{"@context":{"@language":"en","Affiliation":"http:\/\/vivoweb.org\/ontology\/core#departmentOrSchool","AggregatedSourceRepository":"http:\/\/www.europeana.eu\/schemas\/edm\/dataProvider","Campus":"https:\/\/open.library.ubc.ca\/terms#degreeCampus","Creator":"http:\/\/purl.org\/dc\/terms\/creator","DateAvailable":"http:\/\/purl.org\/dc\/terms\/issued","DateIssued":"http:\/\/purl.org\/dc\/terms\/issued","Degree":"http:\/\/vivoweb.org\/ontology\/core#relatedDegree","DegreeGrantor":"https:\/\/open.library.ubc.ca\/terms#degreeGrantor","Description":"http:\/\/purl.org\/dc\/terms\/description","DigitalResourceOriginalRecord":"http:\/\/www.europeana.eu\/schemas\/edm\/aggregatedCHO","FullText":"http:\/\/www.w3.org\/2009\/08\/skos-reference\/skos.html#note","Genre":"http:\/\/www.europeana.eu\/schemas\/edm\/hasType","GraduationDate":"http:\/\/vivoweb.org\/ontology\/core#dateIssued","IsShownAt":"http:\/\/www.europeana.eu\/schemas\/edm\/isShownAt","Language":"http:\/\/purl.org\/dc\/terms\/language","Program":"https:\/\/open.library.ubc.ca\/terms#degreeDiscipline","Provider":"http:\/\/www.europeana.eu\/schemas\/edm\/provider","Publisher":"http:\/\/purl.org\/dc\/terms\/publisher","Rights":"http:\/\/purl.org\/dc\/terms\/rights","RightsURI":"https:\/\/open.library.ubc.ca\/terms#rightsURI","ScholarlyLevel":"https:\/\/open.library.ubc.ca\/terms#scholarLevel","Title":"http:\/\/purl.org\/dc\/terms\/title","Type":"http:\/\/purl.org\/dc\/terms\/type","URI":"https:\/\/open.library.ubc.ca\/terms#identifierURI","SortDate":"http:\/\/purl.org\/dc\/terms\/date"},"Affiliation":[{"@value":"Arts, Faculty of","@language":"en"},{"@value":"Linguistics, Department of","@language":"en"}],"AggregatedSourceRepository":[{"@value":"DSpace","@language":"en"}],"Campus":[{"@value":"UBCV","@language":"en"}],"Creator":[{"@value":"Osa G\u00f3mez del Campo, Adriana","@language":"en"}],"DateAvailable":[{"@value":"2020-08-31T22:13:31Z","@language":"en"}],"DateIssued":[{"@value":"2020","@language":"en"}],"Degree":[{"@value":"Doctor of Philosophy - PhD","@language":"en"}],"DegreeGrantor":[{"@value":"University of British Columbia","@language":"en"}],"Description":[{"@value":"In this dissertation, I explore how speakers of Peninsular Spanish keep epistemic alignment during a conversation by using two discourse markers: sentence initial 'si' and sentence final 'no' (Ortega, 1985; Montol\u00edo Dur\u00e1n, 1999; Schwenter, 2000b; Monta\u00f1ez Mesas, 2015; Schwenter, 2016b). I show that these two discourse markers operate at different conversational levels: 'si' keeps an alignment in the set of shared, public propositions, whereas 'no' keeps an alignment between the information states of the interlocutors.\r\nFirst, I focus on 'si' and its distribution. I show that 'si' is an illocutionary operator that marks that a proposition was already known by all interlocutors, hence forcing an alignment within the Common Ground. The sentence in which 'si' occurs is used in the conversation to reject the previous contribution due to a violation of a felicity condition. I use 'si' to show that speakers have two versions of the Common Ground: one representing what they themselves know, and the other representing what they believe the Addressee to know.\r\nI also discuss the distribution of 'no', which is used to align both beliefs and outcomes among interlocutors. It does so by requesting confirmation from the Addressee that the contextual update proposed by the Speaker is shareable and public. The sentence in which 'no' occurs is used in the conversation to confirm a bias held by the Speaker, either with respect to a belief or an outcome that is preferred.\r\nFinally, I offer a syntactic analysis of both discourse markers. I adopt an analysis based on proposals to include utterance-level information in the left periphery. I propose that 'si' is an illocutionary operator situated within the Complementizer Phrase, as high as ForceP, based on prosodic and syntactic evidence. I propose that 'no' occupies an even higher position, outside of Complementizer Phrase but still within the same clause. I follow recent analyses in the syntactic literature in calling this higher structure the Grounding layer (Thoma, 2016; Wiltschko and Heim, 2016). Within it, 'no' occupies the layer that represents the Addressee\u2019s information state (as perceived by the Speaker).","@language":"en"}],"DigitalResourceOriginalRecord":[{"@value":"https:\/\/circle.library.ubc.ca\/rest\/handle\/2429\/75792?expand=metadata","@language":"en"}],"FullText":[{"@value":"Epistemic (mis)alignment in discourse:what Spanish discourse markers revealbyAdriana Osa Go\u00b4mez del CampoB.A. English Philology, Complutense University of Madrid, 2012A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFDoctor of PhilosophyinTHE FACULTY OF GRADUATE AND POSTDOCTORALSTUDIES(Linguistics)The University of British Columbia(Vancouver)August 2020c\u00a9 Adriana Osa Go\u00b4mez del Campo, 2020The following individuals certify that they have read, and recommend to the Fac-ulty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled:Epistemic (mis)alignment in discourse: what Spanish discourse markers revealsubmitted by Adriana Osa Go\u00b4mez del Campo in partial fulfillment of the require-ments forthe degree of Doctor of Philosophyin LinguisticsExamining CommitteeRose-Marie De\u00b4chaine, LinguisticsSupervisorHotze Rullmann, LinguisticsSupervisory Committee MemberMartina Wiltschko, LinguisticsSupervisory Committee MemberLisa Matthewson, LinguisticsUniversity ExaminerJon Beasley-Murray, French, Hispanic, and Italian StudiesUniversity ExaminerDonka F. Farkas (Linguistics, UC Santa Cruz)External ExamineriiAbstractIn this dissertation, I explore how speakers of Peninsular Spanish keep epistemicalignment during a conversation by using two discourse markers: sentence ini-tial si and sentence final no (Ortega, 1985; Montol\u0131\u00b4o Dura\u00b4n, 1999; Schwenter,2000b; Montan\u02dcez Mesas, 2015; Schwenter, 2016b). I show that these two dis-course markers operate at different conversational levels: si keeps an alignment inthe set of shared, public propositions, whereas no keeps an alignment between theinformation states of the interlocutors.First, I focus on si and its distribution. I show that si is an illocutionary oper-ator that marks that a proposition was already known by all interlocutors, henceforcing an alignment within the Common Ground. The sentence in which si oc-curs is used in the conversation to reject the previous contribution due to a vio-lation of a felicity condition. I use si to show that speakers have two versions ofthe Common Ground: one representing what they themselves know, and the otherrepresenting what they believe the Addressee to know.I also discuss the distribution of no, which is used to align both beliefs andoutcomes among interlocutors. It does so by requesting confirmation from theAddressee that the contextual update proposed by the Speaker is shareable andpublic. The sentence in which no occurs is used in the conversation to confirma bias held by the Speaker, either with respect to a belief or an outcome that ispreferred.Finally, I offer a syntactic analysis of both discourse markers. I adopt an analy-sis based on proposals to include utterance-level information in the left periphery.iiiI propose that si is an illocutionary operator situated within the ComplementizerPhrase, as high as ForceP, based on prosodic and syntactic evidence. I proposethat no occupies an even higher position, outside of Complementizer Phrase butstill within the same clause. I follow recent analyses in the syntactic literaturein calling this higher structure the Grounding layer (Thoma, 2016; Wiltschko andHeim, 2016). Within it, no occupies the layer that represents the Addressee\u2019sinformation state (as perceived by the Speaker).ivLay SummaryAs language users, one of our conversational aims is to stay aligned\u2013that is, insync\u2013with our audience. This alignment includes the information we know andthe actions that we want to perform. But sometimes this alignment is in danger.In this dissertation, I focus on how speakers of Peninsular Spanish re-establishor maintain alignment by using certain words called discourse markers (DMs),in particular si and no. I divide my analysis into two parts: (i) what the wholesentence with the DM does, and (ii) what the DM does to the sentence it appearsin. I show that the use of these DMs also depends on the appropriateness of theprevious contribution to the conversation (DM si) and the bias of the Speaker infavour of a belief or an outcome (DM no). Finally, I show how these insights havetheir representation in the structure of the sentence.vPrefaceThe work presented in this thesis is original research conducted by the author,Adriana Osa Go\u00b4mez del Campo. It builds on work done on interactional languageby Dr. Martina Wiltschko\u2019s \u2018Eh Lab\u2019 (syntaxofspeechacts.linguistics.ubc.ca) boththeoretically and empirically (part of the elicitation methods used were based onstoryboards developed in this project, see 1.5).The elicitations undertaken for this dissertation are covered under ethics ap-proval for the project \u201cKnowledge asymmetries in discourse: what Spanish dis-course markers can tell [Spanish DMs]\u201d (H16-03266) granted to the supervisor,Dr. Rose-Marie De\u00b4chaine. Elicitations were conducted by the author, who alsoanalyzed the data.Parts of this dissertation have been presented at different venues, including:\u2022 An earlier versions of the analysis of one of the discourse markers (DM no)was published in the proceedings of the 21st Workshop on the Semanticsand Pragmatics of Dialogue, which took place in Saarbru\u00a8cken in June 2017.