thief.’5. Thanks to Angela Katabaro for providing me with the Haya data and for sharing herlanguage with me.248The D requirements for argument nominals support the cross-linguistic general-ization about argument-predicate partition (see Chapter 3).In the remainder of this section, I discuss the denotation of Ds; I argue thatHaya D choice is based on the speaker’s belief of existence. The different Haya Dproperties suggest that Haya Ds are fit to be analyzed as belief-of-existence Ds assummarized in Table 5.4.Table 5.3: Haya Ds and their correlation with other D systemsDiagnostics AOE BOE HayaD encodes definiteness 7 7 7D encodes specificity 7 7 7Speaker’s personal knowledge required 3 7 7Used in cultural assumptions 7 3 3Used in possible worlds contexts 7 3 3Used with non-materialized referents 7 3 35.2.2.1 Haya Ds do not contrast for definitenessHaya data support Riedel’s (2011) observation that Haya augments do not contrastfor definiteness. The overt D is used in novel contexts, (364) and the same D mayalso be used in familiar contexts, (365).(364) [Novel/indefinite context: B is telling her friends about what happenedon her way home:] [Haya, A.K.]M-ba-gamb-ile?FOC-1sg-tell-SUBJVo=mu-shaijaD=C1-manyanyemelezeSM1.PST.1SG-stopo=mu-kiandaD=C18-way‘Should I tell you? A man stopped me on the way.’249(365) [Familiar/definte context: A day after, B brings up the same story to thesame people.] [Haya, A.K.]o=mu-shaijaD=C1-manowaSM1-RELya-ny-emelez-eSM1-1SG-stop-TAMo=mu-kiandaD=C18-wayn-ka-mu-bona1sg-PST-OM-see‘I saw the man who stopped me on the way.’In both contexts the speaker has an entity in mind that corresponds to the NPdescription, which is consistent with the current proposal that overt Ds encodeexistence. Turning to specificity, I show that Haya Ds do not distinguish Ds basedon this notion either.5.2.2.2 Haya Ds do not encode specificityThe strongest argument against the specificity account comes from the lack of Dchoice based on (non)specificity. The overt D can be used in specific contexts suchas (366), and the same D can also be used in non-specific contexts, (367).(366) [Specific context: B: I wish I had a cup. I would drink from this stream.You:] [Haya, A.K.]Ky-ali-yoSM7-be-LOCe=ki-kompeD=C7-cupo=mu-liREL=C18-bee=n-shalaganyaD=C9-plastic.bag‘There is a cup in that plastic bag.’ (Lit. ...which is in that plastic bag)(367) [Non-specific context: There are several cups in the cupboard. B asks achild...] [Haya, A.K.]a. N-det-ele1sg-bring-SUBJVe=ki-kompeD=C7-cup‘Bring me a cup.’b. *N-det-ele1sg-bring-SUBJVki-kompeC7-cupIntended: ‘Bring me a cup.’250In the current account, both the specific and the non-specific interpretationsderive from a single generalization: speakers believe that the NP referent exists.That is, there exists at least one cup that satisfies the NP description. In whatfollows I discuss how D choice in Haya is forced by the notion of existence, anddiscuss a variety of contexts in which belief-of-existence Ds are used.5.2.2.3 D distinction in HayaThe familiar signature of existence Ds is found with Haya Ds as well. The crucialdistinction Haya Ds make is that an overt D always encodes a speaker’s belief ofexistence while the polarity D∅ indicates that the speaker fails to convey a beliefof existence. In (368a) the DP appears with the overt D, in which case it denotesthat o=mutambi ‘a/the doctor’ exists. In (368b) the DP with the overt D has widescope over negation, in which case it preserves the existential interpretation. Notethe use of OM in (368b) which always forces the definite reading:(368) D choice in Haya [Adapted from Chagas 1977: 41]a. n-a-bona1sg-PST-seeo=mu-tambiD=C7-doctor‘I saw a/the doctor.’b. ti-n-a-mu-bonaNEG-1sg-PST-OM-seeo=mu-tambiD=C7-doctor‘I didn’t see him, the doctor.’On the other hand, if the speaker does not wish to commit to the existence ofa doctor, the non-belief-of-existence D is used, as in (369a). We see consistentlythat the polarity D∅ must be used under the scope of some non-factual operatorfor the non-belief-of-existence reading to hold, hence (369b) is ungrammatical.(369) D choice in Haya [Adapted from Chagas 1977: 41]a. ti-n-a-bonaNEG-1sg-PST-seemu-tambiC7-doctor(wena-w-ena)(RED-C1-any)‘I didn’t see a/any doctor.’b. *n-a-bona1sg-PST-seemu-tambiC7-doctorIntended: ‘I saw a/the doctor.’251This distinction obtains with subject DPs also, (370a). In fact, it is possible forNEG to scope over the subject and object DPs, as (370b) shows.(370) a. o=mu-kaziD=C1-womana-sindik-ileSM1.PST-push-PFVo=mw-aanao=mw-aana‘A/the woman pushed a/the child.’b. ta-li-yoNEG-be-LOCmu-kaziC1-womana-sindik-ileSM1.PST-push-PFVmw-aanaC1-child‘No woman pushed a/any child.’In (370a) the speaker believes in the existence of the referents; hence, the overtD is used. Data such as (370b) show that in Haya the speaker can use the covertD only if they do not have a belief of existence of an entity of the NP.Riedel (2011) illustrates that with the augment, the NPI/‘any’ reading (in neg-ative sentences) is impossible, as shown in (371a); however, the use of a FreeChoice Item (FCI) (in positive sentences) requires an augment, as in (372a).(371) Polarity contexts: no overt D [Haya, adapt. Riedel 2011: 8]a. *ti-n-a-bonaNEG-1sg-PST-seee=ki-ntuD=C7-thingkyona-ky-onaRED-C7-anyIntended: ‘I didn’t see anything.’b. ti-n-a-bonaNEG-1sg-PST-seeki-ntuC7-thingkyona-ky-onaRED-C7-any‘I didn’t see anything.’(372) FCI context: Overt D is required [Haya, Adapt. Riedel 2011: 9]a. tu-la-gi-shemerwa2sg-FUT-OM4-be.pleased.bye=mi-changoD=C4-contributionyona-y-onaRED-C4-any‘We will be happy about any contributions.’b. *tu-la-gi-shemerwa2sg-FUT-OM4-be.pleased.bymi-changoC4-contributionyona-y-onaRED-C4-any‘We will be happy about any contributions.’252These data provide further support for the D choice based on the notion ofexistence. The NPI/FCI element kyona-kyona induces maximal domain widening(see Kadmon and Landman 1993; Kratzer and Shimoyama 2002; Chierchia 2006,2013, Osa-Gómez 2016; and others)6. Thus, the polarity D∅ in (371b) is consistentwith the speaker’s non-belief of existence in that the NP domain does not containany entity. For (372a), the opposite is true.In what follows I present general contexts that show that like Nata, Hayaovert Ds are used in all belief-of-existence contexts, which sets them apart froman assertion-of-existence system like St’át’imcets.5.2.2.4 Overt Ds as belief-of-existence DsHaya overt Ds pass the diagnostics for belief-of-existence Ds used in this thesis.The overt D can be used in surmising contexts, such as (373), and can also be usedto refer to referents that have not materialized yet, as in (374):(373) [Haya, A.K.][Surmising context: There is a sun-shower. B says:]a. e=n-taleD=C9-lionesse-i-zal-aSM9.T-give.birth-FV‘A lioness is giving birth’.b. * n-taleC9-lionesse-i-zal-aSM9.T-give.birth-FVIntended: ‘A lioness is giving birth’(374) [Future referent context: B is considering donating a chair to a newschool. He believes he can find one to buy.]a. N-da-ba-tw-el-a1sg-FUT-2pl-bring-APPL-FVe=ki-tebe [Haya, A.K.]D=C7-chair‘I will bring you a chair’.b. *N-da-ba-tw-el-a1sg-FUT-2pl-bring-APPL-FVki-tebeC7-chair‘I will bring you a chair’.6. The NPI in many Bantu languages is made up out of a reduplication process having areduplicant and a base (a class prefix and an NPI radical).253Finally, overt Ds can be used in possible world contexts with the attitude verbswith the meanings want or look for. For instance, (375) shows that the overt D isused with iga ‘look for.’(375) [Haya, A.K.]Belief of existence in possible worldsa. Na-ig-a3sg.TA-look.for-FVa=ma-nyembeD=C6-mangoes‘She is looking for mangoes.’b. *Na-ig-a3sg.TA-look.for-FVma-nyembeC6-mangoesIntended: ‘She is looking for mangoes.’The data clearly show that Haya overt Ds can be freely used in contexts whereexistence is only conveyed presumptively. In the final section on Haya, I discusssome data that have been claimed to present a problem for the previous analyses.I will show that the account developed here may offer a promising solution.5.2.2.5 Accounting for residual cases in HayaThe first data set I consider is the case listed in Riedel (2011), in which an augment-less nominal baana ‘children’ appears in a small clause structure: i.e., sentencesfound in embedded contexts but without overt tense marking (see Déchaine 1993;and others), as in (376):(376) [Haya, Riedel 2011: 5]Predicates in small clausesN-ka-bonaSM1sg-PST-see/considerba-anaC2-childboonaC2.all‘I consider them all children.’*‘I saw/considered all (the) children.’This case is not problematic if we analyze baana as a nominal predicate,in which case it is predicted to lack a DP shell, consistent with my argument–predicate generalization discussed in Chapter 3. The predicate here denotes aproperty (i.e., having the quality of children), not individuals. Analyzing this ele-ment inside a small clause as a predicate is the standard analysis of small clauses(see Déchaine 1993; Longobardi 1994 for discusion).254The second and last case concerns some DPs with a polarity D∅ rendered withan apparent definite reading in examples such as (377) and (378). Both Chagas(1977) and Riedel (2011) present translations which make it appear that the po-larity D∅ can also be interpreted with a specific/definite construal. This cannotbe correct, as I argue below.(377) The translation of D∅ [Haya, adapt. Riedel 2011: 8]ti-n-a-bonaNEG-1sg-PST-seeba-naC1-childba-toC2-small‘I didn’t see the small children.’(378) The translation of D∅ [Haya, adapt. Chagas 1977: 42]ti-ŋ-kumanyaNEG-1s-knowmw-anaC1-childy-a-ku-ha-il-e3sg-PST-2sg-give-PFVkitabobook‘I don’t know the child who gave you a book.’If the DPs containing the polarity D∅ involved a definite or existential inter-pretation, the theory developed in this thesis would be meaningless, i.e., it wouldobscure the contrast between the overt D and the polarity sensitive D∅. However,based on my Haya consultant’s data and the general property of Ds in Haya, Iargue that the polarity D∅ in Haya does not receive a definite interpretation. Iargue instead that the above readings arise from conversational implicatures andcannot be the literal meaning of the sentences.In Grice’s (1975) Cooperative Principle of conversation, implicatures are in-ferences arising from pragmatics and are not tied to any structural configurationsor lexical items (Grice 1975; Levinson 1983; Gillon 2006). The implicatures mayarise either by (i) the speaker flouting/violating a conversational maxim or (ii)the speaker being cooperative with/obedient to the conversation maxims. Thenotion of implicature applies to the puzzling Haya data in (377)–(378) above inthe following way. I illustrate first using Nata data.Let us consider a similar situation from Nata. In (379a), the speaker obeysGrice’s maxim of Quantity by giving exactly as much information as is required:s/he will not carry the definite child in the context. In (379b), on the other hand,also a possible answer in this context, the speaker is violating the maxim of Quan-tity by being nonspecific. The speaker is providing more information than is re-quired, by literally saying that s/he will not carry any child:255(379) [Context: A parent wants a babysitter to take care of his child, but shehas no time.] [Adapted from Osa-Gómez (2016: 12, ex 41)]a. Ne-te-ku-ghegh-aN-te-kuɣɛɣ-a1sg-NEG-carry-FVu=mw-aana [by obeying Quantity]u=mu-anáD=C1-child‘I won’t carry the child.’b. N-te-ku-ghegh-aNe-te-kuɣɛɣ-a1sg-NEG-carry-FVmw-aana [by violating Quantity]mu-anáC1-child‘I won’t carry a/any child’ (literal meaning).‘The speaker won’t carry the child’ (implied meaning).Obviously, the speaker is committed to the literal meaning which violates themaxim of Quantity: ‘Do not make your contribution more informative than isrequired’ (Grice 1975; Levinson 1983). But why would speakers choose to violatethe maxim in this context? Deviating from using the overt D/choosing not topick out the referent under discussion (i.e., by using the more general denial) is adistinct communicative style in many languages (see Brown and Levinson 1987;Bousfield 2008; and others on politeness). Nata speakers particularly use this stylewhen they do not want to sound ‘mean’, ‘rude’ or ‘unfriendly’. In this contextwhere all speech participants know there is a familiar child, the literal meaning ‘Iwon’t carry any child’ it implies that ’I won’t carry the particular child in context.’If we analyze the Haya cases with the literal meaning we see that there isno argument about a lack of semantic predictability of Haya Ds, as Chagas orRiedel thinks. There is also no need to appeal to a different analysis such as theone Chagas proposes. Chagas 1977, for instance, tries to unify the analysis ofHaya kinship terms, proper names (which lack a D) and data with polarity Dsappearing in the context of domain restriction by representing the polarity D∅with the implied meaning. He writes:Using the hypothesis that deletion of the PP [augment] indicates oldinformation on the part of both speaker and hearer, the explanationfor the absence of the PP in [...] is obvious [Chagas 1977: 40].256I do not agree with this view. I propose instead that the examples with thepolarity Ds such as (380), presented by Chagas, should be re-glossed with theliteral meaning, in which case the speaker intends an empty reference for the NPdomain under consideration.(380) NPI contexts: D∅ required [Haya, Adapt. Chagas 1977: 42; A.K.]ti-ŋ-kumanyaNEG-1s-knowmw-anaC1-childy-a-ku-ha-il-e3sg-PST-2sg-give-PFVkitabobook‘I don’t know any child who gave you a book.’ (literal)‘I don’t know the child who gave you a book.’ (implied)Likewise, based on my Haya consultant’s intuitions, I have added the literalmeaning of the sentences from Riedel in order to be consistent with the generalbehaviour of the Haya polarity sensitive D∅.(381) NPI contexts: D∅ required [Haya, A.K; adapt. Riedel 2011: 8]a. ti-n-a-bonaNEG-1sg-PST-seeba-naC1-childba-toC2-small‘I didn’t see any small children’ (literal meaning).’‘I didn’t see the small children’ (implied meaning).’b. ti-n-a-bonaNEG-1sg-PST-seeba-naC1-childba-ngeC2-my‘I didn’t see any of my children’ (literal meaning).’‘I didn’t see my children’ (implied meaning).’c. ti-n-a-bonaNEG-1sg-PST-seeba-naC1-childba2ASSOCKato1Kato‘I didn’t see any of Kato’s children’ (literal meaning).’‘I didn’t see Kato’s children’ (implied meaning).’I conclude that the apparent definite reading arises as an entailment of a literalmeaning. The use of a polarity D∅ in these contexts is evidenced by the fact thatnon-speaker-oriented existence Ds do not deny existence of an entity; rather, theyindicate that the speaker fails to convey a belief of existence of the entities denotedin the relevant NP domain, i.e., entities that correspond to the NP description andthe main predicate. Thus, I didn’t see any small children, (381a), literally assertsthat the speaker does not believe in the existence of small children that s/he saw,257and implies therefore that the speaker did not see the particular small childrenin the discourse context either. Assuming that the non-belief-of-existence readingholds for all polarity Ds, I treat the polarity Ds in Haya as denoting non-belief ofexistence as their literal meaning, consistent with the analysis developed here.5.2.3 Existence Ds in LugandaScholars of Luganda state that the function of augments (a.k.a initial vowels) can-not be stated in a single semantic generalization (Ashton 1954; Dewees 1971;Mould 1974; Hyman and Katamba 1993). Ashton et al. (1954: 402) say that aug-ments have to be studied under arbitrary conditions or as indicating “some specialimplications... frequently emphasis.” Dewees (1971) is of the opinion that...its omission and use sometimes correspond to the poles of generaland particular, indefinite and definite, or unemphatic and emphatic.However, there are situations where these observations are contra-dicted. The article characteristics are overridden by morphologicallyand syntactically conditioned rules. (Dewees 1971: 13-14)Mould (1974) reports that augments in Luganda correspond to definite-ness/presupposition of existence and specificity/referentiality, particularly whenused in topicalized structures. His conclusion is that “Luganda has developed adefiniteness contrast with this syntactic device [topicalization]” (p.225). Mouldclaims that except for generic sentences, indefinites correspond to the non-use ofaugments. I will claim instead that Luganda is a speaker-oriented D system, andthat features such as definiteness or specificity do not come from the semantics ofaugments, rather they come from elsewhere.Hyman and Katamba (1993) have argued against previous proposals that thefunction of augments can be reduced to semantic/pragmatic primes. The data I dis-cuss below support the idea that a successful D account cannot be one postulatingD features like (in)definiteness or (non)referentiality/(non)specificity. However,in spite of this I will show that there is a possible unified semantic account forLuganda Ds.I use Luganda data to motivate my semantic proposal and show that it doesseem to be the case that the choice between different Ds in Luganda is forcedby the notion of existence in the sense discussed in this thesis. While it is truethat other factors such as focus marking interact with the augment/D in a waynot observed in Nata (see Chapter 6 for details), such parameters can be dealtwith independently of the current semantic proposal. The correlation between258semantic properties of Luganda Ds and other existence Ds I discuss is given in thechart below:Table 5.4: Luganda Ds and their correlation with other existence DsDiagnostics AOE BOE LugandaD encodes definiteness 7 7 7D encodes specificity 7 7 7Speaker’s personal knowledge required 3 7 7Used in cultural assumptions 7 3 3Used in possible worlds contexts 7 3 3Used with non-materialized referents 7 3 3I show that, even taking into account the focus facts to be discussed in Chapter6, Luganda augments/Ds behave like belief-of-existence Ds. I start by addressingthe usual concerns––whether or not Luganda augments/Ds encode definiteness orspecificity.5.2.3.1 Luganda Ds do not encode definitenessDefiniteness is not a condition for Luganda Ds. The augment is used both in novel,(382a), as well as in familiar contexts, (383)7. This rules out an analysis claimingthat the D is linked to definiteness:(382) [Novel/indefinite context: A boy is telling his girlfriend about what hap-pened to him today:] [Luganda, R.M.]a. O-manyi?2SG-knowo=mu-kaziD=C1-womanyanyimirizaSM1.PST.1SG.stopleerotoday‘You know (what)? A woman stopped me today.’b. *O-manyi?2SG-knowmu-kaziC1-womanyanyimirizaSM1.PST.1SG.stopleerotodayIntended: ‘You know (what)? A woman stopped me today.’7. Thanks to Eddie Amaitum and Richard Musoke for providing me with extra Lugandadata when I needed to fill some gaps.259(383) [Familiar/definite context: A day after, the boy asks his girlfriend.][Luganda, R.M]o=mu-kaziD=C1-womaneyanyimirizaREL.SM1.TM.1sg.stop.CAUS.FVegguloyesterdaywamulabye?2SG.PST.OM1.seeLit: ‘The woman who stopped me yesterday, did you see her?’‘Did you see the woman who stopped me yesterday?’Under the current account, the use of the overt D in o=mukazi ‘a/the woman’is indicative of the fact that the speaker believes in the existence of a woman whostopped him yesterday. I show next that Luganda Ds do not encode specificityeither.5.2.3.2 Luganda Ds do not encode specificityThe data below rule out the possibility that Luganda Ds encode specificity. Asshown, the overt D may be used in specific contexts such as (384); the same D isalso used in non-specific contexts as in (385), as well as in generics, (386)8.(384) [Specific context: B: I wish I had a cup. I would drink from this stream.You:] [Luganda, E.A.]a. Mu-nsawoLOC18-bagmulimuC18.there.ise=ki-kopoD=C7-cup‘In the bag there is a cup.’b. *Mu-nsawoLOC18-bagmulimuC18.there.iski-kopoC7-cupIntended: ‘In the bag there is a cup.’8. Note that objectives of locative prepositions in Luganda do not take a augment, henceMu-nsawo in the bag, has no augment. See Chapter 6 for further discussion260(385) [Non-specific context: There are several cups on a table; you can grabany one.] [Luganda, E.A.]a. Mpereza1sg.passe=ki-kopoD=C7-cup‘Pass me a cup.’b. *Mpereza1sg.passki-kopoC7-cupIntended: ‘Pass me a cup.’(386) [Luganda, E.A.]Generic statements: overt D is OKa. e=biwuguluD=owlsbiyigahunte=biwojoloD=butterflies‘Owls hunt butterflies.’b. * biwuguluowlsbiyigahunte=biwojoloD=butterflies‘Owls hunt butterflies.’What both the specific and the non-specific interpretations have in common isthat the speaker is conveying the belief of existence of a cup or cups. Next, I showthat the Luganda D contrast is necessitated by the core notion of existence.5.2.3.3 D distinctions in LugandaI claim that D choice in Luganda is forced by the notion of existence, evidencedby examples such as (387). In (387a), when the speaker believes that an entitythat satisfies the NP description exists, the overt D is used. On the other hand, ifthe speaker does not wish to commit to the belief of existince of a referent, thechoice is to use the polarity D, as in (387b), in which case it must be interpretedunder the scope of a non-factual operator. If there is no non-factual operator tolicense the D, the result is ungrammaticality, as in (387c).261(387) D distinction in Luganda [Luganda, adapt. Mould 1974:225]a. Nalaba1sg.PST.seeo=mu-sawoD=mu-sawo‘I saw a/the doctor.’b. saalabaNEG.1sg.PST.seemu-sawomu-sawo[¬ >DP]‘I didn’t see a/any doctor.’c. *Nalaba1sg.PST.seemu-sawomu-sawoIntended: ‘I saw a/the doctor.’Mould argues that the D in declarative sentences such as (387a) is ambiguousbetween a definite and an indefinite interpretation. As I propose above for otherlanguages, the Luganda augment is neutral with regards to definiteness/specificity;therefore, the definiteness feature comes from elsewhere, e.g., from the anaphoricuse of OM, as Mould correctly illustrates.5.2.3.4 Overt Ds as belief-of-existence Ds in LugandaThe Luganda D system behaves like a belief-of-existence D system. Overt Ds arerequired in surmising contexts, (388a), in possible worlds, (390), and are also usedto talk about non-materialized referents, as in (390a)9.(388) [Luganda, E.A.][Surmising context: There is a sun-shower. B says:]9. Hyman and Katamba (1993) observe that in grammatical contexts inducing contrastivefocus, the unaugmented form may be used. Mould (1974) also has the same observation.For instance, Mould gives the following example:(i) Focus marking in Luganda [Luganda, Mould 1974: 227]njagalaI.wantmugaatibread‘I want some BREAD.’In Chapter 6 I will discuss the focus parameter in Bantu and show that focus marking utilizesaugmentless nominals, consistent with a predicate focus account (Zimmermann 2008, 2016,see also Zerbian 2006). I will focus on covert Ds here and argue that they are always licensedby a non-factual operator.262a. e=n-goD=C9-leopardezalaSM9.TM-give.birth‘A leopard is giving birth.’b. * n-goC9-leopardezalaSM9.TM-give.birthIntended: ‘A leopard is giving birth.’(389) D choice with attitude verbs [Luganda, Mould 1974: 226]a. Nnonya1sg.lookingo=musawoD=doctor‘I am looking for a/the doctor.’b. *Nnonya1sg.lookingmusawodoctorIntended: ‘I am looking for a/any doctor.’(390) [Future referent context: B is thinking of ordering books to donate to anew school. He doesn’t know which company he will buy from, but hepledged.] [Luganda, E.A.]a. N-ja-kutona1sg.FUT.donate/givee=bi-taboD=C8-book‘I will donate books.’b. *N-ja-kutona1sg.FUT.donate/givebi-taboC8-bookIntended: ‘I will donate books.’These examples show that speakers are not asserting existence of a referent butare only conveying a belief of existence of a referent of the NP. The hypothesis thatthe D distinction is based on the core notion of existence is upheld in Luganda. TheD properties discussed here reveal that Luganda Ds behave like the Nata belief-of-existence Ds. Having defined the Luganda D system, let us now consider anexample presented in Hyman and Katamba (1993) as posing a challenge for thesemantic predictability of augments/Ds.2635.2.3.5 Accounting for residual issues in LugandaThe use of the overt D in (391) is argued to present a problem for the theory thatclaims that augments/Ds encode specificity or definiteness. Hyman and Katamba(1993: 219) offer a solution that the acceptability of the overt augment here hasto do with the type of clause it is in, and not specificity or definiteness.(391) Subjunctive/If-clauses [Luganda, H&K 1993: 218]ànáákóló=kíhe.will.do=whatbwèànáálábàif.he.seesè=n-jóvúD=C9-elephant‘What will he do if he sees an elephant?’Lit., ‘He will do what, if he sees an elephant?’Hyman and Katamba are correct that the overt D in è=njóvú ‘elephant’ cannotbe explained using specificity/referentiality or definiteness, given that it has a non-specific reading. However, by treating the Luganda D system as having propertiesof a belief-of-existence system, we can account for the use of an existence D here.In (391) the speaker is not asserting that there is an elephant; rather, the speakeris presumptively indicating based on his experience/belief that there will see one.As we saw in Chapter 3, Nata would allow similar structures, as seen below:(392) [Context: In Nata elephants often storm villagers’ farms. B knows thatMasato is scared of elephants. But B wants to take Masato to the park andthere is a possibility of seeing one there. B decides to check with Masato’smom first:] [Nata]Ne=bhweNe=βhweCOP=WHa-ghu-kọr-aa-ɣu-kɔŕ-a3g-FUT-do-FVa-raa-rọr-ẹa-ráá-rɔr-ɛ3sg-PROG-see-SUBJVa=n-choghua=n-tʃóɣuD=C9-elephant‘What will he do if he sees an elephant?’Now consider a context in which the speaker does not believe that there will beany elephants. Note that the subjunctive mood here licenses the polarity D, as wesaw in Chapter 3.(393) [Context: We are setting up to go for an adventure in Yoho National Parkin British Columbia. Eli is scared of big animals. I believe there are no264elephants there, but I ask my wife anyway to see her reaction:] [Nata]Ni-nga-h-ẹNi-ŋga-h-ɛCOP-COND-WH-SUBJVa-rọr-ẹ=họa-rɔŕ-ɛ=hɔ3sg-see-SUBJV=LOCn-choghun-tʃóɣuC9-elephant‘How about if he sees an elephant?’Unfortunately the contrasting data with the polarity D was not elicited in Lu-ganda. However, Hyman and Katamba’s point about clause type becomes relevantfor (393). This question is an if-clause (with an inverted protasis); therefore, therelevant non-factual operator here must be the conditional Op. For the Lugandaand Nata cases with overt Ds (where existence holds), belief of existence is en-tailed. This can be accounted for by the analysis suggested in Chapter 4.5.2.4 Existence Ds in KinandeI wish to show that Kinande augments/Ds behave like belief-of-existence Ds inNata, R/Rukiga, Haya and Luganda. No previous semantic work exists on theaugments in this language. A brief remark on the function of Kinande augmentsfrom Schneider–Zioga and Mutaka (2014) is as follows:In Kinande, the presence of the augment is somewhat correlated withdefiniteness and specificity... Indefinite noun phrases may or may nothave augments depending on the semantic and syntactic context...[Schneider–Zioga and Mutaka 2014: 5-6]]I argue here for a stronger generalization. Kinande overt Ds are compatiblewith definite, indefinite, specific and non-specific contexts, however, definitenessor specificity are not part of their semantics. I show that what forces D choice inKinande is speaker’s belief of existence of an NP referent. The summary of resultsin Table 5.5 will guide our discussion, leading to the conclusion that Kinande Dsbehave as belief-of-existence Ds.265Table 5.5: Kinande Ds and their correlation with other existence DsCore diagnostics AOE BOE KinandeD encodes definiteness 7 7 7D encodes specificity 7 7 7Speaker’s personal knowledge required 3 7 7Used in cultural assumptions 7 3 3Used in possible worlds contexts 7 3 3Used with non-materialized referents 7 3 3I discuss these properties below. I start by showing that D choice in Kinandeis not forced by definiteness or specificity.5.2.4.1 Kinande Ds do not encode definitenessThe overt D in Kinande can be used in both novel, (394), as well as familiar dis-course contexts, (395). This rules out the possibility that the D distinction inKinande relies on the notion of definiteness10:(394) [Novel/indefinite context: A girl is telling her friends about whathappened to her today:] (Kinande, P.M.)u-n-asi2sg-TM-knowkwe?whato=mu-pulisiD=C1-police mo-a-ka-nyi-manayamo-SM-PST-1SG-stopmunabwiretoday‘You know what? A police officer stopped me today.’(395) [Familiar context: A day after, one of her friends asks the girl:][Kinande, P.M.]o=mu-pulisiD=C1-polceo-yoC1-Dema-lia-lu-ir-aSM-TM-come-APPL-FVhayiwherekwehi?Q‘Where did that police officer come from?’10. Thanks to Philip Mutaka and Jack Mutaka for providing me with the Kinande datawhen I needed it. My primary Kinande consultant was Philip but some data were alsoelicited from Jack. It was a very happy coincidence that I happened to work with Philip,the same language consultant who Ljilyana Progovac worked with and whose data led tothe famous paper on Kinande augments as NPIs (Progovac 1993).266In both interpretations the speakers agree to convey the existence of a referentthat satisfies the nominal property. That is, in both cases the Ds convey the beliefthat there is such a police officer who stopped the girl. I turn next to specificity.5.2.4.2 Kinande Ds do not encode specificityThe fact that an overt D may be used in specific contexts such as in (396), and thesame D is also used in non-specific contexts such as (397), rules out the hypothesisthat Kinande augments convey specificity:(396) [Specific context: I wish I had a cup. I would drink from this stream. B:][Kinande, P.M.]e=kopoD=cupyineLKo=moD=C18n-gunzaC9-bage-yoC9-that‘A cup is in that bag.’(397) [Non-specific context: There are several cups on a table, equidistant fromthe speaker] [Kinande, P.M.]m-berer-ay1sg-pass.APPL-FVe=kopoD=C7-cup‘Pass me a cup.’In the next section I show that the relevant notion that forces the different Dchoices in Kinande is existence.5.2.4.3 D distinction in KinandeProgovac (1993) has proposed that Kinande augmentless NPs (DPs with the polar-ity D) should be analyzed as negative polarity items (NPIs). Her evidence comesfrom the licensing environments: negation, interrogatives, and conditional sen-tences, in which the licensing environments are parallel to those for English ‘any.’I relate my current proposal to Kinande D data, and show that the syntactic licens-ing environments discussed in Progovac (1993) also have a reflex in the semanticsof Ds. Throughout this thesis, the following syntax-semantic generalization is up-held11:11. I have slightly departed from Progovac in not treating argument DPs containing thepolarity sensitive D∅ as ‘bare NPs’ or augmentless Ns, as they do have a covert D which hasa semantic function (see Chapter 3 for discussion).267(398) Syntax-semantic generalization for the polarity DThe polarity D∅ must fall under the c-command domain/scope of a non-factual operator (eg., negation, conditional, modal, Q-morpheme, etc.)The crucial difference between the current account and Progovac’s is that inthe current account the non-factual operators are semantically defined but thelicensing is syntactic; in Progovac’s account, these operators are only defined syn-tactically, hence are not correlated to the D interpretive contrast. If we allowsyntax to talk to the semantic component, we see that the interpretive contrast ofDs corresponds to a distinction between overt DPs, which always escape the scopeof a non-factual operator, as in (399), and polarity-sensitive DPs which obligatorilyappear inside the scope of a non-factual operator, as in (400).(399) Overt D: BOE [Kinande, adapt. Progovac 1993: 260]a. o=mu-kaliD=C1-womansinotanzirelikesYohani [DP >¬]John‘The woman does not like John.’b. YohaniJohnsinotanzirelikeo=mu-kali [DP >¬]D=C1-woman‘John does not like the woman.’(400) NEG licenses D∅: Non-BOE [Adapt. Progovac 1993: 260-262]a. sinothalithere-ismu-kaliC1-womanwanzirelikesYohani [¬ >D∅P]John’‘No woman likes John.’b. YohanaJohnsinotanzirelikemu-kali [¬ >D∅P]C1-woman‘John does not like any woman.’if the speaker does not wish to commit to believing in existence, they use the polar-ity D. These D distinctions/semantic contrasts are not explained under Progovac’ssyntactic account. I claim here that if the speaker wishes to convey a belief ofexistence, s/he always uses the overt D, as in (399). If the speaker does not wishto commit to believing in existence, they use the polarity D, (400).268In the absence of these licensing operators, the result is a declarative sentence,which always induces an existential interpretation; hence, the overt D must beused, as in (401). The ungrammaticality of (402) arises because there is no non-factual operator in the sentence to license the polarity D∅, hence violating (398).(401) Overt D: BOE [Kinande, adapt. Progovac 1993: 260]a. o=mu-kaliD=C1-womananzirelikesYohaniJohn’‘The woman likes John.’b. YohaniJohnanzirelikeso=mu-kaliD=C1-woman‘John likes the woman.’(402) No D∅ in declaratives [Kinande, adapt. Progovac 1993: 260]a. * mu-kaliC1-womananzirelikesYohanaJohnIntended: ‘The woman likes John.’b. *YohaniJohnanzirelikesmu-kaliC1-womanIntended: ‘John likes the woman.’Unlike Nata in which licensing takes place before movement, Progovac illus-trates that licensing in Kinande is a surface-oriented phenomenon, which meansthat the overt c-command matches the semantic scope. Progovac’s non-negationdata reveal also that the interpretive contrast obtains with yes/no questions anda conditional operator, which is predicted by my analysis (see Progovac 1993).5.2.4.4 Overt Ds behave as belief-of-existence DsI focus on the core diagnostics for belief-of-existence Ds in which speakers donot assert existence, but commit to existence based on their belief that an entityexists. Unlike in an assertion-of-existence D system, Kinande overt Ds are usedin surmising contexts, as in (403), in possible world contexts with attitude verbslike want or look for, as in (404), as well as in contexts where referents have notmaterialized yet, such as (405).269(403) [Kinande, P.M.]Surmising context: There is a sun-shower. B:a. e=temboD=C9-elephant ya-ama-but-aSM-TM-give.birth-FV‘An elephant is giving birth.’b. * temboC9-elephant ya-ama-but-aSM-TM-give.birth-FVIntended: ‘An elephant is giving birth.’(404) [Kinande, P.M.]Belief of existence in a possible world contexta. a-ka-rond-aSM-TM-look-FVa=ma-sabuD=C6-milk‘He is looking for (some) milk.’b. *a-ka-rond-aSM-TM-look-FVma-sabuC6-milkIntended: ‘He is looking for (some) milk.’(405) [Kinande, P.M.]Non-materialized referentsa. a-kandisya-lung-aSM-FUT-marry-FVnawitho=mu-nduD=C1-personwaofhalifar‘She will get married to someone from far away.’b. *a-kandisya-lung-aSM-FUT-marry-FVnawithmu-nduC1-personwaofhalifarIntended: ‘She will get married to someone from far away.’Note also that overt Ds in Kinande can be used in generics. In (406) thespeaker is not asserting the existence of owls or butterflies, but rather is conveyinga belief that in every possible world with owls, they hunt butterflies.270(406) Generic sentences: overt Ds are required [Kinande, P.M.]a. e=bi-bukuluD=C8-owlsbi-ka-higaSM8-HAB-hunte=bi-nyurugunzuo=C8-butterflies‘Owls (always) hunt butterflies.’b. * bi-bukuluC8-owlsbi-ka-higaSM8-HAB-huntbi-nyurugunzuC8-butterfliesIntended: ‘Owls (always) hunt butterflies.’The hypothesis that Kinande makes D distinctions based on the notion of ex-istence is upheld. We now can extend the current account to cases that Progovacpresents as problematic for her account, and show that such cases are not prob-lematic at all.5.2.4.5 Accounting for residual issues in KinandeOne of the cases that Progovac points out as problematic for understanding thelicensing environments in Kinande is with nominals following affirmative copulas(see also Krifka and Zerbian 2008), as in (407a). Progovac says that the nominalmulimi does not appear with the augment while there is no licensor, i.e., the nom-inal follows an affirmative copula12. I have added the case in (407b) and the dataset in (408) to bring out the argument-predicate contrast.(407) a. YohanaJohnniismulimi [Adapted from Progovac 1993: 267]farmer‘John is a farmer.’b. YohanaJohny.ô=mulimi [Elicited, P.M]is.D=farmer‘John is the farmer.’12. Baker (2003) and Krifka and Zerbian (2008) point out that some copula constructionsin Kinande involve Focus marking. See Chapter 6 for the discussion of Focus marking inBantu.271(408) a. amagenda3sg.PROG.walkngaasmukule [Kinande, P.M.]pregnant‘She walks as if she is pregnant.’b. amagenda3sg.PROG.walkngaaso=mu-kuleD=C1-pregnant‘She walks like the pregnant one.’The syntax-semantics mapping proposed in Chapter 3 comes to the rescue here.These cases are not problematic, as they involve a predicate-argument distinctionconsistent with Longobardi (1994). I analyse mulimi in (407a) and mukule ‘preg-nant’ in (408a) as predicates that denote a property, not an individual; hence theylack the D layer. As Philip Mutaka (p.c) points out, the copula yo in (407b) isinvariable and is a fusion of a copula and the augment; hence, it is best presentedunderlyingly as yó-ò=mulimi, which means the yo-phrase has a DP complement.Thus, the cases in (a) contrast with the individual-denoting argument nominalsin the (b) cases, which appear with argument DPs (see also Schneider–Zioga andMutaka 2014; Baker 2003b).Progovac (1993) also lists the lack of the overt D in the by-phrases of passiveconstructions such as (409) as being problematic for her licensing account. Exam-ple (409b), in which the covert D is licensed, seems to involve a null modal withmeanings must/might. Progovac reports this case with must only. In my elicitation,the Kinande speaker accepted the use of the null D with the might interpretationas well as indicated in (409b).(409) By-phrases in Kinande [Adapted from Progovac 1993: 267]a. e=ki-taboD=C7-bookkyaAGRheribauawas.lostnabyo=mu-kaliD=C1-woman.‘The book was lost by the woman.’b. e=ki-taboD=C7-bookkyaAGRheribauawas.lostnabymu-kaliC1-woman‘The book must/[might] have been lost by a woman.’Nata does not allow the formation of by-phrases of this nature. However, epis-temic modals can license the polarity D∅ in other syntactic structures. In (410a),the strong modal particle n, glossed as a strong assertion marker (SAM) rendering272the must interpretation, is used with the overt D; the overt epistemic modal verbtora ‘may’/‘might’, on the other hand, licenses the polarity D∅ in (410b) (refer toChapter 3 for modal licensing).(410) a. n-a-a-kubh-abfn-a-a-kuβ-aSAM-SA7-PST-be-FVa-bwin-ea-βwin-eSA1-find-FVu=mw-aana [Nata]o=mw-aanáD=C1-child‘S/he must have found a/the child.’b. a-a-gho-tor-aa-a-ɣó-tor-aSA7-PST-INF-might-FVku-bh-aku-β-aINF-be-FVa-bwin-ea-βwin-eSA1-find-FVmw-aanamw-aanáC1-child‘S/he might have found a child.’Both in Nata and in Kinande, the overt D is used when a speaker is indicat-ing that s/he is conveying a commitment to the belief of existence of an entity,(409a)/(410a); when the polarity D∅ is used, the speaker is unsure and thereforeprefers not to commit to the belief of existence of an entity, as in (409b)/(410b).While covert D licensing by modals in Kinande remains an area for futureresearch, licensing of a polarity D by an epistemic modal seems to be a plausi-ble assumption. If we argue that the polarity D in Kinande is licensed by a nullepistemic modal, this may explain why the polarity D is used in (409b). It isnot immediately clear to me whether in Kinande the must modal interpretationcan license the polarity D∅. One language where the must modal interpretationcan be used with the polarity D is St’át’imcets (see Matthewson 1998; 1999). InSt’át’imcets, an assertion-of-existence D system, the meaning contrast expressed byKinande by-phrases is expressed by special evidential clitics which mark speakerknowledge (see Chapter 4 for discussion). Thus, if the speaker does not have per-sonal knowledge about the individual s/he cannot commit to the existence of areferent. In this case the speaker-knowledge clitic k’a ‘surmise’ is used to licensethe non-assertion-of-existence D ku, (411):(411) Speaker knowledge: surmise [St’át’imcets, Matthewson 1998: 204]kʷanən-š-túmkwanen-s-túmcatch-TR-PASSk’ak’asurmisetu?tu7COMPL[kʷu plíšmən][ku plísmen][NON.EXIS.DET policeman]‘He must have got caught by a policeman.’273As Matthewson illustrates, if the speaker has personal knowledge of the refer-ent, the surmise clitic will be dropped and an assertion-of-existence D X...a used.Kinande does not seem to track speaker knowledge; however, the modal contextsin which the polarity D is used are crucial in understanding what forces the differ-ent D choice in (409). This case leaves us with three critical questions for futureresearch: why would by-phrases license a covert modal in Kinande? Why does the‘must’ modal license a null D in Kinande (if it can), and not in Nata? What othermodal contrasts exist in Kinande and what are their correlations with the (c)overtaugments?Finally, during my elicitation, Philip Mutaka, the Kinande scholar and speaker-linguist who was my Kinande consultant, asked about cases such as (412) (not inProgovac’s paper):(412) a. *(o=mú-kali!(D=)C1-woman!umbá2sg.1sg.givekubyo [Kinande, P.M.]some‘Woman! Give me some.’b. mú-kali!C1-woman!umbá2sg.1sg.givekubyosome‘Woman! Give me some.’The (412b) example is not problematic if it is analyzed as a vocative structure.There is cross-linguistic evidence that vocatives tend to appear with no D (see DeBlois 1970; De Dreu 2008; Ndayiragije et al. 2012; Espina 2009; and Chapter 6for further discussion).5.2.5 Existence Ds in XhosaXhosa (IsiXhosa) data seem to support the current semantic proposal that the aug-ment/D encodes the speaker’s belief of existence. I show that the reported D con-trast in Xhosa is consistent with the current proposal. The summary chart belowindicates that Xhosa patterns as a belief-of-existence D system.274Table 5.6: Xhosa Ds and their correlation with other existence DsCore diagnostics AOE BOE XhosaD encodes definiteness 7 7 7D encodes specificity 7 7 7Speaker’s personal knowledge required 3 7 7Used in cultural assumptions 7 3 3Used in possible worlds contexts 7 3 3Used with non-materialized referents 7 3 3As per this chapter’s format, I start with some crucial data to show that theXhosa D distinction is not forced by definiteness or specificity; those features comefrom elsewhere.5.2.5.1 Xhosa Ds do not encode definitenessIn Xhosa, Ds do not contrast for definiteness. The overt D is used in novel contextsas in (413a), but the same D is also used in familiar contexts, (414)13.(413) [Novel/indefinite context: A girl is telling her friends about what hap-pened to her today:] [Xhosa, M.N]a. Yhaziknowyintoni?whatNdi-mis-w-e1sg-stop-PASS-PSTi=ndodaD=9mannamhlanjetoday‘You know what? I got stopped by a man today.’b. *...Ndi-mis-w-e...1sg-stop-PASS-PSTndoda9mannamhlanjetodayIntended: ‘...I got stopped by a man today.’13. Elicited data come from Lutsha Bata and Mandisa Ndlovu. I thank these two for theirtime and passion to share their beautiful language with me.275(414) [Familiar/definite context: A day after, one of her friends follows up.][Xhosa, M.N.]U-yi-khangel-e2sg-OM1-see-PSTi=ndodai=9mane-ku-mis-eSM1-2sg-stop-PSTizolo?yesterday‘Did you see the man who stopped you yesterday?’These data support the argument that the augment/D is not the locus of def-initeness in Xhosa. In the current proposal, the use of the overt D in both noveland familiar discourse contexts is predicted under the assumption that the speakerbelieves that there is a man who stopped her. Next, I show that Xhosa Ds do notcontrast for specificity either.5.2.5.2 Xhosa Ds do not encode specificityThe overt D is used where the speaker has a specific cup in mind, as in (415), andthe same D is used in non-specific contexts, such as (416).(415) [Specific context: B: I wish I had a cup. I would drink from this stream.You:] [Xhosa, L.B.]Kun’Kunathere.isi=nkomityii=nkomityiD=9cupphaphainkwesokwesothatsi-khwama.si-khwamaC7-bag‘There is a cup in that bag.’(416) [Non-specific context: There are several cups on the table. B asks achild:] [Xhosa, L.B.]a. Ndi-nikeNdi-nike1sg-passi=nkomityii=nkomityiD=9cup‘Pass me a cup.’b. *Ndi-nike*Ndi-nike1sg-passnkomityinkomityi9cupIntended: ‘Pass me a cup.’276Data from generics provide further evidence that overt Ds may be used in non-specific contexts. In (417) the speaker may not be talking about a specific groupof owls or butterflies:(417) Overt D in generic contexts [Xhosa, L.B.]a. i=sikhovai=sikhovaD=8owlszi-zingel-azi-zingel-aSM8-hunt-FVa=mabhabhathane.a=mabhabhathaneD=6butterflies’Owls hunt butterflies.’b. * sikhova* sikhova8owlszi-zingel-azi-zingel-aSM8-hunt-FVmabhabhathane.mabhabhathane6butterfliesIntended: ‘Owls hunt butterflies.’These data can be accounted for by the single generalization that the use ofan overt D indicates that the speaker is committing to the belief of existence of areferent/referents. The lack of a specificity contrast in Xhosa challenges Visser’s(2008) position that overt Ds in Xhosa are [+spec]. What is important aboutVisser’s account is the fact that specificity and definiteness can also be introducedin the grammar by other (morpho)syntactic devices, e.g., OMs, demonstratives,etc. I have concluded that augments do not contribute to specificity or definitenessat all. I now turn to show that the D distinction in Xhosa is based on the core notionof existence.5.2.5.3 D distinction in XhosaXhosa data support the current proposal that D choice is forced by the notion ofbelief of existence. In (418a)-(418b), the speaker is committed to the existence ofentities contained in the proposition (i.e., a=bantwana ‘(the) children’) and thebelief-of-existence D is used. When the speaker does not want to commit to exis-tence, the non-belief-of-existence D is used, (418c). (Note non-original glossing)14:14. I abandon the Xhosa scholars’ tradition of using the term ‘augmentless (–A) nominals’for DPs containing a non-polarity D. Carstens and Mletshe (2015) argue that [−A] nominalshave a negative concord feature uNeg which must Agree with a negative licenser iNeg;and in some contexts [–A] nominals are licensed by a [+focus] feature. Since I am notfocusing on Agree relations under Minimalism, I remain neutral about their account. Forall I know their account works fine, like other previous licensing accounts/binding accounts277(418) [Carstens & Mletshe 2015: 262, M.B.)]D Choice in Xhosaa. Ndi-bon-é1sSA-see-CONJ1.PSTa=ba-ntwanaD=2-child‘I saw (the) children.’b. A-ndi-bon-angaNEG-1sSA-see-NEG.PSTa=ba-ntwanaD=2-child‘I didn’t see the children.’c. A-ndi-bon-angaNEG-SA1-see-NEG.PSTba-ntwana2-child‘I didn’t see any children.’d. *Ndi-bon-éSA1-see-CONJ1.PSTba-ntwana2-childIntended: ‘I saw children.’As usual, in (418b) the overt DP takes wide scope over a non-factual operator,here negation, to preserve the existential interpretation. In all cases where thepolarity D∅ is used it must fall within the c-command domain/scope of a non-factual operator, (418c). I have claimed that DPs can only have a non-existentialinterpretation if they are interpreted under the scope of a non-factual operator.As is well known, polarity Ds cannot appear in declarative sentences which lack alicensing operator, (418d). The same D distinction obtains with subject DPs:(419) [Xhosa, M.N.]D choice in Xhosaa. i=ndodaD=10mani-fund-eSM1-read-PSTi=ncwadiD=9book‘A/the man read a/the book.’b. a-khoNEG-bemntu10mano-fund-eSM1-read-PSTi=ncwadiD=9book‘Nobody read a/the book.’do (see Progovac’s 1993; Hyman and Katamba’s 1993). My interest is the syntax-semanticscorrelation of Ds. However, see Chapter 6 for further discussion of restrictions with theFocus parameter.278c. a-khoNEG-bemntu10mano-fund-eSM1-read-PSTncwadi9book‘Nobody read a/any book.’The object DP i=ncwadi ‘a/the book’ in (419b) is associated with a belief-of-existence interpretation, hence, it is interpreted with wide scope in Logical Form.Beyond negation, as data from Carstens and Mletshe (2015) and also Visser(2008) show, the D distinction also obtains with other non-factual operators suchas conditionals, Q-morphemes, and a modal, which is consistent with my analysis.In the interests of full disclosure, in the next chapter I will discuss independentsyntactic constraints that ban DPs with a covert D from appearing in certain syn-tactic configurations in Xhosa15. I finish my discussion here by presenting datawhich show that Xhosa overt Ds indeed behave as belief-of-existence Ds.5.2.5.4 Overt Ds as belief-of-existence DsXhosa overt Ds, like Nata, R/Rukiga, Haya, Luganda and Kinande existence Ds,can be used in contexts in which a speaker does not assert existence. Xhosa overtDs are used when a speaker is only surmising the existence of referents such asmonkeys, as in (420a). Overt Ds can also be used to refer to referents that arefuture possibilities (non-materialized referents), as in (421a).(420) [Xhosa, M.N.][Surmising context: There is a sun-shower. B says:]a. ku-tshat-aku-tshat-a17S-get.married-FVi=inkawui=inkawu.D=10monkeys‘Monkeys are getting married (mating).’b. *ku-tshat-a*ku-tshat-a17S-get.married-FVinkawuinkawu.10monkeysIntended: ‘Monkeys are getting married (mating).’15. These syntactic constraints are known in Zulu and Xhosa as the four taboo positions,which are: (i) pre-verbal subjects in SVO clauses, (ii) the position of applied objects andcausees of transitive expletive constructions (TECs), (iii) direct object positions in mono-transitive TECs, and (iv) right-dislocated position.279(421) [Future referent context: B is considering donating a chair to a newschool. He believes he can find one to buy.] [Xhosa, M.N.]a. Ndi-za-ku-leth-aNdi-za-ku-leth-a1SG-come-INF-bring-FVi=situloi=situloD=7chair‘I will come to bring a chair.’b. *Ndi-za-ku-leth-a*Ndi-za-ku-leth-a1SG-come-INF-bring-FVsitulositulo7chairIntended: ‘I will come to bring a chair.’Finally, overt Ds are used to refer to referents in possible worlds, as seen by thefact that attitude predicates such as funa ‘want’ in (422a) and khangela ‘look(ing)for’ in (423a) cannot license the polarity D∅. In the (a) cases, the speaker believesthat there is a world compatible with his/her belief about the existence of fish ormangoes.(422) [Xhosa, M.N.]BOE in attitude verbsa. NdifunaNdi-fun-a1SG-want-FVukutyaukutyaINFT.eati=nhlanzii=nhlanziD=C9-fishnamuhlanamuhlatoday‘I want to eat fish today.’b. *Ndifuna*Ndifuna1SG-want-FVukutyaukutyaINFT.eatnhlanzinhlanziC9-fishnamuhlanamuhlatodayIntended: ‘I want to eat fish today.’(423) [Xhosa, M.N.]BOE in attitude verbsa. u=Maryu-MaryD=Maryu-khangel-au-khangel-aSM1-PROG.look.for-FVi=mangoi=mangoD=4mangoes‘Mary is looking for mangoes.’280b. *u=Mary*u=MaryD=Maryu-khangel-au-khangel-aSM1-PROG.look.for-FVmangomango4mangoesIntended: ‘Mary is looking for mangoes.’These data show that Xhosa Ds do not track the speaker’s personal knowledge;for example, the speaker does not necessarily know that there are fish availablefor him/her to eat. It is only a belief that fish or mangoes exist. I conclude that Dchoice in Xhosa is necessitated by the speaker’s belief of existence.5.2.6 Existence Ds in ZuluThere are a number of works on the subject of the Zulu augment (see Doke 1992;Mzolo 1968; von Staden 1993 and references therein; De Dreu 2008; Buell 2009;Cheng and Downing 2009; Adams 2010; Halpert 2012; Carstens and Mletshe 2015;to mention but a few). Among these, I follow the syntactic view that the Zuluaugment is a realization of the functional category D (cf. De Dreu 2008; Adams2010; Carstens and Mletshe 2015 and others). (For the view that the Zulu augmentcorresponds to Case, see Halpert (2012), but see Carstens and Mletshe (2015) fora counter proposal, and Chapter 2 for arguments against the Case proposal forNata.) I focus on the syntax-semantic arguments about the D contrast, where Ishow that Zulu data support the current proposal that D choice in Zulu is basedon the core notion of existence. A summary chart showing where Zulu Ds sit inrelation to other existence Ds is given below.Table 5.7: Zulu Ds and their correlation with other existence DsCore diagnostics AOE BOE ZuluD encodes definiteness 7 7 7D encodes specificity 7 7 7Speaker’s personal knowledge required 3 7 7Used in cultural assumptions 7 3 3Used in possible worlds contexts 7 3 3Used with non-materialized referents 7 3 3First, I show that definiteness and specificity are not realized on Zulu Ds.2815.2.6.1 Zulu Ds do not encode definitenessZulu does not switch Ds based on novelty-familiarity (see Adams 2010 for thisclaim). The overt D can be used in novel contexts, (424), and the same D is alsoavailable for use in familiar contexts, as in (425)16.(424) [Novel/indefinite context: A girl is telling her friends about what hap-pened to her today:] [Zulu, M.B.]qagelaqagel-aguess-FVkwenzakaleni!kwenzakale-ni!happened-WH!i=phoyisai=phoyisaD=5policelingivimbileli-ngi-vimb-ile5s-1sg-stop-PSTnamhlanjenamhlanjetoday‘Guess what happened! A police officer stopped me today.’(425) [Familiar/definite context: A day after, one of her friends follows up.][Zulu, M.B]ulibonileu-li-bon-ile2sg-OM5-see-PSTi=phoyisai=phoyisaD=9policeekumisilee-ku-mis-ileREL-2s-stop-PSTizolo?izolo?yesterday‘Did you see the police officer who stopped you yesterday?’What both the novel and familiar reading have in common is that the speakeris willing to commit to the belief of existence of a police officer. Next, I show thatZulu Ds do not force D choice based on the notion of specificity.5.2.6.2 Zulu Ds do not encode specificityZulu also does not seem to distinguish Ds based on the notion of specificity (seeAdams 2010 for this claim). Zulu data show that overt Ds may be used both inspecific contexts, (426), as well as in non-specific contexts, (427). This rules outthe possibility that Ds have a specificity contrast.16. Thanks to my Zulu consultant Mthuli Buthelezi for sharing his language with me whenI needed to fill in some data gaps, and for glossing them.282(426) [Specific context: B: I wish I had a cup. I would drink from this stream.You:] [Zulu, M.B.]KukhonaKukhonathere.isi=nkomishii=nkomishiD=9cupku-lesoku-lesoin-thatsikhwama.sikhwama7bag‘There is a cup in that bag.’(427) [Non-specific context: There are several cups in the cupboard. I ask...][Zulu, M.B.]Ngi-pheNgi-phe1sg-give/passi=nkomishii=nkomishiD=9cup‘Give me a cup.’[Consultant comment: I will bring you one of the cups]Overt Ds can also be used in generics, which obtain a non-specific interpretation:(428) Overt D in generic contexts [Zulu, M.B.]a. i=zikhovai=sikhovaD=8owlszi-zingel-azi-zingel-aSM8-hunt-FVi=zimvemvane.i=zimvemvaneD=8butterflies’Owls hunt butterflies.’b. * zikhova* sikhova8owlszi-zingel-azi-zingel-aSM8-hunt-FVzimvemvane.zimvemvane6butterfliesIntended: ‘Owls hunt butterflies.’These data support the notion that the overt D does not encode specificity.Next, I show that the Zulu D distinction is forced by the notion of belief of exis-tence.5.2.6.3 D distinction in ZuluVon Staden’s (1973) view that overt Ds do not encode definiteness, and that theyare used if a speaker has a ‘particular’ object in mind, is akin to the notion ofspecificity. Von Staden writes:283Nouns with initial vowels denote particular, individual objects... de-pending on the nature of the objects referred to. The term ‘object’ isto be understood to denote anything that can be referred to by a noun.[von Staden, 1973: 165]Von Staden states further that the distinction displayed by Zulu augments in-volves a particularization feature contrast, [+IND]/[-IND], which cannot be ren-dered by any English grammatical devices. Von Staden writes that the overt D isloosely represented using the English articles a(n) or the coupled with the rendi-tion particular, as in (429a); its counterpart can be rendered with no, any, nothing(not anything) like, (429b).(429) a. [Zulu, von Staden 1973: 166]A-ku-fik-angaNEG-17s-arrive-NEG.PSTa=bahambiD=2travellers‘No (particular, individual) travellers arrived’Lit: ‘There did not arrive (particular, individual) travellers (e.g., trav-ellers whom the speaker knows would have come, or travellers re-ferred to previously).’b. A-ku-fik-angaNEG-17s-arrive-NEG.PSTbahambi2travellers‘No (nothing like) travellers arrived.’Data with object DPs from De Dreu (2008) show a similar D contrast:(430) a. a-ka-limaz-aNEG-SA1-hurt-FVa=bantwana [Zulu, de Dreu 2008: 18]D=2children‘He doesn’t hurt (some particular) children.’b. a-ka-limaz-aNEG-SA1-hurt-FVbantwana2children‘He doesn’t hurt any children.’I re-analyze the particularization/individualization contrast, [+IND]/[-IND],that Von Staden posits for Zulu augments as a D distinction based on the corenotion of existence. More precisely, Ds in Zulu encode the speaker’s belief of ex-istence. The D contrast based on the notion of existence is supported by the datagiven above. In the (a) cases, the speaker agrees to commit to the existence of an284entity; therefore, the DPs containing an overt D escape the scope of the non-factualoperator to render the existential interpretation. Where the speaker does not wishto commit to the existence of travellers/children, as in the (b) cases, the polarityD is used, in which case it is interpreted under the scope of the non-factual oper-ator. The licensing condition of the polarity D is consistent with Halpert’s (2012)argument that ‘augmentless nominals’ (here, DPs containing polarity Ds) must bevP internal where they are licensed by a L(icenser). I have claimed in Chapter 2that the non-factual operators are the overt realization of Halpert’s abstract headL (cf. De Dreu 2008; Buell 2009; Adams 2010; Carstens and Mletshe 2015).Declarative sentences containing no non-factual operators induce existentialinterpretations and they are compatible with belief of existence Ds, as in (431a).In such sentences there is no licensor to license the polarity Ds; hence, (431b) isungrammatical17.(431) [Zulu, M.B.]Overt D is required in declarativesa. a=bahambiD=2travellersba-a-fik-aSA2-PST-arrive-FV‘(The) travellers arrived.’b. * bahambi2travellersba-a-fik-aSA2-PST-arrive-FVIntended: ‘(The) travellers arrived.’Halpert (2012; 2015) and Carstens and Mletshe (2016) report a number of syn-tactic constraints that ban the licensing of arguments in certain syntactic positions.I postpone the discussion of some of these constraints until Chapter 6.No data are available showing licensing with conditionals or modals in Zulu18.Nonetheless, in interrogatives, the question operator may also license the polarityD. A representative case comes from yes/no questions, (432b):17. Von Staden (1973: 165) claims further that nouns lacking augments (polarity Ds) “referto objects in indefinite or generic contexts.” I do not adopt the use of the term ‘generic’for polarity Ds for two reasons: (i) it is inconsistent with a standard definition of genericity(Carlson and Pelletier 1995; Krifka 2003; Hawkins, 1978; Déchaine et al., 2014); (ii) OvertDs in Zulu are used in generic contexts (see ex. 428) which contradicts von Staden’s Ddistinctions.18. Carstens and Mletshe (2015) argue that Zulu [−A] nominals (i.e., containing polaritysensitive DPs, in my account), behave as antimorphic forms (e.g., (super) strong NPIs or n-words), like the ‘broad’ English NPI ‘any’ forms. Based on this, they say that Giannakidou’s(1998, 2011) non-veridical licensing environment would be too broad for Zulu.285(432) [Von Staden 1973: 1967]Zulu, D choice in polar questionsa. N-azi2sg‐knowu=muntuu=1personnga-zwiINSTR‐5.voicel-a-khe?5‐POSS‐1pron‘Do you know a/the (particular, individual) person by his voice?’b. N-azi2sg‐knowmuntu1personnga-zwiINSTR‐5.voicel-a-khe?5‐POSS‐1pron‘Do you know anyone by his voice?’I now turn to showing that Zulu Ds are comparable with belief-of-existence Dsand do not behave as assertion-of-existence Ds.5.2.6.4 Overt Ds as belief-of-existence DsZulu overt Ds may be used in contexts where an assertion of existence does nothold. Zulu overt Ds are used in surmising contexts in which the speaker is onlyconveying a belief, given cultural assumptions, that there is such a thing as mon-keys mating out there in the rain, (433). Overt Ds are also used when talkingabout non-materialized referents, e.g., to convey a belief that a chair donation isa possibility in the future, (434):(433) [Zulu, M.B.][Surmising context: There is a sun-shower. B says:]a. ku-shad-aku-shad-a17S-marry-FVi=zimfenei=zimfeneD=10monkeysnamhlanjenamhlanje.today‘Monkeys are getting married (mating) today.’b. *ku-shad-a*ku-shad-a17S-marry-FVzimfenezimfene10monkeysnamhlanjenamhlanje.today‘Monkeys are getting married (mating) today.’