\u2022 An earlier version of the analysis of one of the discourse markers (DM si)was presented at the Hispanic Linguistics Symposium 2019, which tookplace at El Paso in October 2019.\u2022 The dialogue model based on three phases was presented as a poster byDr. Johannes Heim and the author at the Canadian Linguistics Association,which took place in Vancouver in June 2019.viTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xivAbbreviations and definitions . . . . . . . . . . . . . . . . . . . . . . . . xixAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 A tale of two discourse markers . . . . . . . . . . . . . . . . . . . 41.2 The protagonists: DM si and DM no . . . . . . . . . . . . . . . . 81.2.1 The \u2018rebel\u2019 complementizer: the Peninsular Spanish DM si 81.2.2 The negotiator: Peninsular Spanish DM no . . . . . . . . 111.3 The setting: conversational updates . . . . . . . . . . . . . . . . . 131.3.1 What gets updated in a conversation . . . . . . . . . . . . 141.3.1.1 Shared: The Common Ground . . . . . . . . . . 15vii1.3.1.2 Shared: Common To-Do-List . . . . . . . . . . 171.3.1.3 Individualized component: discourse commit-ments . . . . . . . . . . . . . . . . . . . . . . . 171.3.1.4 Individualized component: To-Do-Lists . . . . . 181.3.2 Canonical conversation updates . . . . . . . . . . . . . . 191.3.3 Complex conversation updates: two cases . . . . . . . . . 231.3.3.1 Mending epistemic misalignment: Peninsular Span-ish DM si . . . . . . . . . . . . . . . . . . . . . 231.3.3.2 Confirming epistemic alignment: the Peninsularspanish DM no . . . . . . . . . . . . . . . . . . 261.3.4 The link between speech acts and conversational updates . 301.4 The road ahead . . . . . . . . . . . . . . . . . . . . . . . . . . . 331.4.1 Where we find si and no and what they do . . . . . . . . . 331.4.2 Analysis of DM si: monitoring the CG . . . . . . . . . . . 341.4.3 Analysis of DM no: monitoring the Speaker-Addresseerelation . . . . . . . . . . . . . . . . . . . . . . . . . . . 351.4.4 What\u2019s syntax got to do with it? . . . . . . . . . . . . . . 361.5 On data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402 The Peninsular Spanish DMs si and no: where we find them andwhat they do . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422.1 Peninsular Spanish si: where we find it and previous accounts . . . 432.1.1 Where we find si . . . . . . . . . . . . . . . . . . . . . . 452.1.1.1 Si and dependent CPs: interrogative and condi-tional si . . . . . . . . . . . . . . . . . . . . . . 452.1.1.2 Root CPs: distinguishing exclamative si fromDM si . . . . . . . . . . . . . . . . . . . . . . . 462.1.1.3 Si and root CPs: DM si . . . . . . . . . . . . . . 492.1.2 What DM si does: previous accounts . . . . . . . . . . . . 502.1.2.1 Si-utterances are adversative . . . . . . . . . . . 50viii2.1.2.1.1 Independent si-clauses refute implica-tures . . . . . . . . . . . . . . . . . . 522.1.2.1.2 Independent si-clauses refute inferencesfrom context . . . . . . . . . . . . . . 532.1.2.1.3 Independent si-clauses refute the formof the previous contribution . . . . . . 532.1.2.1.4 Independent si-clauses refute previouslyheld beliefs . . . . . . . . . . . . . . . 542.1.2.2 DM si marks obvious truth . . . . . . . . . . . . 542.1.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 552.2 Peninsular Spanish no: where we find it and previous accounts . . 552.2.1 Where we find no . . . . . . . . . . . . . . . . . . . . . . 572.2.1.1 No and propositional polarity . . . . . . . . . . 572.2.1.2 No and discourse marking . . . . . . . . . . . . 622.2.1.2.1 Non-confirmational . . . . . . . . . . 642.2.1.2.2 Confirmational . . . . . . . . . . . . . 652.2.1.3 Declaratives can be anchors of DM no . . . . . . 672.2.1.4 Imperatives can be anchors of DM no . . . . . . 672.2.1.5 Interrogatives can be anchors of DM no . . . . . 682.2.1.6 Exclamatives can be anchors of no . . . . . . . . 702.2.2 What DM no does: previous accounts . . . . . . . . . . . 712.2.2.1 DM no requests confirmation of facts or opinions 712.2.2.2 DM no can be Speaker or Addressee-oriented . . 722.3 Where to go from here . . . . . . . . . . . . . . . . . . . . . . . 732.3.1 A unifying analysis for DM si . . . . . . . . . . . . . . . 742.3.2 A unifying analysis for DM no . . . . . . . . . . . . . . . 772.3.3 Tools for modelling Peninsular Spanish si and no . . . . . 802.3.3.1 Phases to describe the contexts of use . . . . . . 802.3.3.2 Committing to propositions and outcomes . . . . 822.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84ix3 Monitoring the Common Ground: Peninsular Spanish si . . . . . . 853.1 Ingredients of the analysis . . . . . . . . . . . . . . . . . . . . . . 853.2 How to monitor the CG: reminding the Addressee of its content . . 893.2.1 Reminding the Addressee that the proposition is alreadyin the CG . . . . . . . . . . . . . . . . . . . . . . . . . . 903.2.2 The status of the Common Ground . . . . . . . . . . . . . 933.3 Why monitor the CG: rejecting infelicitous speech acts . . . . . . 943.3.1 Si-clauses reject infelicitous assertions . . . . . . . . . . . 963.3.2 Si-utterances reject infelicitous commands . . . . . . . . . 1023.3.3 Si-clauses reject infelicitous questions . . . . . . . . . . . 1083.3.4 Si-clauses reject infelicitous promises . . . . . . . . . . . 1133.3.5 Si-clauses reject infelicitous thank-yous . . . . . . . . . . 1183.3.6 Si-utterances reject infelicitous greetings . . . . . . . . . . 1203.4 How this analysis explains other observations . . . . . . . . . . . 1213.4.1 Si-utterances reject failed implicatures and presuppositions 1213.4.2 Si-clauses do not need to convey contrast . . . . . . . . . 1243.4.3 Si-clauses reject defective inferences from context . . . . . 1253.4.4 Si-clauses reject previously held beliefs by Speaker . . . . 1263.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1274 Monitoring the Speaker-Addressee relation: Peninsular Spanish no 1284.1 Ingredients of the analysis . . . . . . . . . . . . . . . . . . . . . . 1294.2 How to monitor the S-A relation: what speakers can do with DM no1334.2.1 DM no negotiates the update proposed by assertions . . . 1344.2.1.1 How DM no modifies non-evaluative assertions . 1374.2.1.2 How DM no modifies evaluative assertions . . . 1414.2.2 DM no negotiates the update proposed by exclamations . . 1474.2.3 DM no negotiates the update proposed by commands . . . 1504.2.4 DM no negotiates the update proposed by (rhetorical) ques-tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1554.3 What speakers cannot do with DM no . . . . . . . . . . . . . . . 157x4.3.1 DM no is infelicitous with (information-seeking) questions 1584.3.2 DM no is infelicitous with thanking and apologizing speechacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1584.3.3 DM no is infelicitous with commissives . . . . . . . . . . 1604.3.4 DM no is infelicitous with declarations . . . . . . . . . . . 1614.4 Why monitor the S-A relation: confirming a bias . . . . . . . . . . 1624.4.1 DM no conveys a contextual bias . . . . . . . . . . . . . . 1624.4.2 DM no conveys an epistemic bias . . . . . . . . . . . . . 1654.4.3 Bias and anchors beyond assertions . . . . . . . . . . . . 1674.5 How this analysis explains other observations: mitigation . . . . . 1694.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1715 Modelling the pragmatics-syntax interface . . . . . . . . . . . . . . 1725.1 Tools for modelling the pragmatics-syntax interface . . . . . . . . 1765.1.1 Exploded CP: CForce>CFinite . . . . . . . . . . . . . . . . 1765.1.2 Beyond CP: ResponseP>GroundAddresseeP>GroundSpeakerP 1785.1.2.1 First steps: The Speech Act Phrase . . . . . . . 1795.1.2.2 Adding a Call-on-the-Addressee: The Ground-ing and Responding Layer . . . . . . . . . . . . 