(434) [Future referent context: B is considering donating a chair to a newschool. He believes he can find one to buy.] [Zulu, M.B.]286a. Ngi-zo-leth-aNgi-zo-leth-a1SG-FUT-bring-FVi=sihlaloi=7sihlaloD=7chair‘I will bring a chair.’b. *Ngi-zo-leth-a*Ngi-zo-leth-a1SG-FUT-bring-FVsihlalo7sihlalo7chair‘I will bring a chair.’Finally, belief-of-existence Ds can be used with attitude verbs such as funa ‘lookfor’/‘want’/, (435a) to refer to referents in possible worlds.(435) [Zulu, M.B.]Overt D is required with attitude verbsa. u-fun-au-fun-a3sg-want/look.for-FVu=mangou=mangoD=mangoes‘S/he is looking for mangoes.’b. *u-fun-a*u-fun-a3sg-want/look.for-FVmangomangomangoesIntended: ‘S/he is looking for mangoes.’In (435), replacing the overt D with the polarity D will result in ungrammat-icality, which indicates that attitude predicates do not license the polarity D inZulu. I wish now to discuss certain data presented in the previous literature onZulu as posing a challenge for the semantic analysis of Zulu Ds.5.2.6.5 Accounting for residual cases in ZuluPotential obstacles for the current proposal are cases presented in the previousliterature as posing a problem for a semantic analysis. I discuss these cases below.Von Staden (1973) and Buell (2009) present the case in (436) with the Dcontrast involving the verb bika ‘report’ and the numeral one; the problem is thatthere seems to be no licensor for the polarity D:287(436) [Zulu, Von Staden 1973: 169-170]D choice in reportativesa. Zi-bik-e10sm-report-PERFi=zwiD=5messagelinye5one‘They have reported one message.’b. Zi-bik-e10sm-report-PERFzwi5messagelinye5one‘They have reported one message.’The contrast expressed by (436) involves the lexical evidential verb bika ‘re-port’, which is a non-factual operator that can license the polarity D∅ similar tothe verb -bugha ‘report(edly)’ in Nata (see Chapter 3). The meaning of the proposi-tion in (436) may vary depending on the reliability of the information source (seeAikhenvald 2004; Waldie 2012; Matthewson 2007; Déchaine et al. 2016; and oth-ers). My analysis predicts that, in (436a), the speaker has trust in the informationsource and will agree to commit to the belief of existence of a message. Further-more, if the speaker does not trust the source/non-first-hand information, hencedoes not wish to commit to the existence of a message, s/he will use the polarityD, as in (436b)19. While more research may help to arrive at a solid conclusionabout the interpretive difference between (436a) and (436b), it seems that whatforces D choice is whether the speaker believes in the message or not.Another potentially problematic case is (437) which involves the universalquantifier zonke ‘all’, presented in Halpert (2012) (originally from Buell 2008).The problem Halpert points out is that the right-dislocated quantifier zonke ‘all’in (437) takes low scope with respect to negation, an environment that usuallylicenses NPIs. However, the DP set ranged over by universal quantifiers cannot bean empty set; hence, the use of the DP with an overt D follows straightforwardly.Based on Jocken Zeller (p.c) and the Zulu speakers I have worked with, quan-tification over a specific set is another possible reading for (437); hence, I haveincluded this possibility by adding ‘(the)’ in the gloss20.19. Buell (2009) and Carstens and Mletshe (2015) propose a Focus-based analysis in whichall stressed augmentless nominals are focussed and their unstressed counterparts are not.I do not rule out this possibility; however, I am more interested in what forces differentD choices and not only the syntactic distribution of nominals. Here, I will keep the Focusdiscussion in the background, but see Chapter 6 for further discussion of Focus.20. Thanks to Jocken Zeller and Leston Buell for pointing this out. The set ranged overby the quantifier is contextually defined, as the speaker is not referring to children in the288(437) Scope and c-command [Zulu, Adapt. Halpert 2012: 96]i=zi-nganeD=C10-childrena-zi-thand-iNEG-SM10-love-NEGa=ma-swidi]VPD=C6–sweetszonkeall‘Not all children like sweets.’ ¬ >8, *8 >¬3‘Not all (the) children like sweets.’The use of the overt D is consistent with the current account in that (universal)quantifiers always presuppose the existence of their range (the DP cannot denotean empty set) (cf. von Fintel 1994; Matthewson 1998, 2001; Heim 1982; andothers); hence, the DP contains the belief-of-existence D. Thus, the sentence istrue if there is at least one child who does not like candy. The speaker uses theovert D as s/he believes in the existence of a set of children but it is not the casethat all (the) children in this set like sweets. I conclude that the use of the overtD/belief of existence D on the DP quantified over, is consistent with the claim thatZulu overt Ds always encode speaker’s belief of existence.The last potentially problematic case concerns the NPI reading of object-marked DPs used in negative environments21. Object markers (OMs) usually havea familiarity reading; thus under the current account, we would predict an objectDP co-occuring with an OM to have an existential interpretation22. However, inthe famous Zulu elephant example, (438a), Zeller (2008)) argues that speakers ac-cept both the NPI and the familiar interpretations, while Halpert (2012) presentsthe DP with the NPI reading only I didn’t see any elephants. To argue for my case Ihave added the context of use in which I re-elicited the sentences below. The newdata reveal a problem for the claim that overt Ds used in familiar contexts or withOMs in Zulu can yield an NPI interpretation. Note that ] marks dislocated objects.In (438a) the first meaning is the literal meaning of the utterance which en-tails the second/implied meaning. Zeller indicates that in (438b) the OM and thepolarity D∅ are incompatible; the NPI interpretation clashes with the anaphoricinterpretation of the OM, as my analysis predicts. (438c) shows that the DP withthe D∅ must be interpreted under the scope of negation to obtain a non-existentialinterpretation.entire universe (cf. Geurts 1998; von Fintel 2004; Matthewson 1998, 2001; Gillon 2006;and others).21. A case that could also be presented here as a residual problem is sentence fragmentanswers with augmentless nominals/DPs with a polarity D∅ such as muntu ‘Nobody!’, lutho‘Nothing!’, which appear to have no overt licensor. I postpone the discussion of this caseuntil Chapter 6 which deals with parametric variation.22. But see Chapter 6, for a discussion of languages that impose certain constraints on thefamiliarity reading of object marked DPs, i.e., animacy, humanness, wh-questions etc.289[Context: It’s about the familiar story of two elephants that were taken to the zooin Boston. You visit the zoo in Boston and find out the elephants actually died.Now you are back in South Africa. B asks: Did you see the elephants?].(438) Implied or literal?[Zulu, adapt. Zeller 2008: 11; Halpert 2012: 97; M.B]a. A-ngi-zi-bon-anga]NEG-1S-OM10-see-NEG]i=zindlovuD=10elephant‘I didn’t see (them), the elephants.’ [Literal]‘I didn’t see any (of the) elephants’. [Implied]b. *A-ngi-zi-bon-anga]NEG-1S-OM10-see-NEG]zindlovu10elephantIntended: ‘I didn’t see (them), the elephants.’c. A-ngi-bon-angaNEG-1S-see-NEGzindlovu10elephant‘I didn’t see any elephants.’ [Literal]‘I didn’t see even the two elephants.’ [Implied]The Zulu speaker I worked with associated the use of the overt D in (438a) withbeing ‘straight to the point’, and answering using the polarity D∅, as in (438c),as being ‘rude’ or answering the question in ‘a roundabout’ way. Halpert (2012)documents this judgemental effect in reporting on NPI readings with OM construc-tions:I will note here that though the existing literature on Zulu describesthe omission of an augment vowel [the use of a polarity sensitive D∅(J.G)] as a general NPI strategy in the language... These speakers[Durban Zulu speakers] describe augmentless NPIs as being informalto the point of rudeness [Halpert 2012: 90].We may solve the frequently recorded NPI reading in definite contexts suchas (438a) by considering, in line with the intuitions of my Zulu consultant, thatsuch meanings are implied, not literal (i.e., do not come from the semantics ofthe overt D). Cross-linguistically, in familiar/presuppositional contexts, speakersmay use the polarity element by ‘flouting’ the maxim of Quantity, or may chooseto obey the cooperative principles of conversation by using the relevant D (here290the overt D) (see Grice 1975; Levinson 1983 for the discussion of implicatures).Zulu appears to be doing the opposite from Nata regarding the implied vs. literalmeaning in (438a), which uses the overt D to be direct to the point. We sawthat in Nata, speakers used the polarity D∅ to avoid being specific/rude, as theirstrategy of expressing politeness, whereas for Zulu speakers (438c) is a rude andinfelicitous way of responding the question Did you see the elephants? (see Brownand Levinson 1987; Bousfield 2008 on politeness; also refer to §5.2.2.5 for thediscussion with examples from both Nata and Haya).Halpert’s context of use which led her to record A-ngi-zi-bon-anga] i=zindlovu‘I didn’t see any elephants’, here the implied meaning under (438a), is the questionHow many elephants did you see in Boston? Using the same contexts, my Zulu con-sultant responded as in (438a) on the condition that one must be inquiring aboutthe specific (kinds of) elephants (i.e., the elephants sent from Krugar National Parkto Boston which, say, died there). (438b) was consistently rejected; the reason isclear: the DPs used with OM must convey an existential interpretation (see Adams2010). In addition, the Zulu speaker agreed that they would only answer using(438c) in contexts where Boston is a land with no elephants. I conclude that theuse of overt Ds in these cases is consistent with the underlying semantic principlethat derives overt Ds in Zulu: overt Ds encode the speaker’s belief of existence.This goes against the hypothesis that Zulu Ds are semantically vacuous (contraHalpert 2012). While this seems to be a valid conclusion, a more detailed empir-ical study on these issues may be something to pursue in future to fully supportthis position.Having discussed the belief-of-existence D systems, I turn my attention to theassertion-of-existence D systems.5.3 Assertion-of-existence D systemsHere I discuss Bemba, a language that appears to have elements of an assertion-of-existence D system like St’át’imcets.5.3.1 Existence Ds in BembaBemba augments/Ds seem to behave like assertion-of-existence Ds in St’át’imcetswith one exception, that Bemba speakers have/believe in cultural assumptionswhich are not easily found in St’át’imcets. We can summarize the different prop-erties of Bemba Ds across different semantic diagnostics as shown below:291Table 5.8: Bemba Ds and their correlation with other existence DsCore diagnostics AOE Bemba BOED encodes definiteness 7 7 7D encodes specificity 7 7 7Speaker’s personal knowledge required 3 3 7Used in cultural assumptions 7 3 3Used in possible worlds contexts 7 7 3Used with non-materialized referents 7 7 3I first show that Bemba Ds do not encode definiteness and specificity before Idiscuss their existence-encoding properties.5.3.1.1 Bemba Ds do not encode definitenessGivón (1970) states explicitly that Bemba Ds do not contrast for (in)definiteness:...the [VCV/CV] distinction may in English translation converge withthe contrast between def./indef. articles, but this convergence is atbest partial and many times misleading. [Givón 1970: 37]Consistent with Givón, Bemba data show that Ds are not switched based on novel-familiar semantic spaces (cf. Hawkins 1978; Heim 1982; 2011 Schwarz 2009;2012; and others). The overt D in u=mwaume ‘a/the man’ appears in novel con-texts such as (439a), and the same D appears in familiar contexts, as in (440)23:(439) [Novel/indefinite context: A girl is telling her friends about what hap-pened to her today:] (Bemba, M.C.)a. Nawishiba?2sg.knowu=mwaumeD=mwaumeacinjiminikaSM1.PST.1sg.stoplelotoday‘Do you know? A/#the man stopped me today.’23. Thanks to Mr. Musonda Chilengwe, my Bemba language consultant, for sharing hislanguage with me, and for providing me with extra data to fill in the Bemba data gapswhen needed.292b. *Nawishiba?2sg.knowmwaumeC1.manacinjiminikaSM1.PST.1sg.stoplelotodayIntended: ‘Do you know? A man stopped me today.’(440) [Familiar/definite context: A day after, one of her friends asks the girl:](Bemba, M.C.)a. BusheQwalimumonaSM1.PST.OM.seeu=mwaumeD=C1-manwakwiminike?SM1.REL.2sg.stop.PST‘Did you see (him) #a/the man who stopped you?’b. *BusheQwalimumonaSM1.PST.OM.seemwaumeC1-manwakwiminike?SM1.REL.2sg.stop.PSTIntended: ‘Did you see (him) #a/the man who stopped you?’The use of the overt D in both novel and familiar contexts rules out any anal-ysis claiming that the Bemba D system encodes definiteness. Under the currentaccount, the above DPs are consistent with the fact that the speaker is convey-ing the existence of a referent corresponding to the NP; hence, the overt D inu=mwaume ‘a/the man’ is used.5.3.1.2 Bemba Ds do not encode specificityBemba data reveal that overt Ds may be used in specific or non-specific contextsand as such they do not contrast for specificity. In the examples below, the overtD is used in specific contexts such as (441a), but the same D can also feature innon-specific contexts, (442a):(441) [Specific context: A: I wish I had a cup. I would drink from this stream:][Bemba, M.C]a. MuliThere.isi=nkomakiD=C9.cupmuliincilyathati=colaD=C5.bag‘There is a cup in that bag.’b. *MuliThere.isnkomakiC9.cupmuliincilyathatcolaC5.bagIntended: ‘There is a cup in that bag.’293(442) [Non-specific context: There are several cups on a table, equidistant fromspeaker] [Bemba, M.C.]a. MpelakoPass mei=nkomakiD=C9.cup‘Pass me a cup.’[Consultant reaction: I will pass you any]b. *MpelakoPass menkomakiC9.cupIntended: ‘Pass me a cup.’Analyzing Bemba augments/Ds as existence Ds explains why the overt D isused in a variety of contexts. I consider more data that support this contention.5.3.1.3 D distinction in BembaGivón (1978) proposes the notion ‘referentiality’, which I analyze as being anexistence distinction, to capture the overt D/polarity D∅ contrast in Bemba:...the lexicon makes no provision for the definite-indefinite distinctionbut only for that of referential vs. non-referential. The VCV form [theovert D (J.G)] marks referential nouns, while the corresponding CVform of the prefix [the polarity D∅ (J.G)] marks non-referential ones.[Givón 1978: 300]The core notion of existence accounts for the empirical facts concerning the Ddistinction in Bemba. For instance, the data in (443a) show that when a speakerwants to convey the existence of a referent, the overt D must be used. Note that ifa non-factual operator is used, the DP containing the overt D always escapes thescope of the non-factual operator in order to render the existential interpretation,as in (443b)24.(443) Overt D encodes existence [Bemba, adapted from Givón 1978: 301]a. u=mu-anaD=C1-childa-a-someneSM1-PST-readi=ci-taboD=C7-book‘A/the child read a/the book.’24. Givón argues that, except for generics, subjects in Bemba behave as definite DPs, par-ticularly when used in negative contexts. Nonetheless, in my elicitations, both definite andindefinite readings are possible with subjects.)294b. u=mu-anaD=C1-childt-a-a-someneNEG-SM-PST-readi=ci-tabo [DP >¬]D=C7-book‘A/the child did not read a/the book.’If the speaker does not wish to convey existence of a referent, the overt D mustbe replaced by the polarity D∅, (444a). The polarity D always must fall under thec-command/scope of a non-factual operator. (444b) is ungrammatical because itis being used in a declarative context which cannot license the D∅:(444) Licensing of Polarity D∅ [Bemba, adapted from Givón 1978: 301]a. u=mu-anaD=C1-childt-a-a-someneNEG-SM-PST-readci-tabo [¬ >D∅]C7-book‘A/the child did not read any book.’b. *u=mu-anaD=C1-childa-a-someneSM1-PST-readci-taboC7-book*‘The child read any book.’The contrast in terms of existence also obtains with adversative predicates/lex-ical negation which involves a non-factual operator. When the speaker means anexistential interpretation, the overt D is used, and the DP has wide scope withrespect to the lexical negation, as in (445a); however, if s/he does not believe inthe existence of a child that J. saw, the overt D must be replaced by the polarityD∅, as in (445b).(445) Existence Ds [Bemba, adapted from Givón 1970: 43; 2018: 16]a. ndeetwiishika1sg.doubtngathatJ.JaamweeneSM1.PST.seeu=mw-aanaD=C1-child[Overt D >L.Neg]‘I doubt that J. saw the child.’b. ndeetwiishika1sg.doubtngathatJ.JaamweeneSM1.PST.seemw-aanaC1-child[L.Neg >D∅]‘I doubt that J. saw any child.’These data are consistent with the diagnostics for a speaker-oriented existenceD system: (i) the D contrast is based on the notion of existence, and (ii) the D sys-tem is speaker-oriented as opposed to a speaker-hearer system. Next, I show that295the Bemba D system behaves like the assertion-of-existence D system in St’át’imcetsexcept for cultural assumptions which may receive a separate explanation.5.3.1.4 Overt Ds as assertion-of-existence DsBemba Ds seem to me to be an assertion-of-existence D system with a possibleabsence of evidentials and slightly different cultural beliefs about what counts asstrong enough evidence to assert existence. The claim that the Bemba augment sys-tem behaves like assertion-of-existence D is consistent with Matthewson’s (1998)observation:Givón makes the cross-linguistic claim that nominals falling under thescope of a possible modality or negative modality may receive eithera referential (i.e. assertion of existence) or non-referential interpreta-tion; otherwise all nominals are interpreted referentially (1978: 294).This is the case in Bemba as well as in Salish. [...]Givón’s definition isbased on Bemba (Bantu), whose determiner system shows similaritywith Salish systems. [Matthewson 1998: 55, 69]Givón (1970; 1978) analyzes the Bemba Ds as encoding ‘known’ vs ‘unknown’distinctions. Using generic data, Givón (1970) illustrates that when the speakeruses an overt D as in (446) s/he may not be talking about individually knownreferents but rather about a known/specific genus.(446) Specific references in generics [Bemba, adapted from Givón 1970: 47]a. i=m-bwaD=C9/10-dogniCOPnamaC9.animal(i) ‘Dogs are animals.’(ii) ‘The/these dogs are animals.’b. i=m-bwaD=C10-dogsshilalyaSM10.HAB.eati=namaD=C10.meat(i) ‘Dogs eat meat.’(ii) All the dogs in the universe eat meat.’Givón argues that (446a) is “ambiguous with respect to known/unknown, butboth interpretations are specific” (Givón 1970: 47). The conclusion he reaches isthis: ”Bemba indeed judges all subject nouns to be obligatorily specific”, (Givón1970: 48). We already saw that Bemba overt Ds are not specific, eg., referents296such as i=nkomaki ‘a cup’ in (442a) above are not necessarily specific. Accord-ing to Givón, the subject noun phrases in (446) lack a pure generic interpreta-tion and may not be considered as ‘real generics.’ This is similar to St’át’imcets.Matthewson argues that when translating characterizing statements like Bears eathoney, St’át’imcets speakers do not talk about generic bears, rather they refer tosome bears they know. The parallels in generic sentences between Bemba andSt’át’imcets may suggest that a lack of real generics is a diagnostic for assertion ofexistence. However, it is not important that generics be absent for a language tobe an assertion-of-existence system. As Lisa Matthewson (p.c), suggests, as longas a language does not have obligatory deictic Ds, generics should be fine (referto Chapter 4 for the discussion concerning some implications of deictic features).The crucial property of assertion of existence Ds is that the Ds encode speaker’spersonal knowledge, as we saw in St’át’imcets. While the data contrasting for re-ported vs speaker’s personal knowledge is unavailable, there is evidence whichstrongly suggests that Bemba is leaning towards an assertion-of-existence D sys-tem. This consists of data involving Ds with non-materialized referents and inpossible worlds.In Bemba, if the speaker intends to convey the existence of a referent, the overtD will be used with future tense/irrealis modality, as shown in (447a). However,if the speaker does not wish to commit to the existence of a referent, the futuretense/irrealis modality will license the D∅, as in (447b).(447) Licensing of Polarity D∅ [Bemba, Givón 1970: 43; 2018: 16]a. n-ka-mona1s-FUT-wantu=mu-ana [Overtu=C1-childD >FUT]‘I will see a/the child.’b. n-ka-mona1s-FUT-wantmu-ana [FUT>D∅]C1-child‘I will see a child.’Bemba is comparable to the St’át’imcets assertion-of-existence system in this re-spect, since in the Nata belief-of-existence D system, future tense never licensesthe polarity D with non-materialized referents (see Chapter 4).As Givón illustrates, non-factual operators such as attitude verbs translated aslook for or want can also license the D∅, as in (448b). However, if the speaker is297willing to commit to the existence of the relevant DP contained in the proposition,an overt D is always used, (448a).(448) Attitude verbs [Bemba, adapted from Givón 1978: 301]a. u=mu-anaD=C1-childa-a-fwayaSM-PST-wanti=ci-tabo [Overt D >Op]D=C7-book‘A/the child wanted a/the book.’b. u=mu-anaD=C1-childa-a-fwayaSM-PST-wantci-tabo [Op >D∅]C7-book‘A/the child wanted a book (be it any).’The one exception we find in Bemba to what is expected for an assertion-of-existence system is that it does not allow covert Ds to be used in surmising contexts,as is the case with the St’át’imcets non-assertion-of-existence D ku (see Chapter 4).The fact is that in Bemba, overt Ds are obligatorily used in surmising contexts, as(449)-(450) show:(449) [Surmising context: I suddenly have an itchy palm. I say:][Bemba, C.M.]a. Nilapokelel-a1SG.FUT.receive-FVi=fi-sumaD=C8-thing‘I will receive something (good).’b. *Nilapokelel-a1SG.FUT.receive-FVfi-sumaC8-thing‘I will receive something (good).’(450) [Surmising context: Suddenly someone has an itchy foot. S/he says:][Bemba, C.M.]a. a=beniD=C2-visitorsba-liC2-aremu-nshilaLOC18-way‘Visitors are on the way.’298b. * beniC2-visitorsba-liC2-aremu-nshilaLOC18-way‘Visitors are on the way.’(449a) and (450a) are felicitous in a situation where the speaker only hopes/be-lieves that s/he will receive something good/visitors. This slightly differs fromSt’át’imcets in which a speaker cannot use an assertion-of-existence D in contextswhere existence is only believed.It is an open question why Bemba Ds appear to track the speaker’s personalknowledge but at the same time allow overt Ds to be used in surmising contexts inwhich existence is not asserted but believed (the speaker doesn’t know for sure).We may only speculate about the reasons. One speculation may be that Bembaand St’át’imcets just differ in the way speakers process cultural assumptions. Arebeliefs considered as more realities in some cultures than in others? Framing thequestion differently, do Bemba or Bantu speakers more generally consider culturalassumptions as sufficient for assertion? (see Gambarage and Matthewson 2019).I will not try to answer this anthropological question here. Another speculationmay be that perhaps the Bemba language, unlike Salish languages, does not havean evidential or deixis system that interacts with D, hence allows overt Ds to bealso used in beliefs. I leave this issue open for further research. The last case Iturn to is Dzamba which encodes definiteness in its D system.5.4 Ds that do not encode existenceI now present Dzamba data to show that it is the only exception among the lan-guages I discuss here in not encoding existence. The D distinction in Dzamba isbased on novelty–familiarity (definiteness), consistent with Bokamba (1971) andGivón (1978, 2018).5.4.1 Lack of Existence Ds in DzambaDzamba is a true example of an augment language whose D distinction isnot based on belief/assertion of existence; rather, it is based on the novelty-familiarity/definiteness contrast (cf. Bokamba 1971; Givón 1978, 2018). Thechart below summarizes the results for the Dzamba system in relation to the belief-of-existence and assertion-of-existence systems:299Table 5.9: Dzamba Ds and their correlation with existence DsCore diagnostics AOE BOE DzambaD encodes definiteness 7 7 3D encodes specificity 7 7 7Speaker’s personal knowledge required 3 7 7Used in cultural assumptions 7 3 7D encodes existence 3 3 7I show that the major differences between Dzamba and the existence D systemslie in the fact that existence Ds do not access the hearer’s knowledge. This meansthe Dzamba system has a positive setting for the Common Ground Parameter (cf.Matthewson 1998 on this parameter), and other contrasts such as specificity orexistence are completely obscured. I present Dzamba data to show that Dzambais the only exception among the languages I discuss herein that does not encodeexistence.5.4.2 Dzamba Ds contrast for novelty-familiarityThe choice between different Ds in Dzamba is based on the novel-familiar distinc-tion. In (451b), mooto ‘man’ is novel; therefore, the null D is used. If the samenoun is used in familiar contexts, the null D must be switched to the overt D as in(452a)25:(451) [Novel/indefinite context: You meet a man on the way and he stopsyou to ask something. Back home you tell your friend what happened:][Dzamba, E.G.B.]a. #W-eelebi?2sg-knowo=mo-toD=C1-persona-n-tem-y-akiSM1-1sg-stop-CAUS-IMPFlɔɔmetoday#‘You know (what)? The man stopped me today.’25. Thanks to Eyamba Georges Bokamba (E.G.B) who agreed to share his language withme and for providing me with extra data and assisting with the morphological glossing.300b. W-eelebi?2sg-knowmo-toC1-persona-n-tem-y-akiSM1-1sg-stop-CAUS-IMPFlɔɔmetoday‘You know (what)? A man stopped me today.’(452) [Familiar/definite context: A day later, one of your friends inquiresabout the same man:] [Dzamba, E.G.B.]a. W-eena-ki2sg-see-Qo=mo-toD=C1-persono-wa-ko-tem-y-akaREL-SM1-2sg-stop-CAUS-PSTyanayesterday‘Did you see the man who stopped you yesterday?’b. #W-eena-ki2sg-see-Qmo-toC1-persono-wa-ko-tem-y-akaREL-SM1-2sg-stop-CAUS-PSTyanayesterday#‘Did you see a man who stopped you yesterday?’Overt Ds are ruled out in surmising contexts and in generics. Only referentsthat are familiar to the speaker and hearer can take an overt D:(453) [Surmising context: B is eating and s/he bites his/her tongue. S/he says:]a. #a=ba-toD=C2-peopleba-andoSM2-somelo-o-ni-tongo [Dzamba, E.G.B.]COP-SM2-2sg-talkIntended: ‘Some people are talking about me.’b. ba-toC2-peopleba-andoC2-somelo-o-ni-tongoCOP-SM2-2sg-talk‘Some people are talking about me.’(454) Generic contexts: null D required [Dzamba, E.G.B]a. #a=ba-ziD=C2-womenba-ne-tenaSM2-HAB-chopn.konyiC10.woodIntended: ‘Women (customarily) chop wood.’b. ba-ziC2-womenba-ne-tenaSM2-HAB-chopn.konyiC10.wood‘Women (customarily) chop wood.’301Note that if the referent is in the common ground of the discourse, the speakermust choose to use the overt D. Compare (454a) with (455a):(455) [Contexts: The host is inquiring about the two women visitors. Child B:]a. a=ba-ziD=C2-peopleba-ki-tenaSM2-PROG-chopn.konyiC10.wood‘The women are chopping wood.’b. # ba-ziC2-peopleba-ki-tenaSM2-PROG-chopn.konyiC10.woodIntended: ‘The women are chopping wood.’[Consultant comment: ”You can say this if you mean some women.”]These examples fit Bokamba’s (1971) analysis of Dzamba Ds as encoding definite-ness. However, I slightly depart from Bokamba’s additional claim that Dzamba Ds(Noun Pre-prefixes (NPPs), his term) are also specific.Givón has shown, for Bemba, that the CV/VCV contrast is not a mat-ter of [-DEF]/[+DEF], but rather of [+SPEC]/ [-SPEC]. In Dzamba,however, ...it is not only definitiveness but also specificity (referen-tiality) that is here marked by the NPP. [Bokamba 1971: 230/235]I argue that Dzamba Ds are compatible with specific contexts but the D systemdoes not encode (non)specificity.5.4.3 Dzamba Ds do not contrast for specificityBokamba (1971) analyses augments in Dzamba as Ds which encode definiteness orpresupposition of existence, in which case he assumes that they also encode speci-ficity (referentiality). Bokamba argues that the D distinction in Dzamba makesthe following contrasts: [+DEF]/[-DEF]/[+SPEC]/[-SPEC], where there is nopossibility of [+DEF, –SPEC]. He exemplifies these contrasts with the followingexamples:(456) Definiteness contrasts in Dzamba [Dzamba, Bokamba 1971: 220]a. a=mo-ibiD=C1-thief(*mɔɔ)(*one)anyɔlɔklenteredondaku [+DEF, +SPEC]in.the.house‘The thief (*one) entered the house.’302b. mo-ibiC1-thief(mɔɔ)(one)anyɔlɔklenteredondaku [–DEF, +SPEC]in.the.house‘A thief (one) entered the house.’c. mo-ibiC1-thief(mɔɔ)(one)akokicanna-nyɔlɔkland-enterondaku [–DEF, –SPEC]in.the.house‘A thief (one) can enter the house.’According to Bokamba (1971), the difference between the [+SPEC] and [–SPEC] features of moibi ‘a thief’ in (456b) and (456c), respectively, is that the verbnanyɔlɔkl in (456b) is “in the past tense, asserting that the event described hasalready taken place, hence the the subject is referential/specific” (p.220). This isopposed to (456c) which makes no such assertion; hence, the existence of a thief is‘not presupposed.’ As the same covert D marks both specificity and non-specificityin (456b) and (456c), respectively, this rules out the possibility that Dzamba Dscontrast for specificity.Data involving non-factual operators such as negation, adversative predicates,conditionals, and question morphemes show that the novel-familiar contrast holds.I discuss data with negation as representative of licensing cases. (457a) shows thatthe overt D is consistently interpreted as definite. The Dzamba covert D is used inindefinite contexts and in polarity contexts as shown in (457b,c), respectively26.(457) Augment contrast in Dzamba [Dzamba, Bokamba 1971: 220]a. o=mo-ibiD=C1-thief(*mɔɔ)(*one)ta-nyɔlɔklnot-enteredondakuin.the.houseembanot‘The thief (*one) did not enter the house.’ [+DEF, +SPEC]b. mo-ibiC1-thief(mɔɔ)(one)ta-nyɔlɔklnot-enterondakuin.the.houseembanot‘A thief (one) did not enter the house.’ [–DEF, +SPEC]c. ta-nyɔlɔklnot-enternàandmo-ibiC1-thief(mɔɔ)(one)ondakuin.the.houseembanot‘No (single) thief entered the house. ’ [–DEF, –SPEC]26. Note that there are two negation morphemes ta- and emba in Dzamba which constitutea single negative entity, similar to the French ne...pas, except that in Dzamba the secondelement is always sentence final.303Bokamba correctly points out that in (457a,b), the DPs have wide scope overnegation, while in (457c), the D which is [–DEF, –SPEC] takes low scope. I arguethat this distinction correlates with definiteness.I agree with Bokamba’s empirical claim that the overt Ds cannot be non-specific. Bokamba’s assumption is that overt Ds are definite/specific while covertDs can be specific or non-specific. However, if the covert D can be specific ornon-specific, then this obscures the specificity contrast between the overt D andthe covert D, which is also the case in English with a.My argument against specificity marking comes from the fact that the overtD is ruled out in specific indefinite contexts, (458a); and the null D is neutral forspecificity as it can be used in specific contexts, as in (458b), and in non-specificcontexts, as in (459b).(458) [Specific context: A: I wish I had a cup. I would drink from this stream.B says:] [Dzamba, E.G.B.]a. #e=kɔbɔD=C7.cup(yɔ)(one)e-ndo-o-bo-kumbeCOP-LOC-D=C14-plastic.bagIntended: ‘A/one cup is in the plastic bag.’b. kɔbɔC7.cup(yɔ)(one)e-ndo-o-bo-kumbeCOP-LOC-D=C14-plastic.bag‘A/one cup is in the plastic bag.’[Context: There are several cups on a table, equidistant from speaker](459) Non-specific referents: null D is required [Dzamba, E.G.B.]a. #o-n-kominy-el-e2sg-1sg-pass-APPL-SUBJVe=kɔ-bɔe=C7-cup(yɔ)(one)Intended: ‘Pass me a/one cup’b. o-n-kominy-el-e2sg-1sg-pass-APPL-SUBJVkɔ-bɔC7-cup(yɔ)(one)‘Pass me a/one cup’The specific contexts in which the Dzamba overt D in (458a) is infelicitous,would allow a belief/assertion-of-existence D to be used. Given that the Bembaovert D is ruled out in the same contexts, I conclude that Bemba Ds do not contrast304for existence, but rather for novelty-familiarity. Thus Bokamba’s arguments aboutdefiniteness, but not specificity, carry forward.5.5 Summary, remarks and conclusionIn this chapter I have presented empirical evidence to show that the core seman-tics that forces D choice in various Bantu languages is encoding existence. We sawthat existence is not encoded uniformly across languages: a language like Bembaencodes existence by asserting it (i.e., involving speaker’s personal knowledge ofa referent, more similar to St’át’imcets), while other languages (Nata, R/Rukiga,Haya, Luganda, Kinande, Xhosa and Zulu) encode existence rather weakly, interms of a belief (not evidence) that entities exist in the world of discourse.In encoding speaker-oriented existence, a distinction is made between DPswhich obligatorily fall under the scope of a non-factual operator, and those whichare scopally inert, i.e., they always escape the scope of non-factual operators.Extending Matthewson’s (1998) feature compatibility in the assertion of exis-tence D system, we can modal feature compatibility of the speaker-oriented exis-tence D systems discussed here as:(460) Feature compatibility in AOE/BOE D systemsAOE/BOE Ds+Specific+Definite –Definite–Specific–DefiniteNon-AOE/BOE Ds–Specificǀ–DefiniteThis means speaker-oriented existence Ds are compatible with a wide range ofsemantic features. The striking difference between the overt Ds compatible with[–def, –spec] contexts and the covert augment occurring in [–def, –spec] contextsis that the latter must scope under a non-factual operator, in which case it mustbe interpreted under the restriction of the operator.Of the languages I discussed, Dzamba is the only Bantu language that encodesthe novel-familiar D distinction, similar to English. I concluded that the existenceDs in other Bantu languages do not encode the familiar-novel distinction as isthe case with English and Dzamba, as speaker-oriented existence Ds are used in305various contexts. The results of the cross-Bantu discussion are summarized in thetables in Tables 461-463. The shaded regions indicate that it is impossible to usethe D with the non-existential interpretation in familiar contexts.(461) D distinctions in definite systemsEnglish novel familiarexistential interpretation a thenon-existential interpretation aDzamba novel familiarexistential interpretation covert D overt Dnon-existential interpretation covert D(462) D distinctions in AOE systems adapted from Matthewson (1998: 56)St’át’imcets novel familiarassertion of existence X...a X...anon-assertion of existence kuBemba novel familiarbelief of existence overt D overt Dnon-belief of existence covert D∅(463) D distinctions in BOE systemseg., Nata novel familiarbelief of existence overt D overt Dnon-belief of existence covert D∅306We see that English, by utilizing a/the vs any, can optionally display the ex-istence D contrast. This is the case with Dzamba overt/covert Ds also. However,English and Dzamba Ds do not encode speaker-oriented existence. In English, bothany and a can be used in positive statements such as in FCI and in specific contexts.The Dzamba covert D can be used both in existential and non-existential contexts,hence the existence contrast in these languages is disguised. The reason for thedisguise may be that the definite systems utilize both definiteness and existence,but only encode definiteness in their system (see Matthewson 1998).Using the Common Ground Parameter (Matthewson 1998; 1999) we can dividethese languages into a subset-superset relation. For instance, the speaker-orientedsystems (existence D systems) cannot access hearer knowledge hence they are asubset of English/Dzamba which can access such a distinction.While the notion of existence seems to be pertinent to the D systems of theselanguages, a number of constraints appear to disguise this contrast, hence maycause one not to see the semantic predictability we discussed here. In the finalchapter, I discuss certain augment-related constraints and argue that they can beexplained by language independent constraints.307Chapter 6Locus of Variation in Bantu6.1 IntroductionBantu augment languages display a number of (morpho)syntactic and semanticconstraints which may be thought to obscure the semantic analysis of D. The goalsof this chapter are three-fold: (i) to explore the locus of parametric variation withrespect to constraints on the realization of augments, (ii) to present data held bysome Bantuists to be a problem for a semantic analysis and show how far the pro-posed analysis can extend to such cases, (iii) to identify cases that pose a puzzlefor the analysis developed here. I will argue that certain cases held by some Ban-tuists to be a problem to a semantic analysis are non-problems. I will also addresscases that pose a challenge to the analysis developed here. Such constraints mustbe handled independently of the interpretive component. For instance, in someZone J languages like Luganda, stressed DPs mark Focus and appear as augment-less (Hyman and Katamba 1993; Carstens and Mletche 2015; Cheng and Downing2009; and others). This may indicate that the non-use of an overt D has acquiredsome grammatical functions other than lack of commitment to existence as wesaw in Chapter 5. Such DPs are reducible to independent syntactic variation.The chapter is organized as follows: In §6.2 I address parametric variationof D at the DP level. The discussion centres around how the augment as a Dcan compete for the same structural position with other Ds. In §6.3 I discussvarious DP level constraints and show how they interact with the realization ofthe augment as D. In §6.4 I consider the various constraints beyond the DP-level(i.e., DP-external) such as locality, surface c-command, clause restrictions and the308Focus constraint which prohibits the use of overt D in some Zone J languages. §6.5addresses the syntactic-semantic level constraints, specifically, object markers andanaphoricity. The last section, §6.6, suggests areas for future research and presentsmy conclusion.6.2 DP-internal constraintsHere I discuss the parametric variation of the D position with respect to the positionof demonstratives (DEMs), the universal quantifier translated as ‘every’, partitiveelements in weak quantifiers or other modifiers, and locatives. This discussionwill inform us about the Bantu DP structure and show that the use of an overt Dis consistent with my proposal.6.2.1 D and demonstrativesThe lack of augments in certain structures where they are expected has been takenas evidence for the lack of a semantic contrast/predictability of augments (cf. Hy-man and Katamba 1993). The problem is that in some languages, the augmentmay be absent when the pre-nominal DEM is used, which makes it appear that theaugment is not necessary for an existential interpretation. My analysis predictsthat all argument nominals occurring with a DEM will yield an existential inter-pretation, given that DPs appearing with demonstratives are definite descriptions(see Visser 2008; Allen 2014). Some relevant examples are in order.Examples from Zulu (464), Nata (465), and R/Rukiga (466) show the commonsyntactic positions for Bantu DEMs: pre- and post-nominal (see Wald 1973; deDreu 2008; Du Plessis and Visser 1992; Allen 2014). While Zulu and Nata haveonly two DEM positions (pre-nominal and post-nominal), R/Rukiga has three DEMpositions1:(464) Demonstrative positions in Zulu [Zulu, de Dreu 2008: 27]a. leC9.thisn-dodaC9-man‘this man’b. u=m-faziD=C1-womanloC1.this‘this woman’1. Allen (2014) argues that the DEM in on R/Rukiga is composed of the initial vowel/DEMroot (DEMrt) and the class marker ki. I will not discuss deictic features (proximal, medialor distal) associated with DEMs in these languages. See Allen (2015), Wald (1973); Visserand Du Plessis (1992), and Chapter 2 for the discussion of Nata.309(465) Demonstrative positions in Nataa. u-nọ=mu-kariu-nɔ=mú-kari1-DEM=1-woman‘the/this woman’b. o=mu-kario=mu-káriD=1-womanu-nọú-nɔ1-DEM‘this woman’(466) Demonstrative positions in R/Rukiga [R/Rukiga, Allen 2014: 181]a. [Post-nominal]ekia-ki-ØDEMrt-7-PROXkitaboki-tabo7-book‘this book.’b. [Pre-nominal]e=kitaboe=ki-taboD=7-bookekia-ki-ØDEMrt-7-PROX‘this book.’c. [Pre-nominal]ekia-ki-ØDEMrt-7-PROXe=kitaboe=ki-taboD=7-book‘this book.’I argue that there is no augment for the pre-nominal cases in (a) in (464)-(466)and that the D position is filled with other D material (DEM). I extend the viewthat the pre-nominal DEM occupies the same position as the augment/D; this cor-responds to the structure in (467a) (see de Dreu 2008; Carstens and Mletshe 2016;Van de Velde 2005; and others)2. For the post-nominal DEM cases that follow DPswith an overt D in the (b) cases in (464–466), DEMs seem to be heading their ownphrasal projections (DemPs) (see Giusti 1994, Alexidou et al. 2007; de Dreu 2008;Guardiano (2012); Cinque 2005; Windsow 2015; and others). Thus, in the overtsyntax, leftward movement of the DP has taken place. In the spirit of the Uni-versal Base Order Hypothesis (UBOH) (Greenberg 1963; Hawkins 1983; Cinque1995, 2005; Aboh 1999), modifiers are merged in the same hierarchical position2. Since the DEM has an internal structure where it agrees in number with the head noun, Ileave open the question how the whole structure can possibly occupy a D slot (see Déchaineand Wiltschko (2002) for insights.310universally: Dem > Num > A > N, and variation in word order is a result ofsimple and constrained leftward movement of N, in our case NP/DP3 (see Green-berg 1963; Hawkins 1983; Cinque 1995, 2005; Aboh 1999; Valois 1991; Carstens2008; Longobardi 2001; Carstens 2000; Kayne 1994; Ajiboye 2005; Takano 2003;Windsow 2015; Lewis 2016; and many others). Based on this, I propose that inthe (b) cases, the DP started out lower and moved to specifier of DemP as in thestructure in (467b) (see Windsow 2015; Lewis 2016).(467) a. DPDAUGMENT/DEMONSTRATIVEφPmu-kari‘woman’b. DemPDPi DemPDem0 tiThe rare R/Rukiga structure in (466c), in which the pre-nominal DEM precedesthe DP containing the overt D, can be derived by assuming that the DemP-DP wordorder corresponds to the Merge order consistent with the Universal Base OrderHypothesis, which reinforces the notion that syntax is anti-symmetric, thereforeno symmetric order of modifiers in the Merge order (Dem > Num > A > N) ispossible (see Greenberg 1963; Hawkins 1983; Cinque 2005; Kayne 1994):(468) (Greenberg 1963: 87)Greenberg’s Universal 20 (U20)When any or all of the items (demonstrative, numeral, and descriptiveadjective) precede the noun, they are always found in that order. If theyfollow the noun the order is either the same or its exact opposite.Thus, while some languages allow the co-occurence of DEM and D, other lan-guages like Nata do not. This variation is explained by language independent syn-tactic constraints. In all the cases where we get the overt D, the belief-of-existenceinterpretation also obtains as we saw in Chapters 4 and 5 for these languages.6.2.2 D and modifiersIn many Bantu languages, some modifiers such as weak quantifiers (few, many, two,etc.,), adjectives, and modifiers translated as ‘other’ or ‘certain’ seem to appear in3. The fact that what moves here is a phrase and not a head N challenges an assumptionabout head/N movement (see Greenberg 1963; Hawkins 1983).311different syntactic positions within the DP consistent with Cinque (1995, 2005;and later works). In some languages modifiers may appear post-nominally, whilein other languages, we get the mirror image: a modifier appears pre-nominally butafter an augment. In the early Bantu literature (see De Blois 1970, for instance),such nominals present a problem for the syntactic analysis of the augment. I talkabout such structures to argue that such nominals are actually DPs. I propose thatthese variations should be explained within the Universal Base Order Hypothesisstated above. I discuss the variation in syntactic positions of modifiers using -nde/ndi ‘other’, as an example, where I show that nominal arguments obligatorilyrequire a D as my analysis predicts.In Nata, modifiers such as -nde/ndi ‘other’ appear with D, where they createD–doubling structures. In Chapter 3, we saw that D–doubling structures are as-sociated with a partitive reading (i.e., they pick out a subset from a familiar/pre-viously mentioned set). Thus, the non-D-doubling structure in (469a) contrastswith the D-doubling structure in (469b). Notice the order of the DP with respectto (a=)bha-nde is the same in both cases.(469) Partitive by D–doubling [Nata]a. a=bha-anaa=βa-anáD=C2-childbha-ndeβáa-ndeC2-other/certainbha-ka-het-aβa-ka-het-aSA2-PST-pass-FVha-nọha-nɔC16-here‘Some/certain kids passed here.’b. a=bha-anau=βa-anáD=C2-childa=bha-ndea=βaa-ndéD=C2-otherbha-ka-het-aβáa-ndeC2-other/certainha-nọβa-ka-het-aSA2-PST-pass-FV‘(The) other kids passed here.’I propose that (469a) corresponds to the DP structure in (470a), which involvesDP internal modification, and (469b) corresponds to the DP structure in (470b),which is the appositive structure, with DP2 adjoined to DP1.312(470) a. DP1Da=φP1φP1bha-ana‘kids’φP2bhande‘certain’b. DP1DP1a=bhaana‘kids’DP2a=bhaande‘the other’While Nata DPs containing -nde/ndi ‘other’ modifiers exhibit an invariableword order, in other languages, the partitive vs. non-partitive interpretations arerendered by different word orders. For example, in Kirundi, a modifier appearssandwiched between the augment/D and the head noun, (471a), obtaining theDeterminer–Modifier–NP word order. But the modifier can also follow the nounas in (471b). (471) is the famous example from Kirundi which appeared first inMeeussen (1959) and in Hyman and Katamba (1993: 216). Similar examples in(472) come from Kinande.(471) D position with -ndi ‘other’ [Kirundi, adapted from H&K 1993: 216]a. a=báandiD=2other/certainbaantu2people‘(The) other people’b. a=baantuD=2peoplebáandi2other/certain‘Some/a certain people’(472) D position with -ndi ‘other’ [Kinande, H&K 1993: 216]a. o=gundiD=2other/certainmúndu1person‘(The) other person’b. o=múnduD=1persongundi1other/certain‘Some/a certain person’313Considering the flexibility in word order for the cases in (471)–(472), I adoptan analysis involving N-to-D movement for the (b) cases (see de Dreu 2008;Carstens 2008; and others). Based on the Universal Base Order Hypothesis, the (a)examples in Kirundi and in Kinande correspond to the tree in (473a), which showsthat the noun is base generated. (472b) corresponds to the structure in (473b), inwhich N-to-D movement has taken place4.(473) a. DPDo=NPXPgundi‘other’mundu‘person’b. DPDDo=Xmundu‘person’NPXPgundi‘other’tiAs De Blois (1970) observes, several languages allow the Determiner–Modifier–NP word order, including Rwanda, R/Rukiga, and Kinande.6.2.3 D and personal pronounsAnother area in which Bantu languages seem to differ, and sometimes raise ques-tions about the D status on argument nominals, is with respect to the D require-ment with personal pronouns. The examples from Zulu show that a pronoun mayreplace the augment/D, as in (474):(474) Personal pronouns in Zulua. [Zulu, de Dreu 2008: 22]thinaPN1Pmadoda6.men‘We, men’b. [Zulu, Halpert 2012: 130]thinawebantu2peopleabampofuREL2.poor‘We, poor people’4. This version of N-to-D movement is different from Longobardi’s in which movement isbased on complementary distribution between the D and the modifier, and N has to replacethe empty D slot, see §.6.7.1.3.314Furthermore, von Staden (1973) shows that there are two possible configura-tions for personal pronouns in Zulu, either with or without the D. According to vonStaden, the contrasting data in (475) is evidence for the individuation [+IND]/[–IND] distinction. I argued in Chapter 5 that what von Staden calls the individu-ation feature is typically a D contrast forced by belief-of-existence. Here, the DPu=mfundi ‘a/the student’ in (a) has an existential interpretation, but in (b) mfundi‘student’ is a predicate.(475) [Zulu, adapted von Staden, 1973: 168]Personal pronouns in Zulua. naandmi,I,u=mfundi,D=1studentbe-ngi-bonaPAST.IMP-1SG-see‘I, a/the student, was also seeing.’b. naandmi,I,mfundi,1studentbe-ngi-bonaPAST.IMP-1SG-see‘I (in my capacity as) a student was also seeing.’Nata obligatorily requires an augment with DPs occurring with a personal pro-noun. The pronoun can occur either pre-nominally, (476a), or post-nominally(476b), but in either position it will be ungrammatical without a D, (476c).(476) Absolute pronouns in Nataa. itweitwewea=bha-subhea=βa-suβeD=C2-man‘We men’b. a=bha-subhea=βa-suβeD=C2-manitweitwewe‘We, men’c. *(itwe)*(itwe)(we)bha-subheβa-suβeC2-man(itwe)(itwe)(we)Intended: ‘We, men’315The Zulu nouns appearing without a D in (474) and (475b) are not bare nouns,as de Dreu (2008) proposes. The pronoun is a D; hence, the D position in (477a) isfilled by the pronoun (see Postal 1966; Déchaine and Wiltschko 2002; Bhat 2007;and others). Halpert (2012) has a similar opinion that the pronoun is an extendedprojection of a(n) NP/predicate, although she does not explicitly state that it is D5.We can analyze the Nata cases in (476a,b) and the Zulu case in (475a) as formingan appositive structure, as in (477b).(477) a. DPDthina‘we’NPmadonda‘men’b. DPDPitwe‘we’DPa=bhasubhe‘(the) men’It is important to stress that the pronoun in Zulu can function as D, taking theNP as its complement, a situation that is missing with Nata pronouns.6.2.4 D and the universal quantifier ‘every’Augment languages also differ with respect to the position of the universal quan-tifier glossed in English as every. In Haya, Luganda and similar languages, theuniversal quantifier is in complementary distribution with a D, as in (478a). InHaya and Luganda the forms are identical:(478) a. [Haya/Ganda, de Blois 1970: 128]bulieverymu-ntuC1-person‘Every person’b. *bulieveryu=mu-ntuD=C1-personIntended: ‘Every person.’The contribution of QPs with a quantificational D to the existence-based the-ory developed here is that (strong) quantifiers presuppose existence, which entails5. Recall that Halpert does not treat Zulu augments as D but merely as case markers; seeChapter 2 for discussion.316speaker’s commitment to existence (cf. Matthewson 1998 for discussion). Syntac-tically, the D slot is filled with the quantifier buli, as in (479); hence, it takes thesame structural position as the augment (cf. Barwise and Cooper 1981; von Fintel1994; Van de Velde 2005).(479) DPDbuli‘every’NPmuntu‘person’In Nata, the quantifier occurs in post-nominal position only, and it alwaysco-occurs with the D, (480):(480) a. [Nata]u=mw-aanau=mw-aanáD=C1-childw-ọọsẹw-ɔɔsɛ ́C1-all‘Every child.’b. *w-ọọsẹ*w-ɔɔsɛ ́C1-allu=mw-aanau=mw-aanáD=C1-childIntended: ‘Every child.’c. *(w-ọọsẹ)*(w-ɔɔsɛ)́(C1-all)mw-aanamw-aanáC1-child(w-ọọsẹ)(w-ọọsẹ)(C1-all)Intended: ‘Every child.’Finally, Zulu seems to allow the quantifier to co-occur with the D where it mayappear pre-nominally, (481a), or post-nominally (481b), as in Nata:(481) a. [Durban Zulu, Halpert 2012: 39]wonkeeveryu=mu-ntuD=C1-person‘Everyone/every person.’317b. u=mu-ntuD=C1-personwonkeevery‘Everyone/every person.’The syntactic analysis proposed for Luganda and Haya above in which the Qsits in D does not work for languages that do not allow quantificational Ds likeNata and (Durban) Zulu. For these cases we may argue that the quantifier adjoinsto DP, rather than taking NP as a complement. Thus, the surface word orders ofthe Nata and Zulu QPs in which the Q follows the DP (examples (480a)/(481b))can be represented as in (482a). The Zulu surface order in which the quantifierprecedes the DP can be presented as in (482b), where the quantifier is realized onthe left.(482) a. QPDPD NPQb. QPQ DPD NPIn the Zulu and Nata cases, a Q does not replace D; this differs from the situa-tion with the Luganda/Haya quantificational Ds in (478a) above. Since the quan-tifiers always presuppose the existence of their range, belief-of-existence denota-tions in Zulu and Nata follow straightforwardly. Treating quantifiers as adjunctsmay explain the different word orders for Qs in (482). I am not concerned in thisthesis with how different word orders are arrived at, but it seems to be the casethat N movement is not required for these cases, as the cartographic approacheswould suggest (see Cinque 1995, 2005; and others).6.3 DP-external constraintsIn this section, I talk about some syntactic constraints that ban Ds in certain syn-tactic positions. The locus of syntactic variation seems to surround the issues listedin Table 6.2. I will show that these constraints do not operate in Nata, and I arguethat these variations are reducible to independent syntactic variation.318Table 6.1: Parametric variation in the syntaxSyntactic variation on Ds Nata Other Bantu languagesDP with D∅ as fragment answer 7 3Surface c-command 7 3Locality constraint 7 3Clause-mate requirement 7 3Focus marking requirements 7 3I discuss each of these cases to show which cases are accounted for by theanalysis I proposed here and which ones present a puzzle and hence need furtherresearch. I will start with sentence fragment answers.6.3.1 Sentence fragment answersCarstens and Mletshe (2016) discuss Zulu augmentless nominals or negativewords/n-words such as lutho ‘nothing’, muntu ‘nobody’ and ndawo ’nowhere’ thatcan be sentence fragment answers. An illustrative example is given in (483). Theproblem with these cases is that there appears to be no licensor for the polarityD∅.(483) Q: Who did you see? [Zulu, Adapt. C&M 2016: 772]A: muntu!muntu!1person!‘Nobody!’Carstens and Mletshe also discuss data from Xhosa showing that Xhosa doesnot allow n-words to be sentence fragrance answers. This is the case with Nata; aDP containing a D∅ is licensed by an overt NEG6.6. Other Zone E languages like Kuria use an enclitic negation that licenses the polarity D∅(data is from the author as I speak Kuria natively).(i) Q: Who did you see?A: mo-nto=he!mó-nto=he!C1-person=NEG!‘Nobody!’319(484) Q: Who did you see? [Nata]a. A: taa-nyi moo-to!taa-nyí móo-to!NEG-be C1-person!‘Nobody!’b. A: * moo-to!* móo-to!C1-person!‘Nobody!’Carstens and Mletshe analyze such Zulu nominals as having an intrinsic neg-ative force like n-words in Catalan, Italian, French, and English. They argue thatNPIs in Zulu are not a uniform class, i.e., some have external licensors (see Chapter5), some do not. However, it is also possible to propose, under the current syn-tactic analysis, that the covert Ds are licensed by a null negation in the elided vPstructure consistent with approaches to n-words cross-linguistically (cf. Vallduvi1994); Déprez 2000; Giannakidou 2006; Zeijlstra 2004; and others. Semantically,the answers to fragment sentences do not pose any challenge; their interpretationis consistent with the view that D∅ conveys that the speaker is not committing tothe belief of existence of a referent, as I argued in Chapters 4 and 5.6.3.2 Surface c-commandMost syntactic accounts of augment licensing consider surface c-command as thenorm for licensing DPs containing a polarity D∅. The data from Nata challengethis assumption. Recall the c-command definition (Reinhart 1976; Chomsky 1981)I have adopted for Nata in this thesis based on the fact that a non-factual operatoralways c-commands the D∅ and not vice versa:(485) α asymmetrically c-commands β iff:a. The first binary-branching node that dominates α also dominates β.b. Neither α nor β dominate each other.c. α c-commands β but β does not c-command α.The default word order in Bantu affirmative/declarative sentences is SVO. This isexemplified in the Kinande, Luganda, and Zulu examples below:320(486) SVO affirmative sentencesa. o=mu-kaliD=C2-womananzireSA1-likeYohani [Kinande, Progovac 1993]Yohana‘The woman likes John.’b. o=mu-kaziD=C1-womana-yagalaSA1-likeJohn [Luganda, E.A.]John‘A/the woman likes John.’c. u=mu-ntuD=C1-personu-pheka1s-cooki=qanda [Zulu, M.B.]D=C5.egg‘A/the person is cooking an/the egg.’However, this changes immediately to NEG-initial word order when we introducea NEG operator, in which case NEG c-commands the polarity D∅ in the overtsyntax. For the cases in (a) in (487)-(489), the subject DP containing the polar-ity D∅ is vP-internal where it is being licensed by NEG. This means that for thepolarity D∅ to be licensed on subjects, negation has to be sentence-initial7. Theungrammaticality of the cases in (b) shows that a D∅ cannot appear in the pre-NEG environment (cf. Kinande: Progovac 1993; Luganda: Hyman and Katamba1993; Zulu: Adams 2010; Cheng and Downing 2009; Halpert 2012):(487) [Kinande, adapt. Progovac 1993: 260]Licensing of D∅ with NEGa. SiNEGhalithere-ismu-kaliC1-womanwanzirelikesYohaniJohn‘No woman likes John.’b. * mu-kaliC1-womansiNEG(hali)there-isanzirelikesYohaniJohnIntended: ‘No woman likes John.’7. There are different word order cases in these three languages. NEG is part of the mainverb in Zulu, but in Kinande and Luganda, it is part of a light verb. See Baker (2003) fol-lowing Doke (1955), for a discussion about verb raising to T in Zulu and related languages,which might explain how the surface word order obtains.321(488) [Luganda, E.A.]Licensing of D∅ with NEGa. Te-waliNEG-there-ismu-kaziC1-womana-yagalaSA1-likeJohnJohn‘No woman likes John.’b. * mu-kaziC1-womante-waliNEG-there-isa-yagalaSA1-likeJohnJohnIntended: ‘No woman likes John.’(489) [Zulu, adapt. Halpert 2012:164; M.B.]Licensing of D∅ with NEGa. a-ku-pek-angaNEG-17s-cook-NEG.PASTmuntu1personi=qandaD=C5.egg‘Nobody cooked an egg.’b. * muntu1persona-ku-pek-angaNEG-17s-cook-NEG.PASTi=qandaD=qanda‘Intended: Nobody cooked an egg.’Looking at Nata, a different picture emerges. The SVO affirmative sentencein (490a) presents two surface possibilities if changed to a negative sentence, (b)and (c). Now, the surface c-command principle can account for the licensing ofthe polarity D∅ in (490b), but not the one in the pre-verbal position in (490c),which is banned in Kinande, Zulu, Luganda and many other languages:(490) [Nata]Neg-Aux-SVO vs S-Neg-Aux-VOa. o=mu-kario=mu-kário=C2-womana-seegh-irea-sééɣ-ireSA2-likeYohanajohanáJohn‘A/the woman likes John.’b. taa-nyihọtaa-nyíhɔNEG-there-ismu-karimu-káriC2-womana-seegh-irea-sééɣ-ireSA2-likeYohanajohanáJohn‘No woman likes John.’322c. mu-karimu-káriC2-womantaa-nyihọtaa-nyíhɔNEG-there-isa-seegh-irea-sééɣ-ireSA2-likeYohanajohanáJohn‘No woman likes John.’In fact, Nata allows the NEG-VSO structures only with the negative existentialcopula. As the ungrammaticality of (491b) shows, when they occur with a mainverb only, DPs containing the polarity D∅ must be in pre-verbal position in theovert syntax (491a), which is the opposite of Zulu, Kinande, and Luganda:(491) S NEG-V O vs *NEG-V SO [Nata]a. mu-karimu-káriC1-womanta-seegh-ireta-seegh-iréNEG-like-PFVYohaniYohanáJohn‘No woman likes John.’b. *ta-seegh-ire*ta-seegh-iréNEG-like-PFVmu-karimu-káriC1-womanYohanaYohanáJohnIntended: ‘No woman likes John.’The surface c-command principle does not work at all for Nata. The proposalthat unifies all these structures cannot be surface syntactic c-command but ratherwhat I have argued for in this thesis, namely, semantic scope. That is, the D∅is interpreted under the scope of a non-factual operator where it yields a non-existential interpretation. Semantic scope requires c-command but not in the overtsyntax in Nata. I have argued for Nata that licensing happens before Spell-out. Inother languages licensing happens after Spell-out (cf. Halpert 2012 and Buell 2008on this for Zulu). Thus, while in (490b) the subject is in in situ, in (490c) the pre-verbal subject has raised from Spec, vP, where Bantu subjects are generated, to ahigher position in the overt syntax (see Baker 2003; Koopman and Sportiche 1991;Ngonyani 1998; Carstens 2001, 2005; Halpert 2012; and many others).6.3.3 LocalitySome Bantu languages require a 1:1 correspondence between licensors andlicensees (DPs with a D∅), with the licensee being the highest DP in a sentence.323This restriction does not play a role in the grammar of Nata. This fact is evidencedby the possibility of multiple DPs containing a D∅ all licensed by a singlenegation, as in the following example repeated from Chapter 3:[Context: You hear a mentally confused person hallucinating saying A man isteaching kids some language for a visitor. You get outside to calm him down, cor-recting his belief:](492) mo-subhemo-súβeC1-mant-a-kw-eegh-er-it-a-kw-eeɣ-éer-iNEG-SA1-teach-APPL-FVmu-ghenimú-ɣeniC1-visitorbha-anaβa-anáC2-childki-ghambọkí-ɣambɔC7-language‘No man is teaching any kids any language for any visitor.’¬[9xyzq [man(x) & PL.kid(y) & language(z) & visitor (q) & x is teaching yz for q]]Data from various augment languages show that more than two polarity D∅scan be licensed by a single licensor. Representative examples come from Kinandeand Luganda:(493) [Kinande, P.M.]Multiple DPs with D∅ are OKsihaliNEG.there.ismu-kaliC1-womana-kangirirayaSA1-taughtmw-anaC2-childyoofmu-bugeC3-languageokoajilibecauseyaofmu-galimuC1-teacher‘There is no woman who taught a child for a teacher.’[Consultant comment: Here child and teacher are also not specific.](494) [Luganda, E.A.]Three DPs with D∅ are OKTewaliNEG.there.ismu-kaziC1-womanyawagavemw-anaC1-childyenaanyki-ntuC7-thing‘No woman gave any child anything.’324While Nata, Kinande, Haya and Luganda appear not to be sensitive to a 1:1licensing correspondence between the non-factual operator and the polarity D∅,R/Rukiga does appear to be sensitive to this constraint. The surprising behaviourof R/Rukiga is that it can only allow licensing of just one D∅ which must be thehighest argument of the vP inside the clause. As the examples in (495) show, anyremaining argument DP after licensing, must occur with the overt augment forthis syntactic reason. Note that in (495c), among the two object DPs, it is the IOargument that gets licensed, not the DO8.(495) [R/Rukiga, A.A.]Locality: Only the highest D∅ in vP alloweda. TiyaareetaTi-a-a-reet-aNEG-1s-PST-bring-FVkintuki-ntu7-thing‘S/he didn’t bring anything.’b. TihaineTihaineNEG-there-ismu-shaijamu-shaija1-manorikukundao-rikukundaREL-likes*(o=)mukazi*(o=)mu-kaziD=1-woman‘No man likes any/a/the woman.’c. JohnJohnJohntarashomiireta-ra-shom-i-ire NEG-PAST-read-APPL-IMPRFmuntumu-ntu 1-person*(e=)kitabo*(e=)ki-taboD=7-bookLit: ‘John did not read for anyone any/a/the book.’Halpert (2012) reports a similar constraint in Zulu: she found restrictions onthe distribution of augmentless nominals inside vP when the number of nominalsin vP outnumbers the number of licensers (refer to Chapter 2). Other Bantuistsseem to link the kind of syntactic constraint observed here with Focus licensingpositions (cf. Hyman and Katamba 1993; Carsten and Mletshe 2016; and §6.4.4for discussion). More research is needed to test a wide range of constructions tosee how far the licensing can go. I propose that the R/Rukiga restrictions and sim-ilar restrictions in other languages (eg., Zulu as illustrated by Halpert 2012; also8. Asiimwe Allen (p.c) indicates that even adding the domain widener ona ‘any’ on thenon-local arguments would not save the constructions if the arguments contained DPs witha polarity sensitive D∅. Allen (2014) mainly focussed on mono-transitive sentences; notmuch has come to light on this subject.325Carsten and Mletshe 2016) should be explained by independent syntactic con-straints.6.3.4 Clause-mate restrictionsSome languages seem to impose clause-mate restrictions on DPs with a D∅ in em-bedded clauses in negative contexts. Carstens and Mletshe (2016) report data fromZulu and Xhosa showing that object DPs containing a D∅ cannot appear in negativecontexts where they are separated from negation by an indicative clause bound-ary. Consider the data below from Xhosa, (496a), and Zulu, (497a). Accordingto Carstens and Mletshe, if the indicative clause is changed to a subjunctive one9,the Xhosa sentence in (496b) and the Zulu sentence in (497b) are grammatical:(496) [Xhosa, adapted from C&M 2016: 770]Cross-clausal licensinga. *A-ka-tshongoNEG-1SA-say[okokuba[thatu=MandisaD=1Mandisau-fund-is-é]1SA-read-CAUS-CONJ1ba-ntwana2-children]Intended: ‘He didn’t say that Mandisa taught any children.’b. A-ndi-fun-iNEG-1sSA-want-FV[(okokuba)[thatu=SabeloD=1Sabeloa-ty-e1SA-eat-SUBJVku-tya]15-food]‘I don’t want Sabelo to eat any food.’(497) [Zulu, adapted from C&M 2016: 770]Cross-clausal licensinga. *u=SimisoD=1Simisoa-ka-tshongoNEG-1SA-say[ukuthithatu=NothandoD=1Nothandou-theng-é1SA-buy-CONJ1mi-fino]4greensIntended: ‘Simiso didn’t say that Nothando bought any greens.’9. Halpert (2012) reports that in Durban Zulu there is no clause-mate restriction on licens-ing of unaugmented NPs. In her description, the subjunctive mood (SUBJV -e in (497b) isglossed as simply a final vowel (FV).326b. A-ngi-fun-iNEG-1sSA-want-NEG[(ukuthi)[thatu-bon-e2s-see-SUBJVmuntu]1person]‘I don’t want you to see anybody.’In Nata, licensing happens freely, as long as the negation has scope over the po-larity D∅. Licensing obtains in negative contexts when a DP containing the D∅ isseparated from negation by an indicative clause, as in (498); or when it is sepa-rated from negation by a subjunctive clause, as in (499). Both subject and objectDPs are fine10.(498) a. MakuruMakuruMakurut-a-a-bhugh-iret-a-a-βuɣ-iréNEG-3sg-PST-say-PFV[(ango)[(aŋgo)[(that)MasatoMasatoMasaton-a-a-ghor-iren-a-a-ɣór-ireFOC-SA1-PST-buy-PFVma-kuwa]má-kuwa]C6-sugarcane]‘Makuru didn’t say that Masato bought any sugarcane.’b. MakuruMakuruMakurut-a-a-bhugh-iret-a-a-βuɣ-iréNEG-3sg-PST-say-PFV[(ango)[(aŋgo)[(that)mw-aanamw-aanáC1-childn-a-a-ghor-iren-a-a-ɣór-ireFOC-SA1-PST-buy-PFVa=ma-kuwa]a=ma-kuwá]D=C6-sugarcane]‘Makuru didn’t say that any child stole (the) sugarcane.’(499) a. N-te-ghusabh-aN-te-ɣusáβ-a1sg-NEG-wish/pray-FV[Masato[Masato[Masatoa-rọr-ẹa-rɔr-ɛ ́3sg-see-SUBJVmoo-to]móo-to]C1-person]‘I don’t wish Masato to see anyone.’b. N-te-ghusabh-aN-te-ɣusáβ-a1sg-NEG-wish/pray-FV[ moo-to[ móo-to[ C1-persona-rọr-ẹa-rɔr-ɛ ́3sg-see-SUBJVMasato]Masato]Masato]‘I don’t wish anyone to see Masato.’10. The subjunctive mood can only be in the embedded clause if the matrix clause containsnegation.327These data show that Xhosa and Zulu contrast with Nata with respect to restric-tions on object DPs containing the D∅ in the indicative and in embedded clauses innegative contexts (see also Progovac 1993 on the lack of clause-mate restriction inKinande). While my analysis accounts for the Nata data, it does not predict theserestrictions in Zulu and Xhosa. Future research is need to shed more light on whatthe language-specific syntactic/semantic differences might be.6.3.5 Topic markingThe presence or absence of the augment in Kagulu correlates with topicality anddeixis. I show that Nata does not force D choice based on topicality, and that thealternative analysis I propose may offer a promising solution.Petzell (2003) argues that the augment in Kagulu marks topicality, i.e., ‘given’or presupposed information. Petzell gives the examples in (500)-(501) to supportthe topic analysis. Here the overt D correlates with topic material and the covertD correlates with referents in non-topic position:(500) D contrast in Kagulu [Kagulu, Petzell, 2003: 08 ]a. u=mu-hogoD=3-cassavau-oSM2-RELu-ni-ing’h-ileOM3-1SG-give-PST‘The cassava you have given me…’b. ni-ing’he1SG-give+SUBJmu-hogo3-casssavani-diyeSM1.SG-eat+SUBJ‘Give me cassava to eat.’(501) D contrast in Kagulu [Kagulu, Petzell, 2003: 04 ]a. i=mfeleD=-1-womanya-k-w-ambik-il-aSM1-PRES-OM2-cook-APPL-FVa=wanagweD=2-child-POSS‘The woman is cooking for her children.’b. ku-tola15-marrym-fele1-womanku-swanu15-good‘Marrying a woman is good.’328However, the use of the overt augment in DPs that are not in Topic positioncontradicts this claim. For instance, the DPs u=mgunda ‘farm’ in (502) is not intopic position but has the overt D11.(502) No Topic: overt D is Ok [Kagulu, Petzell, 2003: 5/9]si-ku-limaSM1.SG+NEG-PRES-cultivateu=m-gundaD=3-farmw-angu3-POSS‘I am not cultivating my farm.’While I do not have enough data to draw a robust conclusion about the proper-ties of the Kagulu augment data, the D contrast expressed here seems to be trackingthe speaker’s personal knowledge as we saw for Bemba and St’át’imcets in Chap-ter 5. If we propose that Kagulu leans towards an assertion-of-existence D system,(500a) and (501a) would make sense; the speaker has personal knowledge aboutthe ‘cassava’ and ‘woman’, respectively, but does not assert the existence of suchreferents in (500b) or (501b). This means s/he has no first hand evidence for theexistence of the referents. While these initial data may appeal to an existence-based account, further research is needed to pin down factors contributing to thenon-use of an augment in wa-nike ‘kids’ in (503), which is a dialogue between twoneighbours12:(503) D contrast in a dialogue [Kagulu, Petzell, 2003: 6-7 ]a. a=wa-nikeD=2-youthfo-wa-inukahow-SM2-wakenaandm-dala1-wife‘How did [the] kids and wife wake up?’b. kwainkwelifactwa-nikebf2-youthwa-inukaSM2digoyawake‘In fact, [the] kids woke up fine.’Petzell argues that the augment contrast in (503) is not due to topicality butto deixis: the unaugmented nominal in (503b) is deictic hence no augment isused. It is not surprising for an assertion-of-existence D system to encode deictic11. I have glossed the augment as D throughout this thesis, for reasons that were laid outin Chapter 3, hence I replace Petzell’s abbreviation for the pre-prefix Prpr with D=.12. It is not clear why the data were glossed without the definite article in the Englishtranslation.329features, eg., St’át’imcets also encodes spatial-temporal deictic distinctions on itsassertion-of-existence Ds X...a as we saw in Chapter 4. Languages differ in theirrestrictions on Ds in deictic contexts; some languages prohibit Ds and others allowthem (Corver 2008; Espinal 2013). Espinal (2013) argues that Ds are prohibitedwhere nominal expressions are headed by a silent deictic head (DX) or other syn-cretic cases such as vocatives (see also Corver 2008; Ndayiragije et al., 2012; deDreu 2008). While this may support an analysis of (503b) as an NP used in adeictic context, more details on the Kagula D semantics are needed before makingany firm conclusions.Nata, as a belief-of-existence D system, would use the overt D in all the Kagulucontexts given above. The relevant Nata examples are given below.(504) Exchanging greetings: overt D Oka. a=bha-anaa=βa-anaD=C2-youthno-o=mu-karina-o=mu-káriand-D=C1-wifem-ba-ree-re?m-bá-ra(r)-ireFOC-SM2-be.wellLit: ‘Are the kids and wife well?’b. kwakwainkwelikwelifacta=bha-anaa=βa-anaD=C2-youthm-ba-ree-rem-bá-ra(r)-ireFOC-SM2-be.well‘In fact [the] kids are fine.’Nata would only use a predicate nominal in vocative structures such as in(505b). The vocative structure contexts appear to be slightly different from thedeictic usage of wa-nike ‘kids’ in the Kagulu example in (503a) above.(505) a. a=bha-gharúkaa=βa-ɣarúkaD=C2-elderm=ba-re-erem=ba-ra-ireCOP-2p-sleep-PFV‘(The) elders slept fine.’b. bha-ɣarúka!βa-ɣarúka!C2-elderm-mú-re-erem-mú-ra-ireCOP-2p-sleep-PFV‘Have you slept (well), elders?’‘How are you, elders?’330c. *a=bha-gharúka!*a=βa-ɣarúka!D=C2-elderm-mu-re-erem-mu-ra-ireCOP-2p-sleep-PFVIntended: ‘How are you, elders?’The form βaɣarúka in (505b) has an indexical function which is to address/cal-l/greet individuals within the context of a speech act. In the literature, vocativesare often treated as nominal predicates which may predict that they also lack a D(see Longobardi 1994; Moro 2003; Corver 2008; Espinal 2013). This explains whyNata vocatives do not allow the augment/D as (505c) shows. Petzell lists a num-ber of other environments that are puzzling such as objects of locatives, objectsof comitatives, as well as other associative structures, which all do not allow theuse of the (overt) augment. In future research I wish to compare Nata data withobjects of Ps discussed below to determine the extent to which Nata and Kaguluvary with respect to using the overt D in these contexts13.6.3.6 Focus licensingI hinted in Chapter 5 that in some Bantu languages like Luganda, a Focus parameteraccounts for the distribution of augmentless nominals (henceforth [–A] nominals),which I argue may be distinct from polarity sensitive Ds. Consider Hyman andKatamba’s (1993) well-formedness condition in (506) that applies to augmentlessnominals. This rule will be relevant in the current discussion of Focus licensing.(506) Licensing in Luganda [Hyman and Katamba 1993: 224][ –A ] is well-formed only if it is licensed by NEG or FOC.I present the various Focus operations which seem to license [–A] nominalsin Bantu, then show that Nata lacks such operations. I propose that there is adistinction between Focussed nominals and argument DPs containing a polarity13. Petzell and Kühl (2017) investigate Lugulu–a sister language of Kagulu–and report onthe influence of non-linguistic factors affecting augment use. They discuss, among otherfactors, the way the consultants actively manipulate Lugulu by deleting the augment intranslation tasks of the Bible and in the editing process. However, they report that theyounger generation has an overusage of the augment, hence correcting places where theaugment was initially deleted. This makes sense to me given that most if not all Bantu overtaugments are not syntactically restricted. Petzell and Kühl think that, even in Lugulu, thefunction of the augment is marking topicality, givenness and other discourse functions butthey conclude that it does not have a semantic function.331sensitive D∅. The sharp difference between the two nominals is that DPs witha polarity sensitive D∅ always scope under a non-factual operator, while [–A]nominals appearing in focussed positions do not depend on licensing by a non-factual operator but depend on focal stress (see Zimmermann 2008; Carstens andMletshe 2016; and others). While marking stress may help in disambiguation,unfortunately in most Bantu literature, Focussed expressions are not consistentlymarked. I discuss data from Nata, Haya, Luganda, Kinande, and Zulu/Xhosa toargue that Focus licensing is completely missing in Nata.6.3.6.1 Focussed DPs in negative environmentsHyman (1979) and Watters (1979) argue that the dedicated Focus position in somelanguages is “Immediately After the Verb” (IAV) (see also Cheng and Downing2009; Costa and Kula 2008; Buell 2009; van der Wal 2015 and others). Hymanand Katamba (1993) discuss a number of constructions in Luganda for which theyargue that the lack of an augment correlates to Focus. One of the examples theygive is (507a) where they argue that the [–A] nominal modified by DEM can appearin the post-NEG environment, the position for focus. On the other hand, the DPin (507b) is illicit because it appears with an overt D in a focus position. AsHyman and Katamba illustrate, if the DP no longer falls under the c-commanddomain/scope of the FOC operator, as indicated by the right square bracket orexbraciation (their term) after the DEM in (507c)14, it will appear with the overtD, as in (507c). The DP with an overt D which is in a same sentence with theNEG in (507c), is right dislocated/is in topic position. This is consistent withMould (1974) who argues that if a speaker is referring to a familiar referent, DPstructures with an overt D must be used.(507) Focus marking [Luganda, adapted from H&K 1993:226]a. tè-yà-láb-àNEG-3sg.PST-see-FVbì-tábóC8-bookbì-nóC8-DEM‘He didn’t see these books.’b. *tè-yà-láb-àNEG-3sg.PST-see-FVe=bì-tábóD=C8-bookbì-nóC8-DEMIntended: ‘He didn’t see these books.’14. Hyman (p.c) pointed out to me that (507c) had a typo in the original work; hence, Ihave added a [t] which is part of the negation morpheme.332c. [t]è-yà-láb-àNEG-3sg.PST-see-FVbì-nó]C8-DEM]e=bì-tábóe=C8-book‘He didn’t see these books.’One may argue that the [–A] nominals here are DPs containing D∅ given thatthere is a NEG that may have licensed them. This is a weak argument. We knowfrom Chapter 5 that in Luganda, like in many other Bantu languages, when DEMsare used they modify DPs with an overt D, which always are associated with anexistential interpretation. There is no overt D in the focussed DP in (507c).Nata presents a different picture, consistent with the current analysis whichpredicts that an overt D must be used in all structures which induce existential in-terpretation (e.g., declarative sentences, DPs modified with DEMs and OMs, etc.).If we compare/contrast the structures of (507a,b) from Luganda and the ones in(508) from Nata, we notice that in Nata, the DP modified with DEM must oc-cur with an overt D, even when used with a non-factual operator like NEG, as in(508a).(508) a. ta-a-rooch-eta-a-rootʃ-eNEG-SA1-PST-see-FV[e=bhi-tabho[e=βi-taβo[D=C8-bookbhi-nọ] [Nata]βi-nɔ]C8-PROX.DEM]‘He didn’t see these books.’b. *t-a-a-rooch-e*t-a-a-rootʃ-eNEG-SA1-PST-see-FV[ bhi-tabho bhi-nọ][ βi-taβo βi-nɔ][ C8-bookIntended: ‘He didn’t see these books.’In Nata, Focus is morphologically marked using a different strategy, namely acopula nasal N. Structures inducing a focus interpretation have N at the left edgeenclitized to the focussed material, forcing the formation of a cleft sentence. Ido not discuss these cases here since the Luganda cases appear to be very differ-ent syntactically from the focus marked structures in Nata (see Brown 2013 alsoGambarage and Keupdjio 2013 for discussion).6.3.6.2 Focussed DPs in relative clausesMy analysis predicts that DPs modified with a relative clause will take an overt Dand be associated with an existential interpretation. While this is the case in Nata333as I show below, Mould (1974) shows that in Luganda, a focussed DP will appearwith a [–A] nominal even when it is modified with a restrictive relative clause,and when it is associated with an existential interpretation, (509b). Note that thefocussed DP in Luganda must follow negation, and here the verb is marked withan OM, which means the DP refers to a familiar doctor.(509) [Luganda, Mould 1974:225]Focus in Relative clausesa. *saamulabaNEG.1sg.PST.OM.seeo=mu-sawoD=C1-doctore-yajjaREL-SM1.PST.come‘I didn’t see THE DOCTOR who came.’b. saamulabaNEG.1sg.PST.OM.seemu-sawoC1-doctore-yajjaREL-SM1.PST.come‘I didn’t see THE DOCTOR who came.’In Nata, which lacks this kind of Focus operation, the equivalent of (509b) isungrammatical, as (510b) shows. (510a) is good because the DP modified with arelative clause has an overt D, and is marked by an OM:(510) a. [Nata]N-ty-a-a-mo-rooch-eN-ti-a-a-mo-róótʃ-e1sg-NEG-1s-PST-OM-see-PFVo=mu-ghabhoo=mu-ɣabhoD=C1-doctoru-nọu-nɔC1-RELi-i-ch-irei-i-tʃ-íreSA-PST-come-PFV‘I didn’t see (him) the doctor who came.’b. *N-ty-a-a-mo-rooch-e*N-ti-a-a-mo-rootʃ-é1sg-NEG-1s-PST-OM-see-PFVmu-ghabhomú-ɣaβoC1-doctoru-nou-nọC1-RELi-i-ch-irei-i-tʃ-íreSA-PST-come-PFV‘I didn’t see (him) the doctor who came.’It is not clear to me if these examples are focussed, since the focus markerin Nata and NEG are in complementary distribution. Even if they are, they still334show that [–A] nominals are ruled out in all contexts where a belief-of-existenceinterpretation holds, unlike in Luganda focussed constructions. I show below that[–A] nominals are licensed in many other focussed constructions in Luganda andin other languages also.6.3.6.3 Focused DPs with cleftsMy discussion will centre around DPs in clefted positions, which are usually ar-gued to be associated with exhaustive focus (see Higgins 1973; Rochemont (1986,2013); Krifka 1998; 2008 Schwarzschild 1999; Mikkelsen 2005; Lyon 2013; butsee Percus (1997); Davis et al. 2004; Lyon 2013). However, their status in termsof D content seem to vary. For convenience, the structure of the English cleft-sentence is given in (511a), showing three parts: the initial clefting pronoun it;the Focussed DP; and the residue/remnant of the cleft, a CP.(511) [Lyon 2013: 50]English cleft-sentencea. It was the snake that the mongoose caught.b. [Itcleft–pronoun] was [the snakeDP–focus] [that the mongoose caughtresidue].I explore whether polarity sensitive Ds are licensed by a FOC operator in cleftsin Nata like [–A] nominals in some languages discussed here.Starting with Nata, cleft sentences come in two types: (i) Copula initial (focusmarker) with an optional overt complementizer/a relative clause marker (REL),which agrees in Focus as in (512a); and (ii) DP-initial followed with an obligatoryrelative clause marker with focus agreement, (513a). In both cases, the overt Dmust be present on argument DPs receiving an exhaustive focus reading:(512) D in copula-initial cleft sentences [Nata]a. no=o=mo-subhene=o=mo-suβeFOC=D=C1-man(ni-we)(ni-we)(FOC-REL)i-ibh-irea-iβ-ireSA1.PST-steal-PFVa=ma-bheerea=ma-βeereD=C6-milk‘It’s THE MAN who stole the milk.’b. *no= mo-subhe*no= mo-suβeFOC= C1-man(ni-we)(ni-we)(FOC-REL)i-ibh-irea-iβ-ireSA1.PST-steal-PFVa=ma-bheerea=ma-βeereD=C6-milkIntended: ‘It’s THE MAN who stole the milk.’335(513) D in non-copula-initial cleft sentences [Nata]a. o=mo-subheo=mo-suβeD=C1-man*(ni-we)*(ni-we)FOC-RELi-ibh-irea-iβ-ireSA1.PST-steal-PFVa=ma-bheerea=ma-βeereD=C6-milk‘The man is the one who stole the milk.’b. * mo-subhe* mo-suβeC1-man*(ni-we)*(ni-we)FOC-RELi-ibh-irea-iβ-ireSA1.PST-steal-PFVa=ma-bheerea=ma-βeereD=C6-milkIntended: ‘The man is the one who stole the milk.’In either type of cleft sentence, it is impossible to have a focussed DP (whetherpreceding or following the copula) that is not a DP containing an overt D. Basedon the Nata speaker’s comments the (a) cases seem to be picking a man from a setof previously mentioned men, an idea I link with the exhaustivity implicature.Data from R/Rukiga, (514b) show that an overt augment/D is not allowed incleft sentences, hence an [–A] nominal is used15.(514) Clefts: no overt D [R/Rukiga, adapt. Allen 2014: 117]a. *o=mu-ti*D=C3-treeni-gwoCOP-RELgu-gw-ireSM3-fall-PFVIntended: ‘It is the tree that has fallen (down).’b. mu-tiC3-treeni-gwoCOP-RELgu-gw-ireSM3-fall-PFV‘It is the tree that has fallen (down).’5The Kiziba dialect of Haya seems to use the overt D in one structure and the[–A] nominal in the other. Riedel (2010) illustrates that when the full relativedemonstrative cliticized to the copula follows the noun, as we saw in Nata inexample (512) above, the overt D is used, as shown in (515a). However, whenthis COP-RM2 element, which possibly renders exhaustive focus, is missing, the[–A] nominal must be used, as in (515b):15. Note that many Bantu languages in Zone J do not have nasal overt copulas as in Nata.In copula constructions, some other elements may used used, see Hayman and Katamba(1993), Schneider-Zioga and Mutaka (2014)336(515) Clefts sentences in Haya [Haya (Kiziba), Riedel 2010: 03]a. a=Ba-isikiD=C2-girlni-boCOP-RM2y-a-tweeke-ireSM1-P1-send.APPL-P2e=bi-gemuroD=C8-present‘It’s the girls who she sent presents to.’b. Ba-isikiC2-girla-ba-i-ku-zaniraRM-SM2-be-INF-play.APPLa=ha-nyajaD=C16-C9.river‘It’s the girls who are playing by the river.’While the prohibition of the overt D here may be linked to other factors also,e.g., the kind of REL used (see Demuth and Harford 1999 on this), we have seenin R/Rukiga and some Haya dialects, [–A] nominals are licensed in focus contextswhich is not what we found in Nata.6.3.6.4 Focussed DPs in declarative sentencesIn Luganda, Focus is marked with a special intonation/focal stress; when presentedwith sentences containing [–A] nominals as regular declarative sentences (withoutthe intonation/focal stress), the speakers will reject them.(516) [Luganda, Mould 1974:228]Focus DP in declarativesa. Yatunda3sg.PST.sell.FVma-taC6-milk‘He sold SOME/THE MILK’Comment: ‘What he sold was milk.’b. Nayita1sg.PST.call.FVmu-sawoC1-doctor‘I called THE DOCTOR’Comment: Who is the person you called?Based on Mould’s comments, the kind of Focus expressed here seems to have someexhaustivity implicature, a property missing in Nata. Mould says that the focussed‘unaugmented nouns’ are emphasized, but in the same vein he adds that what isbeing focussed denotes “attributes”. It is not clear from his comment if he meansthat the material in Focus are predicates in the sense of predicate focus (Zimmer-mann 2016; 2008).337Comparing the above Luganda example with the Nata example in (517b), wesee that focus marking using the Focus sensitive D∅ does not obtain in Nata. As(517a) shows, focus is marked by a copula nasal at the left edge, in which case theovert D is required on the argument DP.(517) a. [Nata]n=a=ma-bhẹẹrẹn=a=ma-βɛɛ́rɛFOC=D=C6-milka-a-ghor-iria-a-ɣor-iri3sg-PST-sell-PFV‘He sold SOME/THE MILK.’Lit: ‘It’s THE MILK that he sold.’b. *a-ka-ghor-i*a-ka-ɣor-i3sg-PST-sell-FVma-bhẹẹrẹma-βɛɛ́rɛC6-milkIntended: ‘He sold SOME/THE MILK.’The Nata argument DPs appearing in focus positions always take an overt D.Obviously the polarity D∅ could not occur in (517b), given that there is no non-factual operator that can license it.6.3.6.5 Focussed stressed DPs in Xhosa/ZuluCarstens and Mletshe (2016) have recently led the discussion that stressed [–A]nominals in Zulu/Xhosa have a [+Focus] feature. Carstens and Mletshe arguethat stressed DPs have a [+Focus] feature and are ruled out in certain structuralpositions not associated with a Focus feature, e.g., in subject pre-verbal positions(see also Cheng and Downing 2009 for similar observations but with a slightlydifferent conclusion). They argue that the ungrammaticality of (518a) may beexplained by a focus-based analysis and not negation licensing, given that [–A]nominals in object positions in embedded clauses are licensed, (518b).(518) [Xhosa, adapt. C&M 2016: 782]No D∅ in embedded vP-externala. *a-ngi-fun-iNEG-1sSA-want-FV[okokuba[thatm-ntuC1-persona-bon-e1SA-see-SUBJu-Sabelo]D=Sabelo]Intended: ‘I don’t want anybody to see Sabelo.’338b. a-ngi-fun-iNEG-1sSA-want-FV[okokuba[thatu-SabeloD=Sabeloa-bon-e1SA-see-SUBJm-ntu]C1-person‘I don’t want Sabelo to see anybody.’The strongest evidence for a Focus feature presented by Carstens and Mletsheis that the distribution of [–A] nominals shares the same structural restrictions asfocussed materials such as DPs containing kuphela ‘only’, as in (519a), and wh-phrases, as in (519b), which also never appear in subject pre-verbal positions:(519) [Xhosa/Zulu, adapt. C&M 2016: 788-9]kuphela-DP restrictiona. *u=LoyisoaD=1Loyisou-th-é1SA-say-CONJ1(okokuba)(that)[u=Sabelo[D=Sabelokuphela]only]u-fik-ile1SA-arrive-DISJ1Intended: ‘Loyiso said that only Sabelo arrived.’b. *U-bani/bani1-1who/1whou-fik-ile1SA-arrive-DIJS1Intended: ‘Who arrived?’Carstens and Mletshe conclude that the ruled-out subject preverbal DPs are inpositions not associated with Focus16.16. Carstens and Mletshe identify four positional constraints which ban Zulu and XhosaDPs with a D∅:(i) The 4 prohibited locations in Xhosa and Zulua. Preverbal subject position.b. Right-dislocated position.c. Direct object position in a mono-transitive TEC.d. Applied object or causee in TECs.Regarding (c) and (d), I do not discuss Transitive Expletive Constructions (TECs) as Natadoes not have TECs; hence, there is no parallel data to compare. Regarding (b), we alreadyknow that object DPs corresponding with Topic or old information always force objectmarking where DPs are either right or left-dislocated and always require an overt D, whichconvey speaker’s commitment to existence of a referent denoted as old familiar information.339An alternative analysis to Focus licensing for pre-verbal subjects is that ofHalpert (2012), who does not consider licensing restrictions as based on clause-mate restrictions (see §6.3.4) or Focus. Rather, she argues that what makes the [–A] nominal cases like (520a) bad is that licensing of such nominals never obtains invP-external positions. She argues that if the subject of the embedded clause raisesto object/vP-internal position of the matrix clause, it will be licensed, (520b).(520) [Zulu, adapt. Halpert 2012: 164]No D∅ in vP-external positiona. *a-ngi-fun-iNEG-1SG-want-NEG[ukuthi[thatmu-ntuC1-persona-phek-e1SJC-cook-SUBJi=qanda]D=C5.egg]Intended: ‘I don’t want anyone to cook an/the egg.’b. a-ngi-fun-iNEG-1SG-want-NEGmu-ntuiC1-person[ukuthi[thatti a-phek-e1SJC-cook-SUBJ1i=qanda]D=C5.egg]‘I don’t want anyone to cook an egg.’Whether these restrictions derive from raising to object position, as Halpert(2012) claims, or they target Focus licensing locations, as Carstens and Mletshe(2016) claim, or a mix thereof, the lack of similar restrictions in Nata argues forparameterization of such operations in Bantu. Consider, for instance, that NataDPs with a D∅ are allowed in the exact positions where Xhosa and Zulu [–A]nominals are ruled out, (521). While one may argue that in (521a) the DP withthe D∅ has raised to the object position (cf. Halpert 2012), (521b) with an overtcomplementizer shows that the DP with a D∅ is in subject position and is fine.The D∅ of the embedded clause is licensed by NEG in the main clause.