1835.1.3 Integrating DM si and DM no into the clause structure . . 1855.2 (Almost) leftmost and CP-internal: DM si . . . . . . . . . . . . . 1905.2.1 Syntactic distribution: DM si is CP-internal . . . . . . . . 1915.2.1.1 DM si selects for sentence type . . . . . . . . . 1915.2.1.2 Si-clauses are root CPs . . . . . . . . . . . . . . 1925.2.1.3 Si-clauses cannot be negated . . . . . . . . . . . 1935.2.1.4 Si-clauses cannot be conjoined nor disjoined . . 1945.2.2 Linearization: DM si is not leftmost . . . . . . . . . . . . 1955.2.2.1 What DM si precedes . . . . . . . . . . . . . . 1965.2.2.1.1 Topics . . . . . . . . . . . . . . . . . 1965.2.2.1.2 Discourse-oriented adverbs . . . . . . 1975.2.2.2 What DM si follows . . . . . . . . . . . . . . . 202xi5.2.2.2.1 Addressee-oriented discourse activat-ing particles . . . . . . . . . . . . . . 2025.2.2.2.2 Speaker-oriented discourse activatingparticles . . . . . . . . . . . . . . . . 2045.2.2.2.3 Vocatives . . . . . . . . . . . . . . . . 2055.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2065.4 Rightmost and CP-external: DM no . . . . . . . . . . . . . . . . 2095.4.1 Syntactic distribution: DM no is CP-external . . . . . . . 2135.4.1.1 DM no restricts root CPs . . . . . . . . . . . . 2135.4.1.2 DM no does not select for clause-type . . . . . . 2145.4.1.3 No-tagged utterances are undefined relative tonegation . . . . . . . . . . . . . . . . . . . . . 2165.4.1.4 No-tagged utterances are undefined relative toconjunction\/disjunction . . . . . . . . . . . . . 2165.4.2 Linearization: DM no is turn-final . . . . . . . . . . . . . 2175.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2196 Conclusions and outlook . . . . . . . . . . . . . . . . . . . . . . . . . 2226.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2226.2 Comparison with other markers . . . . . . . . . . . . . . . . . . . 2246.2.1 The complementizers si and que . . . . . . . . . . . . . . 2246.2.1.1 Insubordination and DM si . . . . . . . . . . . . 2286.2.2 The DMs no and eh . . . . . . . . . . . . . . . . . . . . . 2316.3 Cross-dialectal variation . . . . . . . . . . . . . . . . . . . . . . . 2346.4 Illocutionary differences between evaluative and non-evaluativeassertions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237xiiList of TablesTable 1.1 Shared and individualized components of a conversation. . . . 15Table 1.2 Felicity of confirmational no with different anchors. . . . . . . 29Table 1.3 Speech act types discussed in this dissertation, based on Searle(1979, 1976) . . . . . . . . . . . . . . . . . . . . . . . . . . . 31Table 1.4 Speech acts, commitment types, and the dialogue elements. . . 32Table 1.5 Searle\u2019s felicity condition types (Searle, 1969). . . . . . . . . . 33Table 2.1 Summary of differences between exclamative and discourse si . 49Table 2.2 Felicity of responding with no to different speech act types. . . 58Table 2.3 Formal differences between phatic and confirmational func-tions of no. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64Table 2.4 Modified from Garc\u0131\u00b4a Vizca\u0131\u00b4no (2005): gradience in the ac-ceptability of no and directives. . . . . . . . . . . . . . . . . . 68Table 2.5 Previous analyses of the use of DM no? (based on Go\u00b4mez Gonza\u00b4lez(2014, p. 114). . . . . . . . . . . . . . . . . . . . . . . . . . . 73Table 2.6 Previous analyses of the use of DM no (based on Go\u00b4mez Gonza\u00b4lez(2014)). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78Table 4.1 Types of speech act and their predicted co-occurrence with DMno. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131Table 5.1 Summary of DM si clause-typing properties. . . . . . . . . . . 192xiiiList of FiguresFigure 1.1 Anchor and DM no. . . . . . . . . . . . . . . . . . . . . . . . 12Figure 1.2 Two possibilities for the nature of CG: as an independent com-ponent (left) or as the intersection of the interlocutor\u2019s com-mitments (right). . . . . . . . . . . . . . . . . . . . . . . . . 16Figure 1.3 The simplest possible model of interaction: A utters p, and pis automatically accepted by B into the CG. . . . . . . . . . . 19Figure 1.4 Two-step model of a dialogue contribution, based on Clarkand Schaefer (1989). . . . . . . . . . . . . . . . . . . . . . . 20Figure 1.5 The Table Model by Farkas and Bruce (2010). . . . . . . . . . 21Figure 1.6 Initial state K1. . . . . . . . . . . . . . . . . . . . . . . . . . 21Figure 1.7 State K2: A utters an assertion. . . . . . . . . . . . . . . . . . 22Figure 1.8 State K3: B accepts the move. . . . . . . . . . . . . . . . . . 22Figure 1.9 State K4: A asks a polar question. . . . . . . . . . . . . . . . 22Figure 1.10 One or two CGs? . . . . . . . . . . . . . . . . . . . . . . . . 25Figure 1.11 How Malamud and Stephenson (2015) analyze RP-tags. Com-ponents with an asterisk (*) represent projected sets. . . . . . 27Figure 1.12 Analysis of Canadian eh in Wiltschko and Heim (2016). . . . 28Figure 1.13 Exploded CP. . . . . . . . . . . . . . . . . . . . . . . . . . . 37Figure 1.14 Proto-neoperformative analysis . . . . . . . . . . . . . . . . . 39Figure 2.1 Schema of the different functions of si. . . . . . . . . . . . . . 43Figure 2.2 Schema of the different functions of no. . . . . . . . . . . . . 56xivFigure 2.3 Anchor and DM no. . . . . . . . . . . . . . . . . . . . . . . . 65Figure 2.4 Two CGs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77Figure 2.5 The Dialogue Board. . . . . . . . . . . . . . . . . . . . . . . 84Figure 3.1 The licensing context for DM si: Two versions of the CG.Crucially, p and q pragmatically contradict each other. . . . . 88Figure 3.2 Phase 0: original alignment in the CG. . . . . . . . . . . . . . 89Figure 3.3 Two possibilities for the nature of CG: as an independent com-ponent (left) or as the intersection of the interlocutor\u2019s beliefsets (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . 93Figure 3.4 Phase 0 (Alignment): both A and B know p according to A. . . 100Figure 3.5 Phase 1 (Misalignment). . . . . . . . . . . . . . . . . . . . . 101Figure 3.6 Phase 2 (Negotiation): A uses a si-clause. . . . . . . . . . . . 101Figure 3.7 Phase 3 (Alignment): both A and B know p. . . . . . . . . . . 102Figure 3.8 Phase 0 (Alignment): Both interlocutors know that A does notspeak Portuguese. . . . . . . . . . . . . . . . . . . . . . . . . 105Figure 3.9 Phase 1 (Misalignment): It becomes apparent that A and B donot share the same CG. . . . . . . . . . . . . . . . . . . . . . 106Figure 3.10 Phase 2: A utters a si-clause. . . . . . . . . . . . . . . . . . . 107Figure 3.11 Phase 3: Both interlocutors know that A does not speak Por-tuguese. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107Figure 3.12 Phase 0 (Alignment): A thinks that both interlocutors knowthat A is the designated driver. . . . . . . . . . . . . . . . . . 111Figure 3.13 Phase 1: It becomes apparent that A and B do not share thesame CG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112Figure 3.14 Phase 2: A utters a si-clause. . . . . . . . . . . . . . . . . . . 112Figure 3.15 Phase 3: A thinks that both interlocutors know that A is thedesignated drived. . . . . . . . . . . . . . . . . . . . . . . . . 113Figure 4.1 Schema of the effect of DM no on the dialogue. . . . . . . . . 130Figure 4.2 Phase 1: Before bare assertion of p=\u2018It is raining\u2019. . . . . . . . 135xvFigure 4.3 Phase 2: Bare assertion of p = \u2018It is raining\u2019. . . . . . . . . . . 136Figure 4.4 Phase 3: After bare assertion of p = \u2018It is raining.\u2019 assumingSpeaker B accepts p. . . . . . . . . . . . . . . . . . . . . . . 137Figure 4.5 Phase 1: Before a modified assertion with DM no. . . . . . . . 139Figure 4.6 Phase 2: Speaker A utters a modified assertion with DM no. . 139Figure 4.7 Phase 3: Acceptance of modified assertion with DM no. . . . 140Figure 4.8 A utters a bare evaluative, judge-independent assertion. . . . . 142Figure 4.9 A utters a bare evaluative, judge-dependent assertion. . . . . . 142Figure 4.10 Phase 1: Before the no-modified assertion of an evaluativeproposition. . . . . . . . . . . . . . . . . . . . . . . . . . . . 145Figure 4.11 Phase 2: Speaker A puts forward a no-modified assertion ofan evaluative proposition. . . . . . . . . . . . . . . . . . . . . 145Figure 4.12 Phase 3: Acceptance of modified assertion of an evaluativeassertion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146Figure 4.13 Phase 2: A utters an exclamation with the DM no. . . . . . . . 149Figure 4.14 Phase 3: Acceptance of no-tagged exclamation. . . . . . . . . 149Figure 4.15 Phase 3(b): Rejection of modified exclamation. p=\u2018It\u2019s a sur-prise.\u2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150Figure 4.16 Speaker A utters a bare command. . . . . . . . . . . . . . . . 151Figure 4.17 Acceptance of a bare command by Speaker B. o=\u2018Study more.\u2019 152Figure 4.18 Phase 1: Before a command modified by DM no. . . . . . . . 154Figure 4.19 Phase 2: Speaker utters command modified with DM no. . . . 154Figure 4.20 Phase 3: Acceptance of command modified with DM no. . . . 155Figure 4.21 Phase 2: Speaker A modified rhetorical question with DM no. 157Figure 5.1 The Complementizer Phrase. . . . . . . . . . . . . . . . . . . 176Figure 5.2 Rizzi\u2019s split CP. . . . . . . . . . . . . . . . . . . . . . . . . . 177Figure 5.3 The Response and Ground layers (Wiltschko and Heim, 2016). 179Figure 5.4 Structure of Speech Act Phrase (SAP) in declarative sentences 180Figure 5.5 Structure of Speech Act Phrase (SAP) in interrogative sentences181xviFigure 5.6 The Response and Grounding layers (Wiltschko and Heim,2016; Wiltschko, 2017) . . . . . . . . . . . . . . . . . . . . . 184Figure 5.7 The extension of the hierarchy of functional categories (\u03ba)in the Universal Spine proposed by Thoma (2016) based onWiltschko (2014). . . . . . . . . . . . . . . . . . . . . . . . . 185Figure 5.8 Position of DM si in ForceP. . . . . . . . . . . . . . . . . . . 186Figure 5.9 Figure from Elvira Garc\u0131\u00b4a (2016, 140; 4.11) shows the syn-tactic integration of DM si in the the clause it introduces. . . . 188Figure 5.10 Proposed syntactic position for DM no. . . . . . . . . . . . . 189Figure 5.11 Position of DM si in ForceP with respect to topics. . . . . . . 197Figure 5.12 Position of DM si in Force with respect to evidential discourseadverbs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198Figure 5.13 Position of DM si in ForceILL with respect to evaluative dis-course adverbs. . . . . . . . . . . . . . . . . . . . . . . . . . 199Figure 5.14 Position of DM si and the speech act level adverb sincera-mente \u2018honestly\u2019. . . . . . . . . . . . . . . . . . . . . . . . . 201Figure 5.15 Position of DM si in Force with respect to speech act discourseadverbs with an intonation break (marked by a comma). . . . . 202Figure 5.16 Proposed syntactic positions for DM si and the Speaker-orienteddiscourse activating particle oye. . . . . . . . . . . . . . . . . 203Figure 5.17 Proposed syntactic positions for DM si and the Speaker-orienteddiscourse activating particle ay. . . . . . . . . . . . . . . . . . 204Figure 5.18 Position of vocatives within the left periphery. . . . . . . . . . 205Figure 5.19 Position of DM si and vocatives. . . . . . . . . . . . . . . . . 206Figure 5.20 Position of si in different projections of the CP. . . . . . . . . 208Figure 5.21 Proposed syntactic positions for DM no. . . . . . . . . . . . . 210Figure 5.22 Movement of the CP to the Specifier of GroundP, followingMunaro and Poletto (2002). . . . . . . . . . . . . . . . . . . . 218Figure 6.1 Position of discourse si and reportative que. . . . . . . . . . . 227Figure 6.2 Parallelism between the two Spanish complementizers. . . . . 228xviiFigure 6.3 si-clause as peripheral adverbial. . . . . . . . . . . . . . . . . 230Figure 6.4 Speaker utters a modified command with DM eh. . . . . . . . 233xviiiAbbreviations and definitionsList of abbreviations used:CG Common GroundCP Complementizer PhraseCOND ConditionalDIM DiminutiveDC Discourse CommitmentsDM Discourse MarkersFORM FormalFUT FutureIMP ImperativeIMPER (Past) ImperfectIMPRS ImpersonalIND IndicativeNEG (Sentential) NegationPST PastxixPL PluralPREP PrepositionPRS PresentREFL ReflexiveSG SingularSBJV SubjunctiveSUPER SuperlativeTDL To-Do-ListList of the definitions of concepts that will be used in this disser-tation:Addressee The person to whom an utterance is directed.Alignment A state where different elements are in a symmetrical position rel-ative to each other. In this dissertation I focus on epistemic alignment: asymmetrical arrangement of the information states of the discourse partici-pants.Anchor The linguistic unit to which an appended element (tag) attaches.Commitment A speaker commits to a proposition when they add it to their Dis-course Commitments\/Common Ground; a speaker commits to an outcomewhen they add it to their To-Do-List\/Common To-Do-List.Commitment (public) A commitment is public when all interlocutors are awareof it.Commitment (shared) A commitment is shared when all interlocutors agree withit.xxConversation Type of discourse. Verbal or manual exchange between two lan-guage users who take turns being the Speaker\/Signer or the Addressee.Discourse Since conversations are the only type of discourse that I will focus on,I will sometimes use \u2018conversation\u2019 and \u2018discourse\u2019 interchangeably.Discourse Move An action (verbal, signed, or gestured) taken by an interlocutor.A discourse move has a purpose and has consequences for future discourse.Discourse Phase A particular moment in discourse.Information State Consists of the beliefs and outcome preferences of each inter-locutor.Interlocutor(s) Conversational partners that take turns in a conversation.Outcome The commitment type of imperatives: a consequence of an action (ex-pressed by the imperative).Proposition The denotation of declarative sentences; \u201cthe sharable objects of theattitudes and the primary bearers of truth and falsity\u201d(McGrath and Frank,2018)Speaker Speech Act participant who holds the floor at a given moment in a con-versation and makes a move.Speech Act A verbal or signed act by which a language user performs an action.Speech act participant Speaker and Addressee.Turn \u201ca time during which a single participant speaks, within a typical, orderlyarrangement in which participants speak with minimal overlap and gap be-tween them.\u201d (Levinson, 1983)Utterance The smallest complete communicative unit; a sentence used in a par-ticular context.xxiAcknowledgmentsMay this be a warning to the reader: this is the only part of the dissertation thathas not been heavily edited, both in terms of grammatical errors and it terms ofdramatic flare.Whenever I felt at a breaking point I would imagine how it would feel to sub-mit the dissertation with excruciating detail, how it would feel to defend, how itwould feel to submit the final version of the dissertation. I even imagined whattype of perfume I would be wearing. It helped me keep going. Out of all the(imagined) situations I made up, never in a million years could I have predictedthat all these steps would happen during a global pandemic. So the following peo-ple who I will be thanking not only helped me through a regular tough period, butduring an particularly tough period.First and foremost, I would like to thank my committee: my research super-visor Rose-Marie De\u00b4chaine, Hotze Rullmann, and Martina Wiltschko. They havehad immense patience, and been extremely supportive. Every conversation withRose-Marie felt like a supernova had exploded in my brain. She would alwaysbring up bigger and more exciting issues that I had foreseen. She was alwayslooking out for me and my health, too, and I am truly gratefully for that. I want tothank her also for always trusting me as a researcher\u2013even when I didn\u2019t. Hotzehas helped me so much during grad school that I don\u2019t even know where to start.During my first years I went through a very tough depression, and although I nevermentioned it by name he helped me through my qualifying paper period by giv-xxiiing me exactly what I needed at that time: small deadlines, celebrations of smallachievements, and constant support. His questions always pushed me to be morethorough: if there is any sign of rigour in my argumentation it is thanks to him.Martina Wiltschko has been a constant inspiration, both at a personal level and ata professional level: she leads her life with courage, and she is never shy to thinkoutside the box.I also couldn\u2019t have asked for better university examiners at my defence: LisaMatthewson and Jon Beasley-Murray. I am so gratefully for their questions andfeedback. And last but surely not least, I could have never imagined that DonkaF. Farkas would be my external examiner: I have admired her work for a longtime, and it inspired so much of my own work. Thank you for your feedback andsupport, and for joining us during the virtual defence.I want to thank everyone at the Department of Linguistics at UBC for theirconstant support and encouragement. I cannot mention every grad student here,but I will always remember the supportive network we built.I am so lucky to have a support network far away physically but not in spirit:my friends back in Spain (and now all over the world). They have kept tabs on mesince high school, and supported me via emails and messages, and they alwaysmade (and make) me laugh: Adriana, Almu, Jojo, Judit, Nela, and Rita. Cristinaand our annual emails is something I always looked forward to. Larissa, alwaysa ray of sunshine when I most needed it. Rudi, who is fearless and stays positivewhatever happens. We were far away, but I could feel their presence from afar.In Vancouver I have also been extremely lucky to have an amazing supportsystem. I want to thank the Eh Lab for constant intellectual stimulus. I want tothank Samuel Akinbo for always making me laugh; Joash Gambarage for sharinghis language with me throughout the years and for being an inspiration for how tohelp a community. Michael Schwan has also helped me with his (maybe uninten-xxiiitional) humour. Raiane Salles, a fighter both academically and personally. MeganKeough, my neighbour and inspiration for so many aspects of my life, I can neverlisten to \u201cLet\u2019s go crazy\u201d by Prince without thinking about you and Julian. Jo-hannes Heim: I took to heart your words at your defence party \u201cGetting a PhD isnot a sign of intellect, but of persistence\u2019\u2013thank you for for rigorous scholarshipand your constant support. Emily Sadlier-Brown has everything a true scholarshould have: creative ideas, inquisitive mind, and a love for beer\u2013she is destinedto do great things. I am so lucky to have Neda Todorovic as a friend, she is a sunnypresence wherever she goes, we have been so blessed to have her as a researcherand teacher (and friend!) in Vancouver.I cannot put into words how much I owe to Oksana Tkachman and HermannKeupdjio. Oksana is probably the first person I tell any good or bad thing thathappens to me, and she has kept me sane so many times. She is such an amazingresearcher and intellectual, and I know (not only believe) that she will change thestatus quo of linguistics. What can I say about Hermann Keupdjio: as he wouldsay, we are basically twins. I remember our conversations on the bus 33, whenthings were dark and we were losing hope, and we would tell each other to hangin there. Now he has a wonderful wife, Christelle, and an amazing son, Nolann;he has such a bright future ahead, and will shake the foundations of theoreticalsyntax!If I managed to keep relatively sane during these last stages of the dissertation,it\u2019s because I had Alex by my side. Whenever I had spent hours trapped in mybedroom writing like a hermit, and I was breaking down, he would make sure thatI was okay and that I would eat and sleep. His unwavering love and patience havebeen my shelter in a tumultuous vortex. Thank you for letting me and Chubbito(my cat, who has also helped me through everything, but it will take me too manypages to acknowledge) into your life.And finally, I want to acknowledge that without my family, I wouldn\u2019t be here,xxivin any shape or form. My friend Oksana has mentioned some times that there issome sort of happiness that radiates off of me, and that it probably comes fromhaving had a warm childhood and close-knit family. I am so extremely lucky tohave the family that I have: my father, Joaqu\u0131\u00b4n, and my mother, Mercedes, whokeep sending me care packages with little notes and gossip magazines. My fatherhas taught me the value of a good vegetable garden: patience, nurturing, and neverforgetting to add some rose bushes even if they are not edible, just for the beautyof the flowers. I\u2019ve taken that to heart. My mother has taught me the power ofwords through literature, and through art; and has also taught me the power ofeducation to fight inequality. I have taken that to heart, too, and hope to continueher teachings in my own career. My brother Luis, who became a father to myniece Sof\u0131\u00b4a during my time away: I learned so much from him, from the nightswhen I couldn\u2019t sleep as a child and he would play guitar for me, to when hetaught me about the three laws of thermodynamics at the pool one summer (I willnever forget actio-reactio!). I wouldn\u2019t have pursued a PhD if I didn\u2019t have him asan example of a true scholar and intellectual. My brother Juan has been my lightand laughter throughout the years, but even more so these last few months. Hehas taught me to always care: about others, and about myself. Most of the bestintrospection comments in this dissertation come from him: he truly knows howpeople relate to each other\u2013something we sometimes forget in Academia. Therehas never been a situation when he couldn\u2019t make me laugh and make me feel athome, even far away. I wouldn\u2019t be the happy person I am nowadays without him.\u00a1Gracias!xxvTo my parents, Joaqu\u0131\u00b4n and Mercedes, for giving me language.And to my brothers, Luis and Juan, for teaching me how to (mis)use it.The inferno of the living is not something that will be; if there is one,it is what is already here, the inferno where we live every day, that weform by being together. There are two ways to escape suffering it. Thefirst is easy for many: accept the inferno and become such a part of itthat you can no longer see it. The second is risky and demands con-stant vigilance and apprehension: seek and learn to recognize whoand what, in the midst of inferno, are not inferno, then make themendure, give them space.\u2014 Italo Calvino, Invisible CitiesxxviChapter 1Introduction- My Dad just came into my room and shouted at me in Klingon. Am Imore embarrassed that he did that or that I know he said I was adisappointment to the empire?- You should be most embarrassed that you\u2019re a disappointment to theempire. \u2014 an Internet user discovering the different dimensions of aconversation.Conversations can be exhausting. Anyone who has had a job interview (ora conversation with a parent-in-law they just met) knows that conversations areso much more than an exchange of information conveyed simply by uttering onesentence after another. During a conversation, language users monitor not onlypropositional information, but sources of information, certainty of information,the emotional state of the interlocutors, even\u2013as the quote at the beginning of thischapter illustrates\u2013the form of the message itself as opposed to the information itconveys. Most importantly, interlocutors want to stay \u201con the same page\u201d: theywant to stay aligned (Clark and Marshall, 1981; Sacks, 1987; Pickering and Gar-rod, 2004; Farkas and Bruce, 2010). Alignment between interlocutors can occurat different linguistic levels, including phonetic, phonological, lexical, syntactic,and semantic (Pickering and Garrod, 2004).In this dissertation, I focus on a linguistic strategy by which interlocutors stay1aligned at the epistemic level, namely by using Discourse Markers (DM).1 I defineepistemic alignment as a symmetrical arrangement of the information states of thediscourse participants, where the information states comprise their beliefs andtheir outcome preferences. I explore two Spanish DMs that are licensed in twocontexts where a Speaker wants to maintain epistemic alignment:i The particle si (DM si henceforth) is used when the Speaker wants to flagepistemic misalignment between the interlocutors. This arises when inter-locutors do not share the same background information as reflected by in-formation conflict in the Common Ground (CG henceforth). An example ofDM si as it is used in Peninsular Spanish is given in (1).ii The particle no (DM no henceforth) is used when the Speaker is trying to as-certain whether there is epistemic alignment between the interlocutors. Thisarises when Speaker wants to confirm the Addressee\u2019s information state. Anexample of DM no as it is used in Peninsular Spanish is given in (2).2(1) Context: A and B are discussing weekend plans. B tells A that they will goto the beach. A says:Sisihacedoesunafr\u0131\u00b4ocoldquethatpela!peels\u2018[Si] it\u2019s freezing!\u20191I have expanded my definition of epistemic in this dissertation to encompass both beliefs andpreferences, although the latter are not typically defined as epistemic.2For ease of reading, I will not add a question mark after no in the body of the text, and I willnot use the initial question mark \u2018\u00bf\u2019 that is used in Spanish orthography.2(2) Context: A and B are relaxing on a patio, enjoying a drink. A says to B:Que\u00b4whatbienwellquethatseIMPRSesta\u00b4isaqu\u0131\u00b4,hereno?no\u2018It is so nice here, [no?] \u2019I propose that these DMs operate at different levels of the conversation: someDMs aim to create informational links, and other DMs aim to have an effect onthe information state of discourse participants (similar to the intentional level inMoore and Pollack (1992))3. I will operationalize these levels to mean that someDMs (like DM si) act at the Common Ground level, other DMs (like a DM no) actat the speech act level. The former links the proposition denoted by the sentence itappears in to the previous discourse, while the latter confirms with the Addresseethe appropriateness of the use of a speech act.The fact that different DMs operate at different levels of the conversation is re-flected in the syntactic structure. DMs have been assigned positions within the leftperiphery of the clause structure, more specifically as part of the ComplementizerPhrase (CP) (Munaro and Poletto, 2002; Zimmermann, 2009; Paul, 2009; Cardi-naletti, 2011). The Complementizer Phrase is the syntactic layer that codes \u201ctheinterface between a propositional content (expressed by the IP) and the superor-dinate structure (a higher clause or, possibly, the articulation of discourse, if weconsider a root clause)\u201d (Rizzi, 1997, p.283); however, this position only conveysthe function of DMs like si, that operate at the informational level. For speechact-level DMs like no, recent analyses in the literature have proposed to extendthe structure of the clause to include discourse and pragmatic information (Speasand Tenny, 2003; Emonds, 2004; Haegeman and Hill, 2013; Wiltschko and Heim,2016).This dissertation contributes to the literature on DMs by proposing a semantico-pragmatic and syntactic analysis of DM si and DM no. These two DMs were3They refer to the mental state of the discourse participants, instead of the information state.3chosen because they seem to encode polar opposites of each other: while DMsi is used as a corrective step (as we have seen in (1)) directly challenging theAddressee\u2019s previous contribution, DM no requests a contribution from the Ad-dressee, including them in the construction of the dialogue. DM si will be an-alyzed as an illocutionary operator, marking that the proposition denoted by thesentence it precedes is already part of the CG. Because of its nature as an illocu-tionary operator, as well as its integration in the clause, I propose that it sits at thehighest projection of CP. On the other hand, DM no\u2019s contribution is to requestfrom the Addressee to confirm that a conversational update can be made publicand can be shared. Since it targets the Addressee\u2019s information state, I position itin a syntactic layer above CP that encodes speech act-level information.1.1 A tale of two discourse markersBefore I introduce the main protagonists of this dissertation and build up the the-oretical setting within which I\u2019ll construct my analysis, let us ask a more generalquestion about DMs: why do we so often use something that is not, strictly speak-ing, necessary to convey the main message? One of the primary characteristics ofDMs is, in fact, that they are optional. This brings up what Degand (2019) calls\u201cthe paradox of discourse markers\u201d: DMs are highly frequent, and yet they arenot strictly required.4 And although they might not be obligatory, they are nec-essary\u2013something that is clear when language users mention that not using DMsmakes them sound \u2018like robots.\u2019 Let us consider the dialogue in (3), taken fromGras and Sansin\u02dcena (2017, p.27; ex. 17):(3) J01:J01:Oyelistencue\u00b4ntametell.meloitdelofLuisLuis(.)(.)loitdelaboutbesokiss\u2018Hey, tell me about Luis, about the kiss\u20194At least in the languages researched so far. There might be a language in the world whereDMs are obligatory.4J02: No\u2018No.\u2019J01:J01:puchadamnqueCOMPsonbe.IND.PRS.3PLmalasbad-F.PL\u2018Damn, how bad you are!\u2019 (taken from Gras and Sansin\u02dcena (2017, p.27; ex.17) (Chilean Spanish)).If we strip from this dialogue what we can identify as DMs (boldfaced above),the result in (4) is, intuitively, harsher: in the first turn by J01, without the openingDM oye \u2018hey\u2019, the command to tell a story about Luis seems too forward as theopening turn, and the second turn by the same J01 is less expressive, and in asense \u2018meaner\u2019. The core message is still there, though: J01 orders J02 to tell astory, J02 refuses, and J01 accuses them of being mean.(4) J01:J01:cue\u00b4ntametell.meloitdelofLuisLuis(.)(.)loitdelaboutbesokiss\u2018Tell me about Luis, about the kiss\u2019J02: No\u2018No.\u2019J01:J01:sonbe.IND.PRS.3PLmalasbad-F.PL\u2018You are bad.\u2019DMs may not convey the core propositional message of an utterance, but theyestablish how that core message is linked to other parts of the message (illustratedin (3) by the opening oye \u2018hey\u2019) and to the interlocutors (illustrated in (3) by the5expressive pucha \u2018damn\u2019). This linking function is not the essence of the messageper se\u2013but it is, in a sense, the essence of communication. This \u2018peripheral\u2019 role ofDMs has been formalized in different ways\u2013using the distinction between propo-sitional and non-propositional meaning, conceptual and procedural meaning, orat-issue and not-at-issue, for example, with DMs conveying non-propositional ornot-at-issue meaning\u2013but the idea is the same: DMs are not needed to convey whatis usually conceived of as the most important part of a message: its propositionalinformation. DMs are in this sense similar to information structuring notions suchas topic and focus: they are not obligatory, but they help package the informationconveyed in a given proposition.DMs are a notoriously hard category to define (see Crible (2017) for an ex-tensive discussion of the terminology): even the term itself is the source of muchdebate, since it is sometimes used interchangeably with other (similar) terms likepragmatic markers or discourse particles.5 DMs form a heterogeneous group:some examples that have been studied as DMs include interjections (Oh!), con-junctions (and), adverbs (frankly), phrases (y\u2019know), verbs (listen), and modalparticles (Standard German ja, doch, wohl). Much of the literature defines themin negative terms: they do not convey truth-conditional meaning, they lack inflec-tional endings, and many of them do not seem to display much syntactic move-ment (Lewis, 2006; Biberauer et al., 2014).6 They are usually grouped togetherbecause they seem to share a common abstract function: linking propositionalcontent to the context. More abstractly, their main defining role is to establishrelationships between units of discourse.Some analyses give a coherence-based explanation of the use of DMs, accord-5I will assume in this dissertation that DMs are a superset of pragmatic markers and discourseparticles.6It is not entirely true that all proposed DMs lack inflectional endings, especially those thatmark agreement with the Speaker or the Addressee (allocutive agreement), such as Upper Aus-trian German goi\/goins\/goits, which changes depending on whether the Addressee is addressedformally, informally, or is there is more than one Addressee (Wiltschko and Heim, 2016). We willsee more examples in Chapter 5.6ing to which they are used by the Speaker to guide the Addressee in assigning thecorrect coherence relations between units of discourse, which can be as large asparagraphs (Sanders and Noordman, 2000). A framework that has given specialattention to the study of DMs is Relevance Theory (Wilson and Sperber, 1990,1999, 2002). In analyses that follow this framework, DMs are viewed as ele-ments that code instructions that help make inferences about the meaning of theutterance\u2013using their terminology, they encode procedural meaning (Blakemore,2001).7 DMs favour the inferential process by pointing explicitly at how codedrepresentations should be connected with information that is already part of thecontext (Blakemore, 2004). Sometimes this includes the Addressee, and thereforemany DMs are said to mark interpersonal contact (Briz, 1998).DMs have also been analyzed as elements that facilitate grounding, a conceptdeveloped by Clark and Schaefer (1989); Clark and Brennan (1991): interlocu-tors want their contributions in a conversation to be understood, and thereforetry to connect them to what is already shared\u2013they ground the utterances in theconversation. This in turn is similar to the idea of Common Ground (CG) Man-agement in the sense of Krifka (2008): some DMs concern themselves with theway the shared (CG) content should develop. In other words: they do not encodetruth-conditional information, but link the truth-conditional information of the ut-terances they appear in to the rest of the conversation.We see therefore that, broadly speaking, there are two types of approachesin the study of DMs: those which look at the relationship between the clausemodified by a DM and its relationship with the rest of the discourse, and thosewhich look at the relationship of the DM to the clause it appears in. One of themain take-away messages of this dissertation is that these are two sides of thecontribution of DMs to the discourse: (i) relating the clause they appear in to therest of the discourse, and (ii) their own specific contribution to the clause itself.These two sides are present in the analyses I offer for the DM si and DM no.87As opposed to conceptual meaning.8The purpose of the analysis presented here is to bring together these two pieces (a speech-act-71.2 The protagonists: DM si and DM no1.2.1 The \u2018rebel\u2019 complementizer: the Peninsular SpanishDM siThere is growing interest in the Romance literature in general and the Spanishliterature in particular about what Corr (2018) calls \u201cillocutionary complemen-tizers\u201d\u2013the use of the complementizers que (\u2018that\u2019) and si (\u2018if, whether\u2019) in rootclauses, as in (6). These two complementizers typically head embedded clausesas in (5): in (a), we see the declarative complementizer que \u2018that\u2019 introducing adeclarative embedded clause, whereas in (b) we see si introducing an interrogativeembedded clause:(5) a. MemedijosaidquethatelthegatocatdeofAdrianaAdrianaesisnegro.black.\u2018She told me that Adriana\u2019s cat is black.\u2019b. NoNEGse\u00b4knowsiifelthegatocatdeofAdrianaAdrianaesisnegro.black\u2018[Si] Adriana\u2019s cat is black.\u2019However, que and si can also be found in root clauses, as the examples in (6)show. The notable thing in (6) is that we find the complementizers que and si inroot clauses, as opposed to embedded clauses as seen in (5):driven analysis and a propositional-driven analysis), but it might be the case that only one of thosepieces may successfully explain all the data presented here. A discussion of these two options, anda potential simplification of the analysis is a future step in my research program.8(6) a. Context: For Adriana\u2019s birthday, A and B have drawn a fancy birthdaycard with a drawing of Adriana\u2019s cat, which they think is white. A findsout that the cat is, in fact, black, but they already gave the card to Adri-ana. A says to B:Tu\u00b4,youquethatelthegatocatdeofAdrianaAdrianaesisnegro.black.\u2018Hey, Adriana\u2019s cat is black.\u2019b. Context: B says they will never borrow black clothes from Adriana be-cause she has a cat and his fur will be visible on the black clothes. So Asays:SisielthegatocatdeofAdrianaAdrianaesisnegro.black\u2018[Si] Adriana\u2019s cat is black.\u2019Not only do the complementizers que and si appear in root clauses in (6a)and (6b), but que in particular can appear multiple times in the same utterance, inseveral syntactic positions within the CP (Demonte and Ferna\u00b4ndez Soriano, 2009;Villa-Garc\u0131\u00b4a, 2015, 2019):(7) QuethatdicesaysAdrianaAdrianaquethatsuhergatocatquethatesisnegro.black\u2018Adriana says that her cat is black.\u2019On the other hand, si does not appear to be as ubiquitous as que, and cannotappear in as many positions as que in (7). However, si can appear in contextswhere instead of linking the clause it heads to the main clause (that is, insteadof linking a non-root clause to a root clause), it links a root clause to a previouselement within the immediate discourse, as we have seen illustrated in (6b). There9are, in fact, two types of constructions were si appears in root clauses: exclama-tives, as in (8a), or in constructions that seem to correct something in the previouscontribution, as in (8b):(8) a. Sisisera\u00b4be.FUT.3SGtonto!dumb\u2018He is SO dumb!\u2019b. Context: B is surprised to hear that an acquaintance is pregnant. A says:Sisiloitsabeknow.PRS.3SGtodoallelthemundo.world\u2018[Si] everybody knows.\u2019This last type of construction is part of the focus of this dissertation. Mostof the literature on this type of clauses highlights their adversative nature (Mon-tol\u0131\u00b4o Dura\u00b4n, 1999; Schwenter, 2000b; Rodr\u0131\u00b4guez Ramalle, 2011) and they areoften said to refute the previous discourse move. However, as Schwenter (2016b)points out, these utterances can be used in situations where at first glance theSpeaker is not correcting the Addressee. This is the case in (9), where the Speakeris in fact agreeing with the previous contribution to the discourse:9(9) Context: B keeps reminding A to be confident if she wants to succeed.S\u0131\u00b4yess\u0131\u00b4,yessisiyaalreadyloitse\u00b4.know.PRS.1SG\u2018Yes, yes, I know.\u20199This example includes an instance of the polarity and response particle s\u0131\u00b4 \u2018yes\u2019. This particleis distinct from the si we are discussing in this thesis both in terms of stress (marked here ortho-graphically), possibility of standing alone or having a prosodic break after it (only stressed s\u0131\u00b4 cando both), as well as etymological origins.10I will refer to these instances as si-clauses following Schwenter (2016a,b), andto the particle itself as DM si given its discursive function. I will present an anal-ysis of DM si that has two main ingredients: (i) the contribution of the si-clauseto the discourse, and (ii) the contribution of the DM si itself to the clause. Thesetwo ingredients are, however, linked and not entirely separable: the contributionof DM si to the sentence will constrain the contribution of the clause as a whole tothe discourse. Regarding the contribution of DM si to the sentence it appears in,I propose that DM si marks that the proposition denoted by the clause it precedesis already in the CG. By re-asserting a proposition the Speaker brings \u2018back\u2019 aproposition from the CG\u2013the effect of this is the \u2018you should have known this\u2019flavour that has been noted in the literature on DM si. Regarding the contributionof the si-clause as a whole, I propose that it is used to reject the previous discoursemove based on its not satisfying felicity conditions. Both ingredients will be re-flected in the syntactic analysis proposed for DM si in Chapter 5: what I call itssuperassertive nature will come from its syntactic position as the head of the high-est Force projection, ForceP; its use to link the clause it heads with the previouscontribution comes from the fact that these constructions are discourse anaphoric:they need a discourse antecedent in order to be interpreted (Cook, 2008).1.2.2 The negotiator: Peninsular Spanish DM noThe form [ANCHOR + no?] (where the anchor is the sentence that precedes DMno) in Peninsular Spanish has been referred to in the literature as a \u201ccomprobative\u201d(comprobativo) marker (Ortega, 1985), an interactive marker (Briz, 1998), a mit-igating device (Fe\u00b4lix-Brasdefer, 2004b,a), a subjective marker (Mo\u00b4ccero, 2010),and as a politeness strategy (Garc\u0131\u00b4a Vizca\u0131\u00b4no, 2005; Rodr\u0131\u00b4guez Mun\u02dcoz, 2009).11Figure 1.1: Anchor and DM no.(10) CONTEXT: B\u2019s cat is meowing and waiting next to his bowl. A tells B:Dalegive.IMP.2SG=himlathecomidafoodya,alreadyno?no\u2018Give him his food, [no?]\u2019Researchers sometimes study DM no in contrast to other confirmational mark-ers, especially eh (Garc\u0131\u00b4a Vizca\u0131\u00b4no, 2005; Montan\u02dcez Mesas, 2015), or in contrastto English tag questions (Go\u00b4mez Gonza\u00b4lez, 2014). Regardless of the frameworkof study, most studies define this construction as consisting of an anchor, followedby the negative particle no with rising intonation. The DM (or tag) is integratedinto the intonational contour of the anchor (Cabedo Nebot, 2013), especially infast speech, although an intonational break between the anchor and the DM ispossible.This dissertation will contribute a systematic description of the possible an-chors of DM no, which will be crucial for the analysis of the contribution of theDM to the discourse structure. I will offer an analysis that encompasses the mean-ings attributed to DM no as a mitigating, interactive, confirming marker: addingthe DM no requests confirmation from the Addressee that the conversational up-date proposed by the Speaker is shareable.121.3 The setting: conversational updatesConversations are sometimes modelled as a type of negotiation or a game (Lewis,1979; Ginzburg, 1996, 2012; Farkas and Bruce, 2010). The players are the in-terlocutors, and they keep score of what has been said, and who said it. Formalmodels of conversation have exploited this depiction of dialogue and have referredto their models as a \u2018scoreboard\u2019 or \u2018gameboard\u2019 (Lewis, 1979; Ginzburg, 1996;Farkas and Bruce, 2010; Ginzburg, 2012; Malamud and Stephenson, 2015). Keep-ing score of what is said is useful in terms of referring back to a piece of informa-tion or entity. The importance of keeping score of who said what is a more recentupdate to the traditional scoreboard model, its importance highlighted by researchon dialogue that includes perspectival