(521) [Nata]D∅ in embedded vP-external position: Oka. N-ti-kwend-aN-te-kwend-a1sg-NEG-want-FV[ moo-to[ moo-to[ c1-persona-terek-ẹa-tɛrɛḱ-ɛ3sg-cook-SUBJri=i=bhurunga]ri=i=βurúuŋga]D=C5-egg]‘I don’t want anyone to cook an/the egg.’340b. MakuruMakuruMakurut-a-a-bhugh-iret-a-a-βuɣ-ireNEG-3sg-PST-say-PFV[ango[aŋgo[thatmw-aanamw-aanaC1-childa-ka-ghor-aa-ka-ɣor-a3sg-PST-buy-FVa=ma-kuwa]a=ma-kuwa]D=C6-sugarcane]‘Makuru didn’t say that any child stole (the) sugarcane.’Unlike DPs modified by kuphela ‘only’ in Zulu and Xhosa, which are bannedin pre-verbal subject position, Nata umwene ‘him/herself/only’ DPs are freely usedin subject preverbal positions:(522) [Nata]Only-DP is alloweda. [(n-)John u=mw-ene][(n-)John u=mw-ene][(FOC)John D=C1-self]a-a-hik-irea-a-hik-ireSA1-PST-arrive-PFV‘(It’s) only John (who) arrived.’b. M.M.M.a-ka-bhugh-aa-ka-βuɣ-a3sg-PST-say-FV(kubha)(kuβa)that[(n-)John[(n-)John[(FOC-)Johnu=mw-ene]u=mw-ene]D=C1-selfa-a-hik-irea-a-hik-ireSA1-PST-arrive-PFV‘M. said that (it’s) (only) John (who) arrived.’Note that Nata uses the focus marker, the nasal N-, on the clefted DP John,unlike Zulu and Xhosa. This brings home the point that Nata does mark Focusmorphologically, and it does not deploy the same strategies as Zulu and Xhosain licensing the [–A] nominals. My analysis also does not capture the focus re-strictions discussed in cases that allow existential interpretation with [–A] focusnominals. My analysis does not also predict why the cases such as (520a) are badgiven that NEG c-commands the [–A] nominal. I propose parameterizing Focus asan independent syntactic constraint.3416.4 Areas of further researchHere I talk about the inconsistent behaviour of proper names in Nata in relationto other Bantu languages. I also discuss prepositional objects to show that there isvariation among languages concerning whether D is required in these structures.6.4.1 D requirement on proper namesMy analysis predicts that proper names would appear with an overt D, which con-veys the speaker’s commitment to existence of an individual identified by a name,based on the fact that names are definite descriptions that uniquely identify anindividual they are associated with every time they are uttered (see Burge 1973;Recanati 1997; Kripke (1977); and others). While languages such as Zulu, Xhosa,Ndali and many others (de Dreu 2008, Visser 2008, Botne 1998) require an aug-ment/D on all noun types including proper names and kinship terms17, (523),Languages like Nata cannot have a D on names or kinship terms, (524):(523) D required on names [Xhosa, Visser 2008: 5; M.N.]a. u=NdlovuD=1Ndlovu‘Ndlovu’b. u=tataD=1father‘Father’(524) Names and kinship terms in Nataa. MakuruMakurúMakurua-ka-bhereker-aa-ka-βéreker-aSA1-PST-call-FVWasatoWasátoWasato‘Makuru called for Wasato.’b. Ghookoɣookógrandmaa-ka-bhereker-aa-ka-βéreker-aSA1-PST-call-FVtaatataatáfather‘Grandma called for Daddy.’17. St’át’imcets shares this property as it requires a D on all argument nominals. Accordingto Davis (2019) St’át’imcets proper names have a dual representation: On the one hand theybehave as predicates and on the other hand nominalized names are arguments of type e.342The difference between Xhosa-type and Nata-type languages seems to be thatin the former, the overt D is obligatory for marking argumenthood even on con-stituents that are directly referring expressions like proper names, while in thelatter a D is not required on names. I consider two hypotheses regarding the lackof D on names. One is to treat names as being of argumental type. This argumentparallels Chierchia (1998) on bare nouns (NPs) in Chinese (see also Gillon andArmoskaite 2013). Alternatively, we can adopt the view that a D needs to be pro-jected for a noun to function as an argument (Déchaine 1993; Longobardi 1994;Matushansky (2008); Alexiadou et al. 2007; Déchaine and Tremblay 2011; andothers). Under this assumption, we can either propose with Longobardi (1994,1999, 2001) that N raises to D to fill the empty D slot (see also Carstens 2008 onBantu); or we can treat names as having a complex structure with a D and a pred-icate, in line with Matushansky (2008) and in part Davis (2018). The diagnosticsbelow reveal that these hypotheses yield inconclusive results which means furtherresearch is needed to discern the status of Nata proper names.Diagnostics Nata Other Ls SourceN is Pred: an Einstein test 7 3 Zamparelli (1995)N can be predicate in SC (3) 3 Matushansky (2008)N undergoes N-to-D in LF ? 3 Longobardi (1994)In Hypothesis I, proper names may be treated as a special class of nominalsthat are arguments by themselves. This may mean also that names are lexicallyfully DPs, they have no functional layer (see Gillon and Armoskaite. 2013 on thishypothesis). However, the common view about arguments is that a D needs to beprojected for a nominal to function as an argument (Déchaine 1993; Longobardi1994; Matushansky 2008; Alexiadou et al. 2007; Déchaine and Tremblay 2011;and others). Furthermore, based on results of diagnostics, it appears that Natanames do not behave as argument DPs per se, and they are also not strictly pred-icates, which means further research is needed. I discuss the mixed properties ofNata names below.6.4.1.1 N is not Pred in the ‘an Einstein’ testNata proper names behave superficially as arguments. First they only appear toallow an individual-denoting reading in contexts where predicates are used. Zam-parelli (1995) argues that names may allow for a predicative interpretation in343certain contexts, such as when they occur after the copula and take an indefiniteD, (525a):(525) Predicate and Kind DPs [English; Zamparelli, 2005: 181]a. That man is an Einstein.b. That man is Einstein.In example (525a), an Einstein is a property (the quality of being like AlbertEinstein) predicated of that man. In contrast, (525b) refers to the genuine AlbertEinstein; hence, it is interpreted as a rigid designator, i.e., picking out the uniqueindividual with such a name (see for instance Kripke 1977; Van de Velde 2019).The predicative use of proper names in English, as in (525a), is not found in Nata.Similar examples reveal that names pick out individuals, hence they behave asarguments rather than predicates, (526)18:(526) a. o=mo-too=mo-toD=C1-manu-yọu-jɔC1-DEMn-angan-aŋgaCOP-likeMakuruMakuruMakuru‘That man is like/resembles (the man called) Makuru.’18. The one context where Nata names appear to take a D is in structures with the ‘humangenitive proclitic’ ɔ=, usually denoting home, family or descendant (cf. Higgins 2011 whorefers to it as simply a ‘locative’ in Ikoma).(i) a. a-ka-bhooria-ka-βoori3sg-PST-askọ=∅-Marwaɔ=∅-MarwaGen.D=C1-Marwa‘S/he asked (the people) at Marwa’s (home).’b. a-ka-bhooria-ka-βoori3sg-PST-askọ=bha-Marwaɔ=βa-MarwaGen.D=C2-Marwa‘S/he asked (the people) at Marwa and company’s (home).’Analyzing the human-genitive proclitic as occupying the D position (recall from Chapter3 that other proclitic elements in Nata take a D also) these data suggest that the propername is taking an overt D, hence it is patterning like a predicate (see Boër 1975). Whilethis seems to be the case here, we see that Nata names are not strictly predicates as arguedbelow, which gives us inconclusive results.344b. o=mo-too=mo-toD=C1-manu-yọu-jɔC1-DEMm=Makurum=MakuruCOP=Makuru7‘That man has the properties of Makuru’3‘That man is called Makuru.’Nata proper names seem to lack a predicative usage, as they always force thename occurring after the relator nanga ‘like’ to be a DP denoting an individual,not a property. That is, the individual denoted by the subject DP is the sameindividual picked by the name Makuru. Leaving nanga ‘like’ out, (526b) is anequative copula construction, with the subject DP and the name making referenceto the same individual.6.4.1.2 N is not strictly Pred in complement clausesAdopting Hypothesis II, we may argue that Nata names are predicates which mayor may not take a D depending on where they are used, in line with Matushansky(2008) (see also Boër 1975). Matushansky uses a small clause test to see if namesmay behave syntactically as predicates (xNPs, where x is a D slot). She shows thatin small clauses, verbs of naming can only appear with a name that is a nominalpredicate of a small clause, (527) (emphasis is mine):(527) Small Clauses [Matushansky 2008: 584-9]a. Earnshaw named [the foundling Heathcliff].b. We call [William Gates Billy].Applying the small clause test to Nata proper names, we see that Nata namescan be interpreted as arguments on the one hand, (528a), and as predicates on theother hand, (528b)(see Davis (2019) for a similar observation in St’át’imcets).(528) a. bha-ka-rok-aβa-ka-rok-a3pl-PST-name-FV[u=mw-aana[u=mw-aana[D=C1-babye=rii-na Makuru]e=rii-na Makuru]D=C5-name Makuru]Lit: ‘They named [the baby the name Makuru].’345b. bha-ka-rok-aβa-ka-rok-a3pl-PST-name-FV[u=mw-aana[u=mw-aana[D=C1-babyMakuru]Makuru]Makuru]Lit: ‘They named [the baby Makuru].’The syntactic environment that Matushansky presents for hosting a predicateyields an appositive structure in (528a), i.e., in (528a) the subject DP is pickingout the the same individual: the baby/Makuru. Thus, in the naming construc-tion in (a) the name behaves as an argument and not a predicate. Note that inEnglish the sentence with the literal meaning *They called [the child the name Eli]is ungrammatical. Matushansky argues that this is because in English “a propername X cannot be replaced with a/the name X in naming constructions, whichmeans that X does not denote the name X”, (p.590)19. Turning to the (b) case,Makuru is not an individual, but rather an attribute/property. This test also givesus inconclusive results.6.4.1.3 The N-to-D movement testLongobardi (1994) argues that in Southern varieties of Italian, proper names thatoccur without a lexically filled D start out as nouns which move into D, unless Dis occupied by other D material. He notes:The specific definite reading of determinerless proper names is ob-tained by raising the head noun to D at some level of representationand leaving the foot of the chain (i.e., the N position) uninterpreted.[Longobardi 1994: 648]Longobardi gives evidence from Romance languages with syntactically condi-tioned alternations between the presence and the absence of the article for humanproper names. He shows that names can appear either after, (529a), or immedi-ately before the possessor, (529b), but never after it when the D is missing, (529c).Thus, in (529b) the name replaces the D:19. This may not be a relevant test for predicates since even in the English cases we getappositive structures not a predicate of a small clause. Rose-Marie Déchaine (p.c) observesthat it is possible in English to say they called the child by the name Eli which is grammatical,and which invalidates Matushansky’s test. The same goes for examples such as We call[William Gates Billy] in (527) where the two names may be forming an appositive structurewith two arguments.346(529) N0-to-D0 movement [adapted from Longobardi 1994: 623]a. [DP[Il]D mio [NP[Gianni]N]] ha finalmente telefonato.the my Gianni has finally called upb. [DP [Gianni]D mio [NP[Gianni]N]] ha finalmente telefonato.Gianni my has finally called upc. *Mio Gianni ha finalmente telefonato.my Gianni has finally called upWhile I do not rule out this possibility, Nata shows neither morphological norsyntactic evidence for raising of proper names to D20. Thus, the N-to-D analysis isalso inconclusive. The behaviour of Nata names calls for further research.6.4.2 D requirement on prepositional objectsIn traditional Bantu, prepositional objects include of objects of locatives, associa-tives and comitatives (see Koopman 2000; Rugemalira 2007). Largely, however,locatives do not seem to behave as prepositions in a number of languages, ratherthey either replace the augment or they pattern as noun class prefixes where theyfollow the augment. In Nata, locatives obligatorily require a complement intro-duced by an augment/D and pattern as prepositions. This section is intended todiscuss these variations and show that further research on locatives is needed. Ido not have data for associatives and comitatives for other languages, however, toshed light on this subject, I discuss Nata data with comitatives (§6.6.2.1) and asso-ciatives (§6.6.2.2); then locatives (§6.6.2.3), before comparing Nata locatives withlocatives in other languages in §6.2.4. I first show that, within Nata, while objectsof comitative and associative Ps behave like normal core arguments in permittingovert D in default environments (contexts not requiring licensing) and polarity D∅in polarity sensitive environments, objects of locative Ps allow only overt D.6.4.2.1 Objects of comitative P permit overt or covert DThe preposition na (and its allomorphs) occurring in comitative structures has ameaning parallel to with. The objects of the comitative behave as nominal argu-ments. In syntactic contexts not allowing licensing, the overt D is required.20. This would mean that Nata has two kinds of covert Ds. We know that in Nata, the covertD has meaning, i.e., it encodes the non-belief of existence of a referent and it is restrictedin its distribution (see Chapter 4 and the next section). Cross-linguistically names do nothave licensing requirements.347(530) a. MariaMariaMariaa-ka-bhin-aa-ka-βín-aSA1-PST-dance-FV[no=*(o=)mo-chuungu][na=*(o=)mo-tʃuungú][with=D=C1-white.person]‘Maria danced with a/the white man.’b. MariaMariaMariaa-ka-bhin-aa-ka-βín-aSA1-PST-dance-FV[na=*(a=)bha-chuungu][na=*(a=)βa-tʃuungú][with=D=C2-white.person]‘Maria danced with (the) white men.’A non-factual operator may license the polarity sensitive D∅ in the object of acomitative P:(531) a. MariaMariaMariata-a-bhin-ireta-a-βín-ireSA1-PST-dance-FV[na= mo-chuungu][na= mo-tʃuungú][with= C1-white.person]‘Maria did not dance with any white man.’b. MariaMariaMariata-a-bhin-ireta-a-βín-ireSA1-PST-dance-FV[na= bha-chuungu][na= βa-tʃuungú][with= C2-white.person]‘Maria did not dance with any white men.’6.4.2.2 Objects of associative P permit overt or covert DAssociative/possessive structures are formed by two DPs (i.e., DPpossessum ofDPpossessor) linked by a preposition with a similar meaning to of in English, whichis the complement of the associative (see Koopman 2000; Carstens 2000, 2008).Associatives permit either overt D or D∅. In (532) the object of P (the possessor)permits overt D:(532) a. N-ka-rọr-aN-ka-rɔr-a1sg-PST-see-FV[e=ghi-tabho[e=ɣí-taβo[D=C7-bookky=o=mu-kári]kj=o=mu-kári]C7.POSS=D=C1-woman]‘I saw a/the book of a/the woman.’348b. N-ka-rọr-aN-ka-rɔr-a1sg-PST-see-FV[e=bhii-tabho[e=βí-taβo[D=C8-bookbhy=a=bha-kári]βi=a=βa-kári]C8.POSS=D=C2-woman]‘I saw (the) books of (the) women.’Note that the possessum also requires a D, as the above examples show21. Li-censing of D∅ is decided on the basis of the syntax of possession. A non-factualoperator may license the polarity sensitive D∅ both on the object of P (the posses-sor) and on the possessum, (533a). In (533b) the possessum alone may be licensed.However, a null D is licensed on the possessor (i.e., the referent of the NP is notbelieved to be existent by the speaker), the possessum DP obligatorily also takesthe null D, which rules out the possibility of licensing the possessor alone, (533c).These results hold for plural DPs as well.(533) a. n-tj-a-rooch-en-ti-a-rootʃ-e1sg-NEG-PST-see-PFV[ ghi-tabho[ ɣí-taβo[ C7-bookke= mu-kári]ke= mu-kári]C7.of= C1-woman]‘I didn’t see any book of any woman.’b. n-tj-a-rooch-en-ti-a-rootʃ-e1sg-NEG-PST-see-PFV[ ghi-tabho[ ɣí-taβo[ C7-bookky=o=mu-kári]kj=o=mu-kári]C7.of=D=C1-woman]‘I didn’t see any book of a/the woman.’c. *n-tj-a-rooch-e*n-ti-a-rootʃ-e1sg-NEG-PST-see-PFV[e=ghi-tabho[e=ɣí-taβo[e=C7-bookke= mu-kári]ke= mu-kári]C7.of= C1-woman]Intended: ‘I didn’t see a/the book of any woman.’The ungrammaticality of (533c) taps into possessor-possessum restrictionswhich is one of the syntactic-semantic puzzles to be researched further in the fu-ture.21. Note that in Bantu the possessor follows the possessum, which contrasts with the En-glish genitive structures that put the possessor before the possessum e.g., John’s book (seeKayne 1994; Ajiboye 2005; and others). Based on the fact that P always agrees in numberwith the possessum DP, some scholars argue that the possessum follows P during Merge(see Carstens 2001, 2008; Koopman 2000; and others).3496.4.2.3 Objects of locative P require overt DLocatives are marked by classes 17, ko, and 18, mo, and can be glossed as ‘on theN’, and ‘in the N’, respectively22. In English, the object of P may permit the useof a/the, (534), but may also permit the licensing of the polarity item any when anon-factual operator is used, (535):(534) a. He threw the grass on a/the roof.b. He added (the) salt to some/the soup.(535) a. He did not throw the grass on any roof.b. They did not add salt to any soup.Nata locatives seem to have special morphosyntactic restrictions that set themapart from other prepositions. I present a variety of syntactic contexts to showthat an object of a locative P always takes the overt D in Nata, unlike those foundin locative structures such as (535) which allow a polarity reading. First, I showthat without non-factual operators, the overt D is required, (536); also with non-factual operators the overt D is still required. This happens in all the licensingenvironments, (537)-(539):(536) a. n-a-a-rẹkẹr-iren-a-a-rɛḱɛr-ireFOC-3sg-PST-throw-FVo=bhu-nyaakio=βu-ɲáakiD=C14-grass[ko=*(o=)ghi-sara][ko=*(e=)ɣi-sára][C17=D=C7-roof]‘He threw (the) grass on the roof.’b. n-a-a-tor-iren-a-a-tóor-ireFOC-3pl-PST-put-FVo=moo-nyoo=moo-ɲóD=C3-salt[mo=*(o=)mo-sori][mo=*(o=)mo-sóri][C18=D=C3-soup]‘She put (the) salt in the soup.’22. The locatives may be realized on the surface as {ko, ku, kw} and {mo, mu, mw},respectively, based on vowel harmony rules (see Anghelescu (to appear)).350(537) Negation: Overt D is required on object of locativea. t-a-a-rẹkẹr-iret-a-a-rɛḱɛr-ireNEG-3sg-PST-throw-PFVo=bhu-nyaakio=βu-ɲáakiD=C14-grass[ko=*(o=)ghi-sara][ko=*(e=)ɣi-sára][C17=D=C7-roof]‘He didn’t throw (the) grass on the roof.’b. t-a-toor-iret-a-tóor-ireNEG-3sg.PST-put-PFVo=moo-nyoo=moo-ɲóD=C3-salt[mo=*(o=)mo-sori][mo=*(o=)mo-sóri][C18=D=C3-soup]‘She didn’t put (the) salt in the soup.’(538) Polar questions: Overt D is required on object of locativea. a-ka-rẹkẹr-aa-ka-rɛḱɛr-aNEG-3sg-PST-throw-FVo=bhu-nyaakio=βu-ɲáakiD=C14-grass[ko=*(o=)ghi-sara?][ko=*(o=)ɣi-sára?][C17=D=C7-roof]‘Didn’t he throw (the) grass on the roof?’b. a-gha-toor-aa-ɣa-tóor-a3pl-PST-put-FVo=moo-nyoo=moo-ɲóD=C3-salt[mo=*(o=)mo-sori?][mo=*(o=)mo-sóri?][C18=D=C3-soup]‘Did she put (the) salt in the soup?’(539) Conditionals: Overt D is required on object of locativea. a-anga-rẹki-irea-aŋga-rɛḱɛ(r)-ire3sg-COND-throw-PFVo=bhu-nyaakio=βu-ɲáakiD=C14-grass[ko=*(o=)ghi-sara,[ko=*(o=)ɣi-sára,[C17=D=C7-roof,n-tw-anga-sẹk-ire]n-tw-aŋga-sɛk-ire]FOC-2pl-COND-laugh-PFV]‘If he threw (the) grass on the roof, we would laugh.’351b. a-anga-toor-irea-aŋga-tóor-ire3pl-COND-put-PFVo=moo-nyoo=moo-ɲóD=C3-salt[mo=*(o=)mo-sori,[mo=*(o=)mo-sóri,[C18=D=C3-soup,n-tw-anga-kum-iri]n-tw-aŋga-kum(i)-iré]FOC-2pl-COND-be.suprised-PFV]‘If she put (the) salt in the soup, we would be surprised.’Nata does not allow the licensing of the polarity D∅ on objects of locatives.The polarity sensitive D∅ is not permitted in negation, (540a), in polar questions,(540b), in conditionals (540c), etc.(540) a. *t-a-a-rẹkẹr-ire*t-a-a-rɛḱɛr-aNEG-3sg-PST-throw-PFVo=bhu-nyaakio=βu-ɲáakiD=C14-grass[ko= ghi-sara][ko= ɣi-sára][C17= C7-roof]Intended: ‘He didn’t throw (the) grass on any roof.’b. *a-ka-rẹkẹr-a*a-ka-rɛḱɛr-a3sg-PST-throw-FVo=bhu-nyaakio=βu-ɲáakiD=C14-grass[ko= ghi-sara?][ko= ɣi-sára?][C17= C7-roof]Intended: ‘Didn’t he throw (the) grass on any roof?’c. *a-anga-rẹki-ire*a-aŋga-rɛḱɛ(r)-ire3sg-COND-throw-PFVo=bhu-nyaakio=βu-ɲáakiD=C14-grass[ko= ghi-sara][ko= ɣi-sára][C17= C7-roofn-tw-anga-sẹk-iren-tw-aŋga-sɛk-ireFOC-2pl-COND-laugh-PFV]Intended: ‘If he threw (the) grass on any roof, we would laugh.’There are two ways in which a polarity D∅ can be licensed. One is that licens-ing of D∅ is possible if the locative is removed and the applicative is used instead.When this happens, what was originally the object of P will now be the appliedobject and will appear immediately after the verb. In this case a direct object (ifavailable) will follow the applied object.352(541) a. t-a-a-rẹkẹr-i-iret-a-a-rɛḱɛr-ɛ(r)-ireNEG-3sg-PST-throw-APPL-PFV[ ghi-sara[ ɣi-sára[ C7-roofo=bhu-nyaaki]o=βu-ɲáaki]D=C14-grass]‘He didn’t throw to any roof (the) grass.’b. a-ka-rẹkẹr-ẹr-aa-ka-rɛḱɛr-ɛr-a3sg-PST-throw-APPL-FV[ ghi-sara[ ɣi-sára[ C7-roofo=bhu-nyaaki?]o=βu-ɲáaki?]D=C14-grass]‘Did he throw to any roof (the) grass?’c. a-anga-rẹki-irea-aŋga-rɛḱɛ(r)-ire3sg-COND-throw-PFV[ ghi-sara[ ɣi-sára[ C7-roofo=bhu-nyaaki,o=βu-ɲáaki,D=C14-grass,n-tw-anga-sẹk-ire]n-tw-aŋga-sɛk-ire]FOC-2pl-COND-laugh-PFV]‘If he threw (the) grass to any roof, we would laugh.’The second way is to use the existential construction coupled with the ap-plicative extension on the main verb as (542) shows. Even with this strategy, thelocative argument must come immediately after the existential verb and the directobject (if available) must follow.(542) a. ghi-ta-nyihọghi-ta-ɲihɔSA7-NEG-there.is[ ghi-sara[ ɣi-sára[ C7-roofa-rẹkẹr-i-irea-rɛḱɛr-i-ire3sg-PST-throw-PPL-PFVo=bhu-nyaaki]o=βu-ɲáaki]D=C14-grass]‘There is no roof that he threw (the) grass to.’353b. N-ke-nyihọN-ke-ɲihɔSA7-NEG-there.is[ ghi-sara[ ɣi-sára[ C7-roofa-rẹkẹr-i-irea-rɛḱɛr-i-ire3sg-PST-throw-APPL-PFVo=bhu-nyaaki?]o=βu-ɲáaki?]D=C14-grass]‘Is there a roof that he threw (the) grass to?’It is possible to have both argument DPs with the D∅ in any of these structures,which indicates that with the applicative strategy, the usual licensing applies. Iwish to propose an idea for a solution of these cases.Locatives, indeed, behave as special cases in not allowing the polarity D. Ipropose that Nata locatives trigger a special pragmatic effect: namely, that of in-ducing a presupposition of existence consistent with the principle of pragmatic ac-commodation (Stalnaker 1974; Heim 1982; Chierchia and McConnell-Ginet 1990;Matthewson 1998; and others):...the hallmark of a presupposition is that it is taken for granted in thesense that its assumed truth is a precondition for felicitous utteranceof the sentence and places a kind of constraint on discourse contextsthat admit the sentence for interpretation(Chierchia and McConnell-Ginet 1990: 283)It seems to be the case that Nata locatives carry an existence presupposition,which is the reason they clash with DP complements containing the polarity D∅.This observation is analogous to Merrindah’s (2016) argument that locatives inBemba (Bantu) introduce definiteness. We should therefore consider the Natastructure rendered as she put salt in the soup, for instance, as having to do with thespeaker’s assumption that the hearer can work out the presupposition that there issomething that the salt is in. If the presupposition is not satisfied in discourse whenthe locative sentence is uttered, this can be accommodated just like any presup-position would. This then must be a felicity condition of all locative sentencesin Nata. While further research on the contribution of locative phrases in Natagoes beyond the scope of this dissertation, I submit that these structures are con-sistent with the core notion of belief of existence, in that the speaker’s belief ofexistence follows straightforwardly from the existence presupposition of locativearguments, which must have an overt D. Next I turn to discuss the behaviour of354locatives in other Bantu languages and show that these languages behave differ-ently from Nata.6.4.3 D and locatives in other Bantu languagesIn this section I discuss variation found with locatives in other languages whichindicates that locatives in Bantu are not a uniform class. Nata and other Zone Elanguages like Kuria, Ngoreme, Zanaki and others fall under (543); most languagesdo not allow an augment with the locative (cf. De Blois 1970). I now turn todiscussing the morphosyntactic/semantic properties of locatives of the (b-c) types.(543) a. Languages that require overt Ds on complements of locatives (i.e.,Nata-type, as discussed above).b. Languages that do not allow overt Ds on complements of locatives(i.e., Bemba/Kinande-type).c. Languages that allow overt Ds on complements of locatives for certainstructures only (i.e., Luganda/Zulu-type).6.4.3.1 Bemba/Kinande-type: objects of locative P prohibit overt DBased on Givón’s (1970) description of Bemba, locatives pattern differently fromother prepositions, in that they cannot ever take a complement introduced by a D:(544) Spatial Adjunct [Bemba, adapted from Givón 1970: 55]a. *A-li3sg-beMU=u=mu-shiLOC=u=C3-villageIntended: ‘He’s in the village.’b. A-li3sg-beMU=mu-shiLOC=C3-village‘He’s in the village.’Givón argues that for these cases, the ‘no-augment’ rule is maintained evenwhen it is under the c-command of a non-factual operator like negation, as (545)illustrates. Thus, even though there is no augment both the existential and non-existential interpretations are available, as shown in (545b). This situation doesnot obtain with Nata as we saw.355(545) Locatives prohibit D [Bemba, adapt. Givón 1970: 55]a. *nshiatumine1sg.NEG.PST.send.FVi=ci-taboD=C10-bookKU-u=mu-anaLOC-D=C1-childIntended: ‘I didn’t send the book to any/the child.’b. nshiatumine1sg.NEG.PST.send.FVi=ci-taboD=C10-bookKU-mu-anaLOC-C1-child‘I didn’t send a/the book to the child.‘I didn’t send a/the book to any child.’Consistent with Givón, I consider the objects of LOC in structures like in (545b)to have no D. This seems to be a case of semantic neutralization (i.e., both readingsare possible).Irimia and Schneider-Zioga (2019) discuss data in Kinande showing that aug-ments do not feature on the complements of locatives. In constructions that allowdifferential marking, they show that the locative is uku in nouns that do not takea D (like pronouns or proper nouns) (546a); and it is oko elsewhere/with commonnouns, (546b). Interestingly, they show that in negative polarity environmentswhere the covert D would have been used, the uku locative form is used, as in(546c), mimicking its use in proper names and pronouns:(546) Locatives [Kinande, Irimia & Schneider-Zioga 2019: 2]a. o=mundúu=1personmw-á-hĩkaAFF-3s-arriveukuUKUYésu1Jesus‘A man approached Jesus.’b. Maryá1Maryasy-á-wíteNEG-3S-haveoko17LOCkitábu7book‘Mary doesn’t have the book.’c. Maryá1Maryasy-á-wíteNEG-3S-haveukuUKUkitábu7book‘Mary doesn’t have any book.’We have seen that in Bemba and in Kinande the objects of locatives are notintroduced by augments, which is the opposite of Nata. As Matthewson (p.c) pointsout, there could be two responses to the proposal I made about presupposition of356existence: either because existence is presupposed, it must be matched with anovert D as is the case of Nata, or conversely, a language could choose to not allowan overt D because existence is already presupposed as in the case of Bemba. Nowfor Kinande, we cannot assume that the speaker’s belief of existence is conveyed bychanging the locative form from oko to uku because proper names and nominalsinterpreted under the scope of negation both use uku even though they do nothave the same semantics. It appears that the speaker-oriented notion of existenceis not encoded in these structures, possibly because differential object marking isin control. A syntactic analysis assuming that the LOC heads its own projectionor that the LOC sits in D is possible (Irimia and Schneider-Zioga 2019; de Dreu2008).6.4.3.2 Zulu/Luganda: objects of locative P permit overt D sometimesLanguage-specific requirements for locatives are also found in Zulu (de Dreu 2008)and in Luganda (Hyman and Katamba 1993), where augments with locatives areonly allowed in some special cases. In Zulu, de Dreu (2008) reports various typesof locatives which do not allow a D in their complements. For instance, locativestructures in Zulu may be composed of a locative prefix ku-, (547a); a locativeprefix E/O; (547b) or a locative prefix and a suffix E/O...ini, (548a). In some casesa locative precedes a noun formed by a pre-nominal DEM, (548b), or the obliqueelement that marks benefactive objects, kwa-, (549) (see Doke 1992; Halpert 2012;de Dreu 2008). Consonant or vowel phonotactics may apply.(547) Locative positions [Zulu, de Dreu 2008: 57/58]a. ku-mu-ntuLOC-C1-person‘near somebody.’b. e-sitoloLOC-7.store‘in the store.’(548) Locative positions [Zulu, de Dreu 2008: 57]a. e-m-buzi-iniLOC-C9-goat-LOC‘on the goat.’b. ku-lolu-hloboLOC-this.11-summer‘in this summer.’357(549) No augment with kwa- [Zulu, adapt. from Halpert 2012: 212]u=1SiphoD=1Siphou-zo-pheka1s-FUT-cooku=kudlaD=C15-foodkwa-zinganeLOC-C10-child‘Sipho will cook food for the children.’However, the grammar may require a D in some special cases in order forobjects to function as DP arguments. De Dreu (2008) points out that the samelocatives kwa and ku may allow a D in special cases such as with kinship terms,(550a), or with very recent class 9 loan nouns not used with E/O or -ini, as shownin (550b). (550c) shows that not all loan nouns receive a D, however.(550) a. kobaba [Zulu, adapt. from de Dreu 2008: 58]ku=o=babaLOC=D=2a.fathers‘near the fathers.’b. kw=i=confederationsku=i=confederationsLOC=D=confederationscupcupcup‘at the confederations cup.’c. kwa=MTNLOC=5.MTNServiceServiceProvider [Halpert, 2012: 212]Provider‘to the MTN Service Provider.’De Dreu, following Koopman (2000) and den Dikken (2006), argues that eventhough locatives have been analyzed traditionally as P/D in Zulu, they do not sitin D position. Rather, these structures are comprised of NPs occurring with someother projections, e.g., the E/O and the ini may project to phrases. However, thereis no explanation why the D is allowed in the ‘special cases’ and not in other cases,which shows that further research is needed.In Luganda, locative structures expressing spatial-temporal domains seem tonot allow the use of D.(551) Locatives with no D [Luganda, H&K 1993: 237]a. *ya-kí-teek-a3sg-PST-put-FV[ku=e=mmééza][C17=D=C9.table]Intended: ‘He put it on the table.’358b. ya-kí-teek-a3sg-PST-put-FV[ku=mmééza][C17=C9.table]‘He put it on the table.’(552) a. *ya-kí-teek-a3sg-PST-put-FV[mu=e=nnyúmba][C18=D=C9.house]Intended: ‘He put it in the house.’b. ya-kí-teek-a3sg-PST-put-FV[mu=nnyúmba][C18=C9.house]‘He put it in the house.’However, in some special cases, spatial VP adjuncts are DPs and the D is re-quired in order for the locative object to function as an argument, as in (553).(553) Locatives with D [Luganda, adapt. H&K 1993: 238]a. ya-lá-ba3sg-PST-put-FVe=bí-tábóD=C8-book[byáá[ofo=ku=mmééza]D=C17=C9.table]‘He saw (the) books kept on the table.’b. ya-lá-ba3sg-PST-put-FVe=bí-tábóD=C8-book[byáá[ofo=mu=nnyúmba]D=C18=C9.house]‘He saw (the) books kept in the house.’The English translations do not indicate the semantic differences between the loca-tive structures with and without an augment. My Luganda informant notes thatthe literal translation for the sentence in (553a) should be #‘He saw the books ofthe table’, where the speaker is locating the books that are usually kept there asopposed to those from the shelf. These comments are in line with my intuition forNata, where an object rather than a space is the meaning of the locative object, asin (554):(554) Spatial-VP adjuncts are DPs [Nata]359a. a-ka-rọr-aa-ka-rɔŕ-a3sg-PST-see-FVe=bhi-tabhoe=βí-taβoD=C8-book[bhe[βe[ofku=u=meecha]ku=a=mɛɛtʃá]C17=D=C9.table]Lit: ‘He saw (the) books of the table (are on the table).’b. a-ka-rọr-aa-ka-rɔŕ-a3sg-PST-see-FVe=bhi-tabhoe=βí-taboD=C8-book[bhe[βe[ofmo=o=nyuumba]mo=a=ɲuumbá]C18=D=C9.house]Lit: ‘He saw (the) books of the house (are in the house).’I propose that for Nata the functional element preceding the locative is thepartitive element. In Nata, as argued above, the locative is the P which takesthe complement introduced by D. In Luganda, the locative does not seem to be apreposition contrary to what we saw in Nata; it behaves as a normal noun classprefix preceding the D as seen in (555a). Even though it appears here that Nataand Luganda both allow a D, the data clearly show that the Luganda order oflocative structures is different from Nata, at least on the surface; i.e., in Lugandathe D precedes the LOC, which is not the case in Nata:(555) a. LugandaFPF0byáá‘of’DPD0o=LocPLOC0ku=NPmmééza‘on the table360b. NataFPF0bhePPP0ku=‘on’DPD0u=NPmẹẹcha‘the table’Myers (1987) and Bresnan and Mchombo (1987a) propose that locatives inChechewa are class prefixes and are syntactic heads (see also Diercks 2010 whoproposes the LocativeP for Lubukusu). However, this does not account the factsin Nata. Carsten (1991, 1993) argues that locatives do not head locative NPs orintroduce gender information and that they are prepositions. This observation isconsistent with Nata data.This section has shown that there are a number of independent constraints onBantu locatives at play. The Bemba locatives, for instance, present a puzzle tothe theory I developed here. My analysis does not account for all the vagaries oflocatives discussed in some of these languages. For Nata, the language I designedmy analysis for, the account is quite successful. Further research on locatives isneeded to determine whether the puzzles from other languages are reducible toindependent variation (see Marten et al. 2007, for instance, for an initial discussionof various parameters in Bantu syntax).6.5 Summary and conclusionThe question I have addressed in this chapter is the locus of parametric varia-tion in a subset of Bantu augment languages. I presented data considered to beproblematic for a semantic account and argued that some cases that have beenconsidered as “old problems” turn out to be non-problematic. I have shown thatsome DP-level elements such as demonstratives, pronouns, and some quantifiersalso may occupy the D slot, hence they can be in complementary distribution withthe augment (cf. Van de Velde 2005; Carstens and Mletshe 2016; and others).With locatives, we also saw that languages differ in terms of D requirements onlocatives and there is no uniform treatment. The solution lies in future research.361In this thesis I have challenged the idea that syntactic licensing is necessarilya surface-oriented phenomenon. For languages with surface-oriented operations,the non-factual operator must c-command the polarity D∅ in the overt syntax,e.g., in Zulu, Xhosa, Luganda, Kinande, Haya, Bemba, and Runyankore-Rukiga.For languages in which surface c-command is not a requirement, we saw thatlicensing takes place before Spell-out. We also saw that subject DPs containing thepolarity D require the licensor to occur above vP; and if the licenser is a lexicalnegative verb only object DPs can be licensed not subject DPs. We can summarizethe similarities and differences across the languages as in Table 6.2:Table 6.2: Syntactic-semantic similarities and variationComponent Nata-type other LsSyntax Licensing by a non-factual Op 3 3Licensor/Op sits above the D∅ 3 3Surface c-command 7 3Licensing is after Spell-out 7 3Locality 7 3[-A]–Focus interaction 7 3Semantics D is a speaker oriented system 3 3/DzambaD choice based on existence 3 3Quite strikingly, we saw that the ‘null D’ is not a uniform class in languages likeZulu, Xhosa, some Haya dialects, Luganda, and R/Rukiga. We saw that the sameD encodes both the speaker’s non-existential interpretation and Focus marking.While positional restrictions may be a way of diagnosing syntactic Case as Halpertclaims or Focus phenomena as Carstens and Mletshe illustrate, we saw that incertain negative contexts, a null D may be ambiguous between the Focus readingand the NPI interpretation, as the Xhosa transitive expletive construction examplein (556) shows.(556) Focussed DP in negative contexts [Xhosa, C&M 2016:765]A-ku-phek-angaNEG-17SA-cook-NEG.PASTmu-ntumu-ntua=ma-qanda!i=6-egg‘NOBODY cooked eggs!’362How does one disambiguate between the Focus reading and, say, the NPI inter-pretation of the D∅ in a case like this? Most augment accounts have in common thenon-factual operators (negation, subjunctive mood, modals, Q-morpheme, etc.),which appear to condition the distribution of augments syntactically. An openquestion is about the link between the non-factual operators and syntactic phe-nomena like Focus. The questions for future research are about how my analysiscan extend to other Bantu languages.Furthermore, there are certain grammatical phenomena that I have not ex-plored due to lack of data from augment languages, eg., animacy effects and obli-gatorily object marking. Aissen (2003), for instance, proposes hierarchies of an-imacy and definiteness showing that, in assertions, the objects highest on bothscales have higher chances of being object-marked than those on the lower end(see also Morimoto 2002; Riedel 2009, 2010; Bax and Diercks (2012); Seidel andDimitriadis 1997; and others).(557) The dimensions for object marking Aissen (2003: 437)a. Animacy: Human > Animate > Inanimateb. Definiteness: Proper name > Pronoun > Definite NP > Indef. Spe-cific NP > Non-specific NP.In a non-augment language like Swahili, obligatory object marking is linkedto animacy (Vitale 1981; Morimoto 2002; Riedel 2009, 2010 and many others).Object-marked structures are used in a variety of contexts, including definite, in-definite, and NPI contexts. In Swahili, it is not possible to use a proper name oran animate thing in an object position without object marking the verb. However,inanimate things need not be object marked unless they are mentioned in the pre-vious discourse. With obligatory object marking for animates, speakers allow anNPI interpretation of the object marked DP, as the co-referenced objects in (558)show.(558) a. ha-ku-*(mj)-pigaNEG.SM-PST-(OM)-hitrisasibullettembojelephantyeyote. [Swahili]any‘He didn’t shoot any elephant.’b. Si-ku-*(mwj)-agizaNEG.SM-PST-(OM)-order-FVma-jiC6-waterm-totojC1-childyeyoteany‘I didn’t order water from any child.’363Nata does not allow object marked DPs to be used in indefinite or in negativepolarity contexts as predicted by my analysis (refer to Chapters 3 and 4). Objectmarked DPs always render existential interpretations and polarity Ds clash withthe anaphoric reading of object markers. Are there augment languages whichallow an NPI reading or an indefinite reading with object marked DPs?Finally, I have argued that the notion of existence is pertinent to other Bantulanguages, and I have shown that while various previous accounts of Bantu as-sumed the mostly known features of specificity or definiteness; only Dzamba en-codes definiteness. Obviously I have investigated only a handful of Bantu augmentlanguages. The question for future research is: are there D systems which havebeen misanalysed or misdescribed as conveying either definiteness or specificity,but it is really speaker-oriented existence?364BibliographyAbney, S. (1987). The English noun phrase in its sentential aspect. PhD thesis. !pages 4, 102Aboh, E. O. (1998). On the syntax of gungbe noun phrases. In Paper presented atthe 27th annual meeting of the African language and linguistics colloquium. LeidenUniversity, Netherland. ! pages 101Adams, N. (2010). The Zulu ditransitive verb phrase. PhD thesis, University ofChicago. ! pages 49, 291Aikhenvald, A. Y. (2000). Classifiers: a Typology of Noun Categorization Devices.Oxford University Press, Oxford. ! pages 23Aikhenvald, A. Y. (2014). The grammar of knowledge: a cross-linguistic view ofevidentials and the expression of information source. In Aikhenvald, A. Y. andDixon, R., editors, The grammar of knowledge: a cross-linguistic typology:explorations in linguistic typology, volume 7, pages 1–50. Oxford UniversityPress. ! pages 167Aikhenvald, A. Y. (2015). Evidentials: their links with other grammaticalcategories. Linguistic Typology, 19(2):239–277. ! pages 167Aissen, J. (2003). Differential object marking: Iconicity vs. economy. NaturalLanguage and Linguistic Theory, 21(3):435–483. ! pages 363Ajiboye, O. J. (2005). Topics on Yoruba nominal expressions. PhD thesis,University of British Columbia. ! pages 103Aldridge, M. V. (1992). The elements of mathematical semantics. ! pages 59Alexiadou, A. (2014). Multiple Determiners and the Structure of DPs, volume 11.John Benjamins Publishing Company. ! pages 138, 139Alexiadou, A. and Wilder, C. (1998). Possessors, Predicates and movement in thedeterminer phrase, chapter Adjectival Modification and Multiple Determiners.John Benjamins, Amsterdam. ! pages 140365Allen, A. (2014). Definiteness and specificity in Runyankore-Rukiga. PhD thesis,Stellenbosch University. ! pages 241, 242Alonso-Ovalle, L. and Menéndez-Benito, P. (2003). Some epistemic indefinites.In Kadowaki, M. and Kawahara, S., editors, Proceedings of NELS 33, pages1–12. Amherst GLSA publications. ! pages 72Anagnostopoulou, E. (1994). Clitic dependencies in Modern Greek. PhD thesis,University of Salzburg. ! pages 141Androutsopoulou, A. (1996). The licensing of adjectival modification. InProceedings of WCCFL, volume 13, pages 17–31. ! pages 139Anghelescu, A. (2013). Morphophonology and tone in Nata. In Black, A. andLouie, M., editors, UBC Working Papers in Linguistics, volume 34, pages 89–103.Department of Linguistics, University of British Columbia. ! pages 14Anghelescu, A., Gambarage, J., Lam, W. M., and Pulleyblank, D. (2017). Nominaland verbal tone in Nata: An allomorphy-based account, chapter 5. OxfordUniversity Press, Oxford. ! pages 14Arkoh, R. and Matthewson, L. (2013). A familiar definite article in akan. Lingua,123:1–30. ! pages 73Asarina, A. (2011). Case in Uyghur and beyond. PhD thesis, MassachusettsInstitute of Technology. ! pages 42Ashton, E. O. (1944). Swahili Grammar (including intonation). Longman, Greenand Co Ltd, London. ! pages 104Ashton, E. O. (1954). A Luganda grammar. Longmans, London, New York. !pages 258Aunio, L. (2010). Ikoma nominal tone. Africana Linguistica, 16:3–30. ! pages 12Baker, M. C. (2003). Agreement, dislocation, and partial configurationality. InCarnie, A., Harley, H., and Willie, M., editors, Formal approaches to function ingrammar: In honor of Eloise Jelinek. John Benjamins, Amsterdam. ! pages 23,323Barwise, J. and Cooper, R. (1981). Generalized quantifiers and natural language.Linguistics and Philosophy, 4:159–219. ! pages 317Bauer, L. (1983). English Word-formation. Cambridge University Press. ! pages101Bauer, L. (2003). Introducing linguistic morphology. Edinburgh University Press,.! pages 136366Bax, A. and Diercks, M. (2012). Information structure constraints on objectmarking in manyika. Southern African Linguistics and Applied Language Studies,30(2):185–202. ! pages 363Bhat, D. N. (2007). Pronouns. Oxford University Press. ! pages 316Boër, S. E. (1975). Proper names as predicates. Philosophical Studies,27(6):389–400. ! pages 344, 345Bokamba, E. G. (1971). Specificity and definiteness in dzamba. SAL,2(3):217–237. ! pages 239Bolinger, D. L. (1972). Linear modification. Lingua, 18:1–34. ! pages 170Bombi, C. (2018). Definiteness in akan: Familiarity and uniqueness revisited.SALT Proceedings: Semantics and Linguistic Theory, 18:141–160. ! pages 73Borer, H. (1988). Yearbook of morphology, chapter On the morphologicalparallelism between compounds and constructs, pages 45–65. Foris,Dordrecht. ! pages 139Borer, H. (2005). In Name only, volume I, chapter Structuring Sense. OxfordUniversity Press, Oxford. ! pages 38Botne, R. (1998). Prosodically-conditioned vowel shortening in chindali. Studiesin African Linguistics, 27(1):97–121. ! pages 342Bousfield, D. (2008). Impoliteness in interaction. John Benjamins, Amsterdam. !pages 256Brame, M. (1982). The head selector theory of lexical specifications and thenon-existence of coarse categories. Linguistic Analysis, 10:321–325. ! pages 4Breheny, R. (2003). A lexical account of (implicit) bound contextualdependence. SALT, VIII. ! pages 71Bresnan, J. and Kanerva, J. (1989). Locative inversion in chichewa. LinguisticInquiry, 20:1–50. ! pages 149Bresnan, J. and Mchombo, S. (1987a). Topic, pronoun and agreement inchichewa. Language, 63:741–782. ! pages 361Bresnan, J. and Mchombo, S. A. (1987b). Topic, pronoun, and agreement inchichewa. Language, 63(4):741–782. ! pages 23Brown, C. (2013). Focus in nata: Denotation vs. discourse-new. In Paperpresented at the Canadian Linguistic Association at the University of Victoria, BCon June 2nd 2013. ! pages 27, 333367Brown, P. and Levinson, S. (1987). Politeness: Some universals in language usage.Cambridge University Press, UK. ! pages 256Bruening, B. (2008). The scope fieldwork project. ! pages 8Bruening, B. (2014). Precede-and-command revisited. Language, 90:342–388. !pages 151Buell, L. (2009). The distribution of the nguni augment: A review. In Paperpresented at workshop on the BantuAugment Workshop, June 17 2009. LeidenUniversity, Amsterdam. ! pages 281Buell, L. (2011). Zulu ngani ‘why’: Postverbal and yet in cp. Lingua, pages805–821. ! pages 138Burge, T. (1973). Reference and proper names. Journal of Philosophy,70(14):425–439. ! pages 342Burton, S. and Matthewson, L. (2015). Semantic Fieldwork Methodology, chapterTargeted construction storyboards in semantic fieldwork, pages 135–156.Oxford University Press, Oxford. ! pages 8Burzio, L. (1986). Italian syntax: A Government-Binding Approach. Dordrecht:Reidel. ! pages 148Carlson, G. and Pelletier, F. J. (1995). The Generic Book. University of ChicagoPress. ! pages 72Carstens, V. (1991). The Morphology and syntax of determiner phrases in Kiswahili.PhD thesis, UCLA. ! pages 100Carstens, V. (2000). Concord in minimalist theory. Linguistic inquiry,31(2):319–355. ! pages 23Carstens, V. (2001). Multiple agreement and case deletion: Againstphi-(in)completeness. Syntax, 3:147–163. ! pages 22Carstens, V. (2005). Agree and epp in bantu. Natural Language and LinguisticTheory, 23:219–279. ! pages 147Carstens, V. (2008). Dp in bantu and romance. In de Cat, C. and Demuth, K.,editors, The Bantu-Romance connection: a comparative study of verbal agreement,DPs, and information structure. John Benjamins Publishing CompanyAmsterdam/Philadelphia. ! pages 23Carstens, V. and Diercks, M. (2011). Agreeing ‘how’? University of Missouri andPomona College, ms. ! pages 24368Carter, H. (1963). Coding, style, and the initial vowel in northern rhodesianplateau tonga: A psycholinguistic study. African Language Studies, 4:1–42. !pages 239Casali, R. (2003). [ATR] value asymmetries and underlying vowel inventorystructure in Niger-Congo and Nilo-Saharan. Linguistic Typology, 7:307–382. !pages 47Chafe, W. (1986). Evidentiality: The linguistic coding of epistemology. Ablex Pub.Corp, New Jersey. ! pages 54Chagas, J. E. (1977). The pre-prefix. In Byarushengo, E., Duranti, A., andHyman, L., editors, Haya grammatical structure, number 6. Southern Californiaoccasional papers in linguistics. ! pages 248, 256Cheng, L. (1991). On the typology of WH-questions. Garland Publishing, NewYork. ! pages 163Cheng, L. and Downing, L. (2009). Information structure, chapter Against FocusP:arguments from Zulu, pages 247–266. Cambridge University Press, Cambridgeand New York Press. ! pages 281Cheng, L. and Rooryck, J. (2000). Licensing wh-in-situ. Syntax, 3:1–19. ! pages163Chierchia, G. (1998). Reference to kinds across language. Natural Languagesemantics, 6(4):339–405. ! pages 38Chierchia, G. (2006). Broaden your views: Implicatures of domain widening andthe “logicality” of language. Linguistic inquiry, 37(4):535–590. ! pages 219Chierchia, G. (2013). Logic in grammar: Polarity, free choice, and intervention,volume 2. Oxford University Press. ! pages 219Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris. !pages 5, 320Chomsky, N. (2008). On phases. In Freidin, R., Otero, C. P., and Zubizarreta,M. L., editors, Foundational Issues in Linguistics Theory, pages 133–166. MITPress, Cambridge. ! pages 42Christophersen, P. (1939). The articles: A study of their theory and use in English.Munksgaard, Copenhagen. ! pages 56Chung, S. and Ladusaw, W. (2004). Restriction and saturation. MIT Press,Cambridge. ! pages 58Cinque, G. (1990). Types of A’-Dependencies. MIT Press, Cambridge. ! pages 143369Cinque, G. (1995). On the evidence for partial n-movement in the romance dp.In G.Cinque, J.Koster, J.-Y.Pollock, L.Rizzi, and R.Zanuttini, editors, PathsTowards Universal Grammar. Studies in Honor of Richard S. Kayne. GeorgetownUniversity Press. ! pages 101Cinque, G. (2005). Deriving greenberg’s universal 20 and its exceptions.Linguistic Inquiry, 36:315–332. ! pages 310Condoravdi, C. (2002). Temporal interpretation of modals: modals for thepresent and for the past. In Beaver, D., Kaufmann, S., Clark, B., and Casillas,L., editors, The Construction of Meaning. CSLI Publications. ! pages 168Contini-Morava, E. (2000). Noun class and number in swahili. InContini-Morava, E. and Tobi, Y., editors, Between Grammar and Lexicon, pages3–30. John Benjamins, Amsterdam. ! pages 22Corbett, G. (1991). Gender. CUP, Cambridge. ! pages 101Corbett, G. (2000). Number. CUP, Cambridge. ! pages 23Corbett, G. (2006). Agreement. CUP, Cambridge. ! pages 23Corver, N. (2008). Uniformity and diversity in the syntax of evaluative vocatives.Journal of Comparative German Linguistics. ! pages 330Costa, J. and Kula, N. (2008). Focus at the interface: Evidence from romance andbantu. In de Cat, C. and Demuth, K., editors, The Bantu-Romance connection.John Benjamins, Amsterdam. ! pages 332Davis, H. (2010). Salish languages lack generalized quantifiers after all! In Li, N.and Lutz, D., editors, Proceeding of Semantics and Linguistic Theory. CornellUniversity. ! pages 216Davis, H. (2019). On proper names in St’át’imcets. In Reisinger, D. K. E. andMellesmoen, G., editors, Papers for the International Conference on Salish andNeighbouring Languages, volume 54. UBC working papers in linguistics. !pages 342, 345Davis, H., Matthewson, L., and Shank, S. (2004). Clefts vs. nominal predicates intwo salish languages. In Gerdts, D. B. and Matthewson, L., editors, Studies inSalish Linguistics in Honour of M. Dale Kinkade. University of Montanaoccasional papers in linguistics. ! pages 335Dayal, V. (1998). Any as inherently modal. Linguistics and philosophy,21(5):433–476. ! pages 71Dayal, V. (2004). The universal force of free choice any. Linguistic variationyearbook, 4(1):5–40. ! pages 71370De Blois, K. F. (1970). The augment in bantu languages. In Africana Linguistica,volume 4, pages 85–165. Tervuren. ! pages 19De Cat, C. and Demuth, K. (2008). The Bantu-Romance connection: a comparativestudy of verbal agreement, DPs, and information structure. John BenjaminsPublishing Company Amsterdam/Philadelphia. ! pages 16de Dreu, M. (2008). The internal structure of the zulu dp. Master’s thesis,Universiteit Leiden. ! pages 49De Haan, F. (1999). Evidentialty and epistemic modality: setting boundaries.Southwest Journal of Linguistics, 18:83–102. ! pages 167De Wolf, P. P. (1971). The noun class system of Proto-Benue-Congo. The Hague,Mouton. ! pages 19Déchaine, R.-M. (1993). Predicates across categories: Towards a category-neutralsyntax. PhD thesis, University of Massachusetts Amherst. ! pages 3Déchaine, R.-M. (2007). The evidential base. Paper presented at NELS 38 at theUniversity of Ottawa. ! pages 54Déchaine, R.-M., Dufresne, M., and Tremblay, M. (2018). The trajectory ofphi-features on old french d and n. Canadian Journal of Linguistics,63(2):167–193. ! pages 3, 35Déchaine, R.-M. and Gambarage, J. J. (2016). Evaluatives in bantu. 47thAssociation of Contemporary African Linguistics, University of CaliforniaBerkeley. ! pages 22Déchaine, R.-M., Girard, R., Mudzingwa, C., and Wiltschko, M. (2014). Theinternal syntax of shona class-prefixes. Language science, 43:18–46. ! pages22, 105Déchaine, R.-M., Si, D., and Gambarage, J. J. (2017). Nata deverbal nouns. InKandybowicz, J. and Torrence, H., editors, Africa’s Endangered Languages:Documentary and Theoretical Approaches. Oxford University Press. ! pages 14Déchaine, R.-M. and Tremblay, M. (2011). Deriving nominal reference. InGaleano, C., Görgülü, E., and Presnyakova, I., editors, Proceedings of the fortiethWestern conference on linguistics, volume 21 of WECOL, pages 419–430. !pages 72Déchaine, R.-M. and Wiltschko, M. (2002). Decomposing pronouns. LinguisticInquiry, 33:409–442. ! pages 310, 316Delsing, L.-O. (1993). The internal structure of noun phrases in Scandinavianlanguages: a comparative study. PhD thesis, University of Lund. ! pages 138371Demirdache, H. (1996). The temporal reference of noun phrases in st’at’imcetsand its implications. paper presented at the 4th annual victoria salishmorphosyntax workshop. In Salish Morphosyntax Workshop. ! pages 215Demirdache, H. and Matthewson, L. (1995). On the universality of syntacticcategories. In Proceedings of NELS 25. ! pages 215Demuth, K. and Harford, C. (1999). Verb raising and subject inversion incomparative bantu. Journal of African Languages and Linguistics, 20(1):41–61.! pages 337Demuth, K. . S. M. (1997). Presentational focus and thematic structure incomparative bantu. Journal of African Languages and Linguistics, 18:1–19. !pages 149den Dikken, M. (2006). Relators and linkers: The syntax of predication, predicateinversion, and copulas. Linguistic Inquiry Monographs, 47. ! pages 108, 358Den Dikken, M. and Giannakidou, A. (2002). From hell to polarity: Aggressivelynon-d-linked wh-phrases as polarity items. Linguistic Inquiry, 33(1):31–61. !pages 113Denny, J. P. and Creider, C. A. (1986). The semantics of noun classes inproto-bantu. In Craig, C., editor, Noun Classes and Categorization, pages217–239. Benjamins, Amsterdam. ! pages 20Déprez, V. (2000). Parallel (a)symmetries and the internal structure of negativeexpressions. Natural Language and Linguistic Theory, 18:253–342. ! pages 320Dewees, J. W. (1971). The role of syntax in the occurrence of the initial vowel inLuganda and some other Bantu languages. PhD thesis, University of Wisconsin.! pages 241, 258Diercks, M. (2010). Agreement with subjects in Lubukusu. PhD thesis, GeorgetownUniversity. ! pages 24Diercks, M. (2012). Parameterizing case: Evidence from bantu. Syntax,15(3):253–286. ! pages 100Diesing, M. (1992). Indefinites. MIT Press, Cambridge. ! pages 58Dimmendaal, G. (2001). Places and people: field sites and informants: Linguisticfieldwork. Cambridge University Press. ! pages 7Doke, C. (1955). Zulu syntax and idiom. Longmans Southern Africa, Cape Town.! pages 321Doke, C. (1992). Textbook of Zulu Grammar. Maskew Miller Longman, CapeTown. ! pages 281372Ebert, C. (2019). Blackwell Companion to Semantics, chapter Wide ScopeIndefinites: Dead Relatives. Blackwell Publishers. ! pages 221Eisner, J. (1994). Every-less in wonderland?: revisiting any. In Proceedings ofESCOL, volume 11, pages 92–103. ! pages 71Embick, D. and Noyer, R. (2001). Movement operations after syntax. LinguisticInquiry, 32(4):555–595. ! pages 28Enç, M. (1991). The semantics of specificity. Linguistic Inquiry, 22:1–26. ! pages28Erteshik-shir, N. (1993). The dynamics of focus structure. Ms., Ben GurionUniversity of the Negev. ! pages 165Espinal, M. (2013). On the structure of vocatives. In Sonnenhauser, B. and Noel,P., editors, Vocative!, Trends in Linguistics Series, chapter On the structure ofvocative. Mouton de Gruyter, Berlin. ! pages 330Espinal, M. (2017). Verb valency changes, chapter Morphosyntactic defectivenessin complex predicate formation, pages 32–49. John Benjamins, Amsterdam.! pages 37Etxeberria, U. and Giannakidou, A. (2010). The nominal structure in Slavic andBeyond, chapter D-head, domain restriction, and variation: From Greek andBasque to Salish, pages 413–440. De Gruyter. ! pages 91Farkas, D. (2002). Varieties of indefinites. In Jackson, B., editor, Proceedings ofSALT 12, Ithaca, pages 59–83. Cornell University/CLC Publications. ! pages142Featherston, S. (2007). Data in generative grammar: The stick and the carrot.Theoretical Linguistics, 33:269–318. ! pages 7Ferch, E. (2012). Scopeless quantity words in shona. Natural Language Semantics,21:373–400. ! pages 74Ferrari-Bridgers, F. (2008). A unified syntactic analysis of italian and lugandanouns. In de Cat, C. and Demuth, K., editors, The Bantu-Romance connection: acomparative study of verbal agreement, DPs, and information structure. ohnBenjamins Publishing Company Amsterdam/Philadelphia. ! pages 101Fodor, J. and Sag, I. (1982). Referential and quantificational indefinites.Linguistics and Philosophy, 5:355–398. ! pages 71Fortune, G. (1984). Shona grammatical constructions, volume 1. Mercury Press,Harare., 3rd edition. ! pages 22373Francis, N. (2014). The misterious nasal in Nata. Term paper for the 2012/2013UBC field methods class (LING431-432/531-532). ! pages 27Frege, G. ([1892]1997). On sense and reference. In Ludlow, P., editor, Readingsin the philosophy of language, pages 563–583. MIT Press, Cambridge. ! pages59Gajewski, J. (2011). A note on licensing strong npis. Natural Language Semantics,19:109–148. ! pages 145Gambarage, J. and Pulleyblank, D. (2017). Tongue root harmony in Nata: anallomorphy-based account. Oxford University Press. ! pages 14Gambarage, J. J. (2011). Mkanganyiko wa dhana za mzizi, kiini na shina katikamofolojia ya kiswahili. Kiswahili, 74(1). ! pages 101Gambarage, J. J. (2012). Context-of-use of augmented and unaugmented nounsin Nata. UBC Working Papers in Linguistics, 34:45–59. ! pages 14, 51Gambarage, J. J. (2013). Vowel harmony in Nata: An assessment of rootfaithfulness. In 2013 CLA Conference Proceedings, pages 1–15. ! pages 14, 17Gambarage, J. J. (2017). Unmasking the Bantu orthographic vowels: the challengefor language documentation and description, chapter 19. Oxford University Press.! pages 17Gambarage, J. J., Anghelescu, A., Burton, S., Dunham, J., Guntly, E., Keupdjio,H., Lam, Z.-M., Osa-Gomez, A., Pulleyblank, D., Si, D., Yoshino, Y., andDéchaine, R.