elements like taste predicates (Malamudand Stephenson, 2015) and that focuses on the idea of dialogue as a way to ex-press varying degrees of commitment to a proposition (Gunlogson, 2008; Farkasand Bruce, 2010; Krifka, 2015; Rudin, 2018; Heim, 2019a). All this informa-tion (who said what) is registered in a scoreboard\u2013which I will call the DialogueBoard\u2013as a way to keep track of the information that sustains the dialogue.But interlocutors \u2018use\u2019 each other for more than tracking and keeping score.They may rely on each other to fully commit to the truth of what they are saying,what Gunlogson (2008) calls \u2018contingent commitment\u2019: the Speaker marks thatshe needs the contribution of the Addressee in order to add the proposition to theCG. There are various linguistic mechanisms that have been analyzed as contribut-ing to this interaction between interlocutors, among others intonation (Gunlogson,2008; Farkas and Roelofsen, 2017; Rudin, 2018; Schlo\u00a8der, 2018; Heim, 2019a)and discourse particles and markers (Reese and Asher, 2007; Zimmermann, 2011;Lam, 2014; Thoma, 2016; Wiltschko and Heim, 2016; Heim, 2019b; Jamieson,2019).In this section, I introduce the different components that form the DialogueBoard, divided into whether they are shared between the interlocutors, or whetherthey are individualized for each discourse participant. Different speech acts willtarget different components through specific effects on the Dialogue Board (Farkas13and Bruce, 2010): knowing the mechanics of simple (canonical) conversationalupdates will help us develop what happens during the complex conversational up-dates that DM si and DM no propose.1.3.1 What gets updated in a conversationOne big question when studying any kind of cooperative enterprise (such as aconversation) is whether interlocutors are accessing a mutually shared set of in-formation or, in contrast, whether they are only accessing information that is ac-cessible to themselves. Specifically for conversations, the distinction betweenshared and individual components of a conversation is central to explaining theeffects of different speech acts (Gunlogson, 2008; Farkas and Bruce, 2010; Mala-mud and Stephenson, 2015; De\u00b4chaine et al., 2017): for instance, a speech act likea promise would only affect an individualized component in that it would add theoutcome of the promise to the Speaker\u2019s \u201clist\u201d of actions she has committed to.10Farkas and Bruce (2010) claim that language users want to expand the beliefs thatwe have in common. In their model (which I will call the Table Model followingRudin (2018)), they distinguish between a shared set of beliefs (the CG) and twoindividualized sets of beliefs (Discourse Commitments (DC)): the driving force ofconversation is to move propositions from the DCs to the CG (Stalnaker, 2002;Farkas and Bruce, 2010; Rudin, 2018).Rudin (2018) proposes the same split for commands: he proposes a shared To-Do-List (TDL) together with individual TDLs for each speech act participant. Thisresults in a Split-Table Model, with one doxastic part (including DC\/CG) and oneteleological part (including TDLs). Each part consists of a shared element (eitherthe CG or the TDL), and two individualized elements (DCA\/B or TDLA\/B). Thismodel, (schematized in Table 1.1), is the basis that I will work with. 1110One could argue that a promise should be public in order to be a felicitous promise. Notethat the fact that something is individualized does not mean that it is not public, only that thecommitment to it is individual.11In Rudin\u2019s model, he defines the contents of the TDL as preferences. Yang and Wiltschko(2016) talk about intentions. I chose outcomes following Beyssade and Marandin (2006), but all14Doxastic TeleologicalShared Common Ground To-Do-ListIndividualized Discourse Commitments (DCA\/B) To-Do-List (TDLA\/B)Table 1.1: Shared and individualized components of a conversation.In the next sections I will describe these shared and individualized compo-nents of the conversation, focusing on the Common Ground (section 1.3.1.1), thecommon To-Do-List (section 1.3.1.2), the individualized Discourse Commitments(section 1.3.1.3) and the individualized To-Do-Lists (section 1.3.1.4). Both thekinds of information they contain, and the fact that they contain either shared orindividualized commitments is a key component of the analysis presented in thisdissertation.1.3.1.1 Shared: The Common GroundConversations are said to happen against a background of shared information(Schiffer, 1972; Stalnaker, 1978; Clark and Marshall, 1981; Clark and Carlson,1981; Lee, 2001; Farkas and Bruce, 2010). This background is ever-changing: asthe conversation moves forward, more information is added to it. We can refer tothis shared background of information as the Common Ground (CG). It representsthe shared, public commitments of the interlocutors during a conversation. Theseare the two characteristics of the CG: it is shared, and it is public. It is shared,because every interlocutor in the conversation is committed (at least publicly) tothe truth of the propositions contained in CG. It is public, because interlocutorsactually work together in order to build the CG. In fact, it is often stated that oneof the main driving forces of a conversation is to add propositions to the CG.The CG can be viewed as an element on its own or as a result of intersect-ing components depending on the approach taken (Gunlogson (2008); Farkas andBruce (2010); Thoma (2016); Wiltschko and Heim (2016), a.o.). The two possi-bilities are sketched out in Figure 1.2.these terms are meant to describe more or less the same concept.15Figure 1.2: Two possibilities for the nature of CG: as an independentcomponent (left) or as the intersection of the interlocutor\u2019scommitments (right).Proponents of the intersection possibility claim that the CG is the result of theintersection between the individual commitments of the interlocutors: this is thesituation sketched in the rightmost part of Figure 1.2 (Gunlogson, 2004, 2008;Thoma, 2016; Wiltschko and Heim, 2016). Whatever proposition is in this in-tersection, it will be common between all interlocutors (in this figure a dyad isassumed for simplicity), and hence it will be considered shared knowledge. Onthe other hand, proponents of having a CG that is separate from any other indi-vidual components use this distinction to capture the effect of assertions on theconversation: adding the asserted content to the CG (Farkas and Bruce, 2010).This dissertation will assume that the CG is a separate component from indi-vidualized discourse commitments, following Farkas and Bruce (2010) and sub-sequent work that uses the model they propose (Malamud and Stephenson, 2015;Rudin, 2018). The CG will include \u201cthe set of propositions that have been agreedupon by all participants (...) together with the propositions that represent theshared background knowledge of the discourse participants.\u201d (Farkas and Bruce,2010, p. 85). This assumption (both on the conception of the CG, and the contentof the CG) is made based on two pieces of evidence from the DMs discussed inthis thesis. Regarding the independent status of the CG, the use of DM no showsthat some complex discourse moves (like requesting confirmation of an assertion)are used specifically to mark that the Speaker wants to move a proposition fromthe individual components to the shared CG. Although this move is the default16for (unmodified) assertion according to Farkas and Bruce (2010), it is made ex-plicit by the use of this DM. Regarding the content of the CG, the use of DMsi in contexts where a felicity condition has been violated shows that this typeof background information on conversational principles needs to be stored (andaccessible), and I will assume that the CG allows us to model this.1.3.1.2 Shared: Common To-Do-ListRudin (2018) proposes to have a similar (publicly shared) element to CG, which isa reservoir of outcomes or preferences: a common To-Do-List (TDL). The notionof TDL comes from Portner (2004), which he defines as a set of functions fromindividuals to sets of properties. He distinguishes, however, two TDLs: one forthe Addressee, one for the Speaker.The role of the shared TDL is the teleological counterpart of the CG: theshared TDL is a publicly shared set of outcomes which includes the outcomesto which each interlocutor (A and B) have publicly committed. I will use thiselement to account for the fact that DM no can be used with imperative anchorsto confirm with the Addressee that they are publicly committed to an outcome.1.3.1.3 Individualized component: discourse commitmentsDistinguishing shared from individualized commitments is important for situa