-M. (2017). The Nata documentation project, chapter 3. OxfordUniversity Press. ! pages 14Gambarage, J. J. and Keupdjio, H. (2013). Wh-in-situ in nata. volume 29 ofNWLC, pages 119–129. University of British Columbia. ! pages 112Gambarage, J. J. and Matthewson, L. (2019). The bantu-salish connection indeterminer semantics. ! pages 220Garrett, E. J. (2001). Evidentiality and Assertion in Tibetan. PhD thesis, Universityof California, Los Angeles. ! pages 54Geurts, B. (1998). Presuppositions and anaphors in attitude contexts. Linguisticsand Philosophy, 21:545–601. ! pages 289Ghomeshi, J., Paul, I., and Wiltschko, M. (2009). Determiners: Universals andvariation. In Determiners: universals and variation. John Benjamins. ! pages103Giannakidou, A. (1998). Polarity sensitivity as (non)veridical dependency. JohnBenjamins, Amsterdam. ! pages 49374Giannakidou, A. (2001). The meaning of free choice. Linguistics and philosophy,24(6):659–735. ! pages 142Giannakidou, A. (2006). N-words and negative concord. In Everaert, M. and vanRiemsdijk, H., editors, The Blackwell Companion to Syntax, volume 3, pages327–391. Blackwell, Malden. ! pages 49Giannakidou, A. (2009). The dependency of the subjunctive revisited: Temporalsemantics and polarity. Lingua, 119(12):1883–1908. ! pages 170Giannakidou, A. (2011). Semantics: An international handbook of natural languagemeaning, chapter Negative and positive polarity items. De Gruyter, Berlin, 2ndedition. ! pages 142Giannakidou, A. and Mari, A. (2018). A unified analysis of the future asepistemic modality: The view from greek and italian. Natural Language andLinguistic Theory. ! pages 168Gillon, C. (2006). The semantics of determiners: domain restriction in Skwxú7mesh.PhD thesis, University of British Columbia, Vancouver, BC. ! pages 51, 82Gillon, C. (2009). Deictic features: evidence from skwxú7mesh. InternationalJournal of American Linguistics, 75:1–25. ! pages 215Gillon, C. (2010). The mass/count distinction in innu-aimun: Implications forthe meaning of plurality. In Rogers, B. and Szakay, A., editors, 15th WSCLAProceedings (UBCWPL), pages 157–164. ! pages 38Gillon, C. and Armoskaite., S. (2013). The semantic import of (c)overt d. InChoi, J., Hogue, E. A., Punske, J., Deniz Tat, J. S., and Trueman, A., editors,Proceedings of the 29th West Coast Conference on Formal Linguistics. ! pages 343Giorgi, A. and Pianesi, F. (1998). Tense and aspect: from semantics tomorphosyntax. Oxford University Press. ! pages 170Giusti, G. (1994). Enclitic articles and double definiteness: a comparativeanalysis of nominal structure in romance and germanic. The Linguistic Review,11(3-4):241–255. ! pages 138Giusti, G. (2008). Agreement and concord in nominal expressions. In de Cat, C.and Demuth, K., editors, The Bantu-Romance connection: a comparative study ofverbal agreement, DPs, and information structure. John Benjamins PublishingCompany Amsterdam/Philadelphia. ! pages 24Givón, T. (1970). Studies in Chibemba and Bantu Grammar. PhD thesis, Universityof California, Los Angeles. ! pages 114Givón, T. (2018). On Understanding Grammar. John Benjamins, Amsterdam,revised edition edition. ! pages 167375Good, J. (2005). Reconstructing morpheme order in bantu: The case ofcausativization and applicativization. Diachronica, 22:3–57. ! pages 101Greenberg, J. (1963). Some universals of grammar with particular reference tothe order of meaningful elements. In Greenberg, J., editor, Universals ofLanguage, pages 73–113. MIT Press, Cambridge. ! pages 10Grice, H. (1975). Logic and conversation. In Cole, P. and Morgan, J., editors,Syntax and semantics: speech acts, volume 3, pages 41–58. San Diego:Academic. ! pages 83Guardiano, C. (2012). Demonstratives, word order and the dp between syntaxand semantics: crosslinguistic remarks. Studies in Greek Linguistics,32:100–115. ! pages 310Guillemin, D. M. (2007). The emergence of the determiner system in mauritiancreole: A syntax semantics mapping. University of Queensland WorkingPapers in Linguistics. ! pages 51Guillemin, D. M. (2009). The Mauritian Creole Noun Phrase: Its Form and Function.PhD thesis, University of Queensland. ! pages 28Guthrie, M. (1948). The Classification of Bantu Languages. Oxford UniversityPress, London. ! pages 10Guthrie, M. (1967-1971). Comparative Bantu: An Introduction to the ComparativeLinguistics and Prehistory of the Bantu Languages, volume 1-4. GreggInternational Publishers Ltd, London: Greggs. ! pages 10Halle, M. and Marantz, A. (1993). Distributed morphology and the pieces ofinflection. In Hale, K. and Keyser, S. J., editors, The View from Building 20. MITPress. ! pages 28Halpert, C. (2012). Argument licensing and agreement in Zulu. PhD thesis, MIT. !pages 40Halpert, C. (2015). Argument licensing and agreement in Zulu. CambridgeUniversity Press. ! pages 40Hanks, W. F. (2005). Explorations in the deictic field. Current Anthropology,46(2). ! pages 51Haspelmath, M. (1997). Indefinite pronouns. Clarendon Press Oxford. ! pages142Hawkins, J. (1978). Definiteness and indefiniteness. Croom Helm, London. !pages 56Hawkins, J. (1983). Word order universals. Academic Presss. ! pages 310376Hedberg, N., Görgülü, E., and Mameni, M. (2009). More on specificity anddefiniteness in english, turkish and persian. Paper presented at MOSAIC at theUniversity of Ottawa, Ottawa, Ontatior. ! pages 70Hedberg, N. and Potter, D. (2010). Equative and predicational copulas in thai.Presented at the Berkeley Linguistics Society at the University of California atBerkeley. ! pages 108Heim, I. (1982). The semantics of definite and indefinite noun phrases. PhD thesis,University of Massachusetts, Amherst. ! pages 56Heim, I. (1988). On the projection problem for presuppositions. In Davis, S.,editor, Reprinted in Pragmatics (1991), pages 114–125. Proceedings of theSecond West Coast Conference on Formal Linguistics. ! pages 28Heim, I. (1992). Presupposition projection and the semantics of attitude verbs.JOS, 9:183–221. ! pages 212Heim, I. and Kratzer, A. (2010). Semantics in generative grammar. Blackwell,Malden, MA, 2nd edition. ! pages 59Heine, B. (1976). Knowledge and use the second language in musoma region: Aqualitative survey. Kiswahili, 46:49–59. ! pages 10Henderson, B. (2006). The syntax and typology of Bantu relative clauses. PhDthesis, University of Illinois at Urbana-Champaign. ! pages 24Higginbotham, J. (1985). On semantics. Linguistic Inquiry, 16:395–420. ! pages106Higgins, F. R. (1973). The Pseudo-cleft Construction In English. PhD thesis,Massachusetts Institute of Technology, Cambridge. ! pages 108Higgins, H. (2010). Ikoma orthography sketch. Unpublished Manuscript. SIL. !pages 14Higgins, H. A. (2011). Ikoma vowel harmony: Phonetics and phonology.Master’s thesis, Trinity Western University. ! pages 17Hill, D., Lindfors, A.-L., Nagler, L., Woodward, M., and Yalonde, R. (2007). Asociolinguistic survey of the bantu languages in mara region, tanzania. SIL,pages 1–212. ! pages 13Hintikka, J. (1969). Knowledge and belief. Cornell University Press. ! pages 212Hirose, T. (2003). The syntax of d-linking. Linguistic Inquiry, 34:499–505. !pages 113377Hoeksema, J. (2012). On the natural history of negative polarity items. LinguisticAnalysis, 38(1-2):1–33. ! pages 145Hooper, J. B. (1975). Syntax and semantics, chapter On assertive predicates,pages 91–124. Academic Press, New York. ! pages 170Hornstein, N., Nunes, J., and Grohmann, K. (2005). Understanding minimalism.Cambridge University Press. ! pages 28Huddleston, R. and Pullum, G. K. (2002). The Cambridge Grammar of the EnglishLanguage. Cambridge University Press. ! pages 116Hyman, L. (1979). Aghem Grammatical Structure. University of SouthernCalifornia Press. ! pages 332Hyman, L. M. (1999). The historical interpretation of vowel harmony in bantu.In Hombert, J.-M. and Hyman, L. M., editors, Bantu Historical Linguistics, pages235–296. CSLI Publications, Leland Stanford Junior University. ! pages 52Hyman, L. M. and Katamba, F. X. (1993). The augment in Luganda: Syntax orpragmatic?, chapter Theoretical Aspect of Bantu Grammar 1, pages 209–256.CSLI Publication, California:. ! pages 16Ionin, T. (2006). ’this’ is definitenly specific: specificity and definiteness inarticle systems. Natural Language Semantics, 14:175–234. ! pages 77Irimia, M. and Schneider-Zioga, P. (2019). Differential object marking inkinande. Paper presented ACAL50 at the University of British Columbia. !pages 356Irimia, M.-A. (2005). Types of secondary predication. Toronto Working Papers inLinguistics, 25:20–29. ! pages 116Jackendoff, R. (1977). X’ syntax: A study of phrase structure. MIT Press,Cambridge. ! pages 5Jakobson, R. (1957). Shifters, verbal categories, and the Russian verb. HavardUuniversity Press, Massachusetts. ! pages 171Jelinek, E. (1995). Quantification in straits salish. In Bach, E., Jelinek, E.,Kratzer, A., and Partee, B., editors, Quantification in Natural Languages, pages487–540. Springer. ! pages 216Johannes, J. G. (2007). The kinata noun structure. University of Dar es Salaam.! pages 17Kadmon, N. and Landman, F. (1993). Any. Linguistics and philosophy, 16(4). !pages 72378Kamp, H. and Reyle, U. (1993). From Discourse to Logic. Kluwer, Dordrecht. !pages 70Kihm, A. (2005). The Oxford handbook of comparative syntax, chapter Noun class,gender, and the lexicon-syntax-morphology interfaces: A comparative study ofNiger-Congo and Romance languages, pages 459–512. Oxford UniversityPress. ! pages 101Kiyota, M. (2009). Situation aspect andviewpoint aspect: From Salish to Japanese.PhD thesis, University of British Columbia, Vancouver, BC. ! pages 16Klima, E. (1964). Negation in English. ! pages 175Kolliakou, D. (2004). Monadic definitesand polydefinites: their form meaningand use. Journal of Linguistics, 40:263–333. ! pages 140Koopman, H. (2000). The syntax of specifiers and heads, chapter Prepositions,postpositions, circumpositions and particles: The structure of Dutch PPs.Routledge, London. ! pages 347Koopman, H. and Sportiche, D. (1991). The position of subjects. Lingua,85:211–258. ! pages 147Kotani, S. and Gambarage, J. J. (2016). Past tense and wh-phrases in nata. Paperpresented ACAL45 at the University of Oregon. ! pages 27Kratzer, A. (1998). Scope or pseudo-scope? are there wide-scope indefinites? InRothstein, S., editor, Events in Grammar. Dordrecht: Kluwer academicpublishers. ! pages 6, 234Kratzer, A. (2005). Indefinites and the operators they depend on: From japaneseto salish. Reference and quantification: The partee effect. ! pages 16Kratzer, A. and Shimoyama, J. (2002). Indeterminate pronouns: the view fromjapanese. In In 3rd Tokyo conference on psycholinguistics. ! pages 218Kremers, J. (2009). Recursive linearization. The Linguistic Review, pages135–166. ! pages 139Krifka, M. (1995). The semantics and pragmatics of polarity items. Linguisticanalysis, 25(3-4):209–257. ! pages 76Krifka, M. (1999). The semantics/pragmatics interface from different points of view,chapter At least some determiners aren’t determiners, pages 257–291. ElsevierScience, Oxford. ! pages 125Krifka, M. (2003). Bare nps: Kind-referring, indefinites, both, or neither? InProceedings of SALT 13, Ithaca. CLC Publications, NY. ! pages 63379Kripke, S. (1977). Speaker’s reference and semantic reference. In French, P.,Uihling, T., and Wettstein, H., editors, Contemporary perspectives in thephilosophy of language, volume 6-27. Minneapolis: University of MinnesotaPress. ! pages 342, 344Kyriakaki, M. (2014). The three degrees of definiteness. In Leung, H., O’Hagan,Z., Bakst, S., Lutzross, A., Manker, J., Rolle, N., and Sardinha, K., editors,Proceedings of the 40th Annuam meeting of the Berkeley Linguistics Society. !pages 35Ladusaw, W. (1979). Polarity sensitivity as inherent scope relations. PhD thesis,University of Texas, Austin. ! pages 56Ladusaw, W. A. (1980). Polarity Sensitivity as Inherent Scope Relations. Garland,NY. ! pages 145Lahiri, U. (1998). Focus and negative polarity in hindi. Natural languagesemantics, 6(1):57–123. ! pages 142Lam, Z. W.-M. (2013). The verbal morphotonology of nata. UBC Working Papersin Linguistics. ! pages 14Lamontagne, G. and Travis, L. (1986). The syntax of adjacency. In Proceedings ofthe 6th West Coast Conference on Formal Linguistics, pages 173–186. CSLIPublications, Stanford. ! pages 45Landman, F. (1991). Structures for Semantics. Kluwer, Dordrecht. ! pages 101Levinson, J. (1978). Properties and related entities. Philosophy andPhenomenological Research, 39:1–22. ! pages 108Levinson, S. (1983). Pragmatics. Cambridge University Press. ! pages 83Levinson, S. (2000). Presumptive Meanings. MIT Press, Cambridge. ! pages 83Link, G. (1983). The logical analysis of plurals and mass terms: Alattice-theoretical approach. In Bäuerle, R., Schwarze, C., and von Stechow,A., editors, Meaning, use and the interpretation of language, pages 303–323.Walter de Gruyter, Berlin, New York. ! pages 63Loebel, E. (1994). Kp/dp-syntax: interaction of case marking with referentialand nominal features. Theoretical Linguistics, 20:38–70. ! pages 45Longobardi, G. (1994). Proper names and the theory of n-movement in syntaxand logical form. Linguistic Inquiry, 25:609–655. ! pages 3Longobardi, G. (2001). The handbook of contemporary syntactic theory, chapterThe structure of DPs: Some principles, parameters, and problems, pages562–603. Blackwell, Oxford. ! pages 3380Longobardi, G. (2008). Essays on Nominal determination: From morphology todiscourse management, chapter Reference to individuals, person, and thevariety of mapping parameters, pages 188–209. John Benjamins. ! pages 3Lyon, J. (2011). The semantics of determiner phrases in okanagan. Paperpresented at Paper presented at the International Conference on Salish andNeighboring Languages (ICSNL). ! pages 29, 65Lyon, J. (2013). Predication and Equation in Okanagan. PhD thesis, University ofBritish Columbia, Vancouver, BC. ! pages 65Lyons, C. (1999). Definiteness. Cambridge University Press. ! pages 55Maclaran, R. (1982). The semantics and pragmatics of the English demonstratives.PhD thesis, Cornell University. ! pages 77Maho, J. (1999). A Comparative Study of Bantu Noun Classes. Acta UniversitatisUpsaliensis, Cothoburgensin. ! pages 19Maho, J. (2003). A classification of the bantu languages: an update of guthrie’sreferential system. In Nurse, D. and Philippson, G., editors, The BantuLanguages, pages 639–651. Routledge, New York. ! pages 10Marantz, A. (1997). No escape from syntax: don’t try morphological analysis inthe privacy of your own lexicon. In the proceedings of the 21st Annual PennLinguistics Colloquium. University of Pennsylvania working papers inlinguistcs. ! pages 100, 102Marantz, A. (2013). Verbal argument structures: Events and participants. Lingua,30:152–168. ! pages 102Marten, L., Kula, N., and Thwala, N. (2007). Parameters of morpho-syntacticvariation in bantu. Transactions of the Philological Society, 105:253–338. !pages 361Matthewson, L. (1998). Determiner Systems and Quantificational Strategies:Evidence from Salish.vidence from Salish. The Hague: Holland AcademicGraphics. ! pages 3Matthewson, L. (1999). On the interpretation of wide scope indefinites. NaturalLanguage Semantics, 7:79–134. ! pages 3Matthewson, L. (2004). On the methodology of semantic fieldwork. InternationalJournal of American Linguistics, 70(4):369–415. ! pages 7Matthewson, L. (2011). Empirical Issues in Syntax and Semantics, volume 8,chapter On apparently non-modal evidentials. Colloque de Syntaxe etSémantique de Paris, Paris. ! pages 54381Matthewson, L., Davis, H., and Rullman, H. (2007). Linguistic Variation Yearbook,volume 7, chapter Evidentials as epistemic modals: Evidence fromSt’át’imcets, pages 201–254. John Benjamins. ! pages 54Matushansky, O. (2008). On the linguistic complexity of proper names.Linguistics and philosophy, 21:573–627. ! pages 343May, R. (1985). Logical Form. MIT Press, Cambridge. ! pages 221Meeussen, A. E. (1967). Bantu grammatical reconstructions. Africana Linguistica,3:81–121. ! pages 20Meinhof, C. (1932). Introduction to the Phonology of the Bantu Languages. DietrichReimer/Ernst Vohsen., Berlin. ! pages 12Mekacha, R. K. (1985). Phonological rules affecting vowels in kinata. Master’sthesis, University of Dar es Salaam. ! pages 13Mekacha, R. K. (1993). The Sociolinguistic Impact of Swahili on Ethnic CommunityLanguages in Tanzania: A Case Study of Ekinata. PhD thesis, University ofBayreuth, Bayreuth. ! pages 13Menéndez-Benito, P. (2010). On universal free choice items. Natural LanguageSemantics, 18(1):33–64. ! pages 71Merrindah, K. (2016). A morpho-semantic analysis of locatives of town bemba.Master’s thesis, University of Zambia. ! pages 354Mikkelsen, L. (2005). Copular clauses: specification, predication, and equation.In Volume 85 of Linguistik Aktuell. Amsterdam: John Benjamins. ! pages 108,109Milsark, G. (1974). Existential sentences in English (published in 1979). PhD thesis,MIT. ! pages 61Morimoto, Y. (2002). Prominence mismatches and differential object marking inbantu. In Butt, M. and King, T. H., editors, Proceedings of the LFG02 Conference,National Technical University of Athens. CSLI Publications. ! pages 363Moro, A. (2003). Romance languages and linguistic theory 2001, chapter Notes onvocative case. A case study in clause structure, pages 247–261. ! pages 331Morwood, J. (2001). Oxford Grammar of Classical Greek. Oxford University Press,New York. ! pages 138Mould, M. (1974). The syntax and semantics of the initial vowel in luganda. InVoeltz, E., editor, Proceedings of the third annual conference on African linguistics,pages 223–230. Indiana University Press, Bloomington. ! pages 241382Mudzingwa, C. (2010). Shona Morphophonemics: Repair Strategies in Karanga andZezuru. PhD thesis, University of British Columbia. ! pages 101Muzale, H. and Rugemalira, J. (2008). Researching and documenting thelanguages of tanzania. Language Documentation and Conservation, 2:68–108. !pages 13Myers, S. (1987). Tone and the Structure of Words in Shona. PhD thesis, Universityof Massachusetts at Amherst. ! pages 361Mzolo, D. (1968). The zulu noun without the initial vowel. African studies,27(4):195–210. ! pages 281Ndayiragije, J., Nikiema, E., and Bhatt., P. (2012). The augment in kirundi:When syntax meets phonology. In Selected Proceedings of the 42nd AnnualConference on African Linguistics. Cascadilla Proceedings Project, Somerville,MA. ! pages 19Ngonyani, D. (1998). V-toi movement in kiswahili. AAP, 55:129–144. ! pages148Ngonyani, D. (1999). Xᵒ-movement in kiswahili relative clause verbs. Linguisticanalysis, 1-2:137–159. ! pages 148Ngonyani, D. (2001). The morphosyntax of negation in kiswahili. AAP,68:17–33. ! pages 148Nurse, D. (1977). Classification of the Chaga dialects. Humburg: Buske. ! pages10Nurse, D. and Philippson, G. (1980). Bantu languages of east africa: alexicostatistical survey. In Polomé, E. C. and Hill, C., editors, Languages inTanzania. Oxford University Press, London. ! pages 10Osa-Gomez, A. (2016). Quantification and freedom of choice in nata. UBCworking papers in linguistics, pages 147–163. ! pages 14, 72Owusu, A. P. and Korsah, S. (2019). Bare nouns in akan revisited. In Paperpresented at TripleA6: The Semantics of African, Asian, and AustronesianLanguages. Massachusetts Institute of Technology. ! pages 73Palmer, F. R. (2001 [1968]). Mood and modality. Cambridge University Press. !pages 170Partee, B. H. (1986). Ambiguous pseudoclefts with unambiguous be. In Berman,S., Choe, J.-W., and McDonough, J., editors, Proceedings of NELS XVI.University of Massachusetts, Amherst. ! pages 108383Percus, O. (1997). Prying open the cleft. In Proceedings of the 27th Annual Meetingof the North Eastern Linguistic Society, pages 337–351. Graduate LinguisticStudent Association, University of Massachusetts, Amherst. ! pages 335Perlmutter, D. M. (1978). Impersonal passives and the unaccusative hypothesis.In Proceedings of the 4th Annual Meeting of the Berkeley Linguistics Society. UCBerkeley, pages 157–189. ! pages 148Pesetsky, D. (1987). Wh-in-situ: movement and unselective binding. In Reuland,E. and ter Meulen, A., editors, The Representation of (in)Definiteness, pages98–129. MIT Press, Cambridge. ! pages 113Petzell, M. (2003). What is the function of the pre-prefix in kagulu? In Karlsson,M. S., editor, Andréasson, langue.doc., Göteborg University Open Archive. !pages 28, 328Petzell, M. and Kühl, K. (2017). The influence of non-linguistic factors on theusage of the pre-prefix in luguru. Linguistic Discovery, 15(1):35–48. ! pages331Portner, P. (1998). The ”progressive” in modal semantics. Language,74:760–787. ! pages 170Portner, P. and Yabushita, K. (2001). Specific indefinites and the informationstructure theory of topics. Journal of Semantics, 18(3):271–297. ! pages 71Postal, P. (1966). On so-called ”pronouns” in english. In Dinneen, F. P., editor,Report of the Seventeenth Annual Round Table Meeting on Linguistics andLanguage Studiesinguistics and language Studies, pages 177–206. GeorgetownUniversity Press, Washington, DC. ! pages 316Potts, C. (2007). The expressive dimension. Theoretical Linguistics,33(2):165–197. ! pages 22Poulos, G. and Msimang, C. T. (1998). A Linguistic Analysis of Zulu. Via Afrika,Pretoria. ! pages 114Progovac, L. (1993). Non-augmented npis in kinande as negative polarity items.In Mchombo, S., editor, Theoretical aspects of Bantu grammar, volume 1.University of Chicago Press/CSLI publications. ! pages 32Progovac, L. (2000). Coordination, c-command, and ‘logophoric’ n-words. InHorn, L. R. and Kato, Y., editors, Studies on negation and polarity, pages88–114. Oxford University Press, Oxford. ! pages 145, 163Pylkkänen, L. (2002). Introducing Arguments. PhD thesis, MIT. ! pages 152Recanati, F. (1997). Direct reference: From language to thought. Blackwell, Oxford.! pages 342384Reinhart, T. (1997). Quantifier scope: how labor is divided between qr andchoice functions. Linguistics and Philosophy, 20:335–397. ! pages 6Reinhart, T. M. (1976). The Syntactic Domain of Anaphora. PhD thesis,Massachusetts Institute of Technology. ! pages 320Riedel, K. (2009). The syntax of object marking in Sambaa: A comparative Bantuperspective. PhD thesis, Universiteit Leiden. ! pages 112, 363Riedel, K. (2010). Relative clauses in haya. In Relative Clauses in Bantu Workshop.! pages 336, 363Riedel, K. (2011). The augment in haya. In Paper presented at the 42nd meetings ofthe Annual Contemporary Association of Linguistics. ! pages 248, 255Ritter, E. (1991). Two functional categories in noun phrases. In Rothstein, S.,editor, Perspectives on Phrase Structure: Heads and Licensing, pages 7–62.Academic Press, California. ! pages 139Rizzi, L. (1997). The fine structure of the left periphery. elements of grammar.Springer, Netherlands, pages 281–337. ! pages 112Rochemont, M. (1986). Focus in generative grammar. Amsterdam: JohnBenjamins. ! pages 335Rochemont, M. (2013). Discourse new, focused and given. In Brandtler, J.,editor, Approaches to Hungarian, volume 13. Benjamins, Amsterdam. ! pages335Rooryck, J. (1994). On two types ofunderspecification: Towards a feature theoryshared by syntax and phonology. Probus, 6:207–233. ! pages 163Rugemalira, J. M. (1993). Bantu multiple “object” constructions. LinguisticAnalysis, 23(3-4):226–252. ! pages 152Rugemalira, J. M. (2007). The structure of the bantu noun phrase. SOAS WorkingPapers in Linguistics, 15:135–148. ! pages 347Rullmann, H. and You, A. (2006). General number and the semantics andpragmatics of indefinite bare nouns in mandarin chinese. In von Heusinger, K.and Turner, K., editors, 175-196, volume 16, pages 175–196. Elsevier,Amsterdam. ! pages 101Russell, B. (1905). On denoting. Mind, 14:479–493. ! pages 62Sadlier-Brown, E. (2013). The nata applied double object constructions:Implications for applicative and (a)symmetrical typologies. CanadianAssociation of Linguistics. ! pages 133385Schadeberg, T. C. (2006). Derivation: The Bantu languages, pages 71–89.Routledge, London. ! pages 21Schneider-Zioga, P. and Mutaka, P. N. (2014). Small clauses and secondarypredication in kinande. Journal of West African Languages, 42(2). ! pages 116Schütze, C. (1997). INFL in child and adult language: Agreement, case and licensing.Massachusetts Institute of Technology. ! pages 40Schwarz, F. (2009). Two types of definites in natural language. PhD thesis,University of Massachusetts Amherst. ! pages 65Schwarz, F. (2012). Situation pronouns in determiner phrases. Natural LanguageSemantics, 20(4):431–475. ! pages 65Schwarzschild, R. (1999). Givenness, avoidf, and other c=constraints on theplacement of accent. Natural Language Semantics, 7:141–177. ! pages 335Schwarzschild, R. (2002). Singleton indefinites. Journal of Semantics,19:289–314. ! pages 71Schwarzschild, R. (2006). The role of dimensions in the syntax of noun phrases.Syntax, 9(1):1–110. ! pages 38Seidel, A. and Dimitriadis, A. (1997). The discourse function of object markingin swahili. CLS, 33:373–389. ! pages 363Sharvy, R. (1980). A more general theory of definite descriptions. PhilosophicalReview, 89(4):607–624. ! pages 62Shetler, B. (1996). Telling Our Own Stories. Local Histories from South Mara,Tanzania. Leiden-Boston, Brill. ! pages 12Siloni, T. (2002). Adjectival constructs and inalienable constructions. In Ouhalla,J. and Shlonsky, U., editors, Themes in Arabic and Hebrew syntax. Kluwer,Dordrecht. ! pages 139Simango, S. R. (2006). Verb agreement and the syntax of cinsenga relativeclauses. Southern African Linguistics and Applied Language Studies, 24:277–190.! pages 112Soames, S. (1989). Presupposition. In Gabbay, D. and Guenthner, F., editors,Handbook of Philosophical Logic, pages 553–616. Dordrecht: Reidel. ! pages59Sportiche, D., Koopman, H., and Stabler, E. (2013). An Introduction to SyntacticAnalysis and Theory. Wiley-Blackwell. ! pages 147386Stalnaker, R. (1974). Pragmatic presuppositions. In Munitz, M. . and Unger, P.,editors, Semantics and philosophy, pages 471–482. New York University Press.! pages 58Stegen, O. (2002). Derivation processes in rangi. Studies in African Linguistics,31(1-2):130–153. ! pages 136Stowell, T. (1981). Origins of Phrase Structure. PhD thesis, Massachusetts Instituteof Technology. ! pages 116Stowell, T. (1989). Subjects, specifiers and x-bar theory. In Baltin, M. and Kroch,A., editors, Alternative conceptions of phrase structure. Chicago and London: TheUniversity of Chicago Press. ! pages 3Strawson, P. (1950). On referring. Mind, 59:320–344. ! pages 59Swart, H. (2009). Negation in early l2: a “window” on language genesis. InBotha, R. and Swart, H., editors, Language evolution: the view from restrictedlinguistic systems. LOT Publications, Utrecht. ! pages 175Szabolsci, A. (1983). The possessor that ran away from home. The LinguisticReview, 3:89–102. ! pages 4Szabolsci, A. (1987). Functional categories in the noun phrase. In Kenesei, I.,editor, Approaches to Hungarian 2, pages 167–190. ! pages 102Taraldsen, K. (2010). The nanosyntax of nguni noun class prefixes and concords.Lingua, 120:1522–1548. ! pages 99Ullrich, J. (2018). Modification, secondary predication and multi-verb constructionsin lakota. The Language Conservancy, Bloomington. ! pages 116Uribe-Etxebarria, M. (1994). Interface licensing conditions ons negative polarityitems: a theory of polarity and tense interactions. PhD thesis, University ofConnecticut. ! pages 151Vallduvi, E. (1994). Structural properties of information packaging in catalan. InKiss, K. E., editor, Discourse configurational languages. Oxford University Press.! pages 320Valois, D. (1991). The internal syntax of dp and adjective placement in frenchand english. In Proceedings of NELS, volume 21, pages 367–382. ! pages 311Van de Velde, M. (2008). Studies in African comparative linguistics, with specialfocus on Bantu and Mande, chapter The order of noun and demonstrative inBantu. Royal Museum for Central Africa, Tervuren. ! pages 103van de Velde, M. (2019). The Bantu languages, chapter Nominal morphology andsyntax. Routledge. ! pages 103387van der Sandt, R. (2019). Presuppositional binding. Blackwell Publishers. ! pages59van der Wal, J. (2015). Bantu syntax. In Lingustics, Morphology and Syntax.Oxford Handbooks Online. ! pages 332Vergnaud, J. R. and Zubizarreta, M. L. (1992). The definite determiner and theinalienable constructions in french and in english. Linguistic Inquiry,23(4):595–652. ! pages 37Villalba, X. (2009). Definite adjective nominalizations in spanish. In Espinal, M.,Leonetti, M., and McNally, L., editors, Proceedings of the Nereus InternationalWorkshop Definiteness and DP Structure in Romance Languages, number IV, pages139–153. ! pages 108Visser, M. (2008). Definiteness and specificity in the isixhosa determiner phrase.South African Journal of African Languages, 1:11–29. ! pages 241Vitale, A. (1981). Swahili Syntax, volume 5 of Publications in Language Sciences,Dordrecht. Foris. ! pages 363von Fintel, K. (1994). Restrictions on quantifier domains. PhD thesis, University ofMassachusetts. ! pages 82von Fintel, K. (1998). The semantics and pragmatics of quantifier domains.Paper read at the Vilem Mathesius Centre, Prague. ! pages 82von Fintel, K. (2004). Comments on beaver: Presupposition accommodation andquantifier domains. In Kamp, H. and Partee, B., editors, Context-dependence inthe analysis of linguistic meaning, pages 405–410. Elsevier, Amsterdam. !pages 289von Staden, P. (1993). The initial vowel in zulu. African studies, 32(3):163–181.! pages 281Wald, B. (1973). Syntactic change in the lake languages of northeast bantu.Studies in African Linguistics, 4(3):237–268. ! pages 309Waldie, R. (2012). Evidentiality in Nuu-chah-nulth. PhD thesis, University ofBritish Columbia, Vancouver, BC. ! pages 54Watanabe, A. (2004). The genesis of negative concord: Syntax and morphologyof negative doubling. Linguistic Inquiry, 35(4):559–612. ! pages 160Watters, J. R. (1979). Aghem Grammatical Structure, volume 7, chapter Focus inAghem, pages 137–197. University of Southern California Occasional Papersin Linguistics. ! pages 332388Westerstahl, D. (1984). Determiners and context sets. In van Benthem, J. and terMeulen, A., editors, Generalized Quantifiers in Natural Languages, pages 45–71.Reidel, Dordrecht. ! pages 82Wilkins, H. and Kimenyi, A. (1975). Strategies of constructing a definitedescription: Some evidence from kinyarwanda. Studies in African Linguistics,6(2):151–169. ! pages 114Williamson, K. (1993). The noun prefixes of new benue-congo. Journal of AfricanLanguages and Linguistics, 14(1). ! pages 19Wiltschko, M. (2009). What’s in d and how did it get there? In Ghomeshi, J.,Paul, I., and Wiltschko, M., editors, Determiners: universals and variation, pages25–66. John Benjamins, Amsterdam. ! pages 38Wiltschko, M. (2013). Descriptive relative clauses in austro-bavarian german.Canadian Journal of Linguistics, 58(2):157–189. ! pages 66Windsow, J. W. (2015). The distribution of irish locatives (seo/sin/siúd): Dp, ap,demp, or other? University of Calgary working papers in Linguistics. ! pages 310Winter, Y. (1997). Choice functions and the scopal semantics of indefinites.Linguistics and Philosophy, 20:399–467. ! pages 6Yoon, S. (2013). Parametric variation in subordinate evaluative negation:Korean/Japanese versus others. PhD thesis, University of Texas at Airlington. !pages 155, 175Zamparelli, R. (2008). On the interpretability of phi-features. In de Cat, C. andDemuth, K., editors, The Bantu-Romance connection: a comparative study ofverbal agreement, DPs, and information structure. John Benjamins PublishingCompany Amsterdam/Philadelphia. ! pages 24Zeller, J. (2008). The subject marker in bantu as an antifocus marker.Stellenbosch papers in linguistics, 38:221–254. ! pages 289Zeller, J. and Ngoboka, J. P. (2013). On parametric variation in bantu, withparticular reference to kinyarwanda. Transactions of the Philological Society,113(2). ! pages 148Zerbian, S. (2006). Expression of information structure in the bantu languagenorthern sotho. ZAS Papers in Linguistics, 45. ! pages 262Zerbian, S. and Krifka, M. (2008). Quantification across bantu languages. InMatthewson, L., editor, Quantification: a crosslinguistic perspective, volume 64,pages 383–414. North Holland Linguistic Series. ! pages 49Zimmermann, M. (2008). Constrastive focus and emphasis. Acta LinguisticaHungarica, 55:347–360. ! pages 262389Zimmermann, M. (2016). Predicate Focus. Oxford Handbooks Online. ! pages262, 337Zwarts, F. (1998). Three types of polarity. In Hamm, F. and Hinrichs, E., editors,Plural quantification, pages 177–238. Dordrecht, Kluwer. ! pages 145Zwicky, A. M. (1977). On clitics. Indiana University Linguistics Club,Bloomington. ! pages 103390"@en ; edm:hasType "Thesis/Dissertation"@en ; vivo:dateIssued "2019-11"@en ; edm:isShownAt "10.14288/1.0384566"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Linguistics"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "Attribution-NonCommercial-NoDerivatives 4.0 International"@* ; ns0:rightsURI "http://creativecommons.org/licenses/by-nc-nd/4.0/"@* ; ns0:scholarLevel "Graduate"@en ; dcterms:title "Belief-of-existence determiners : evidence from the syntax and semantics of Nata augments"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/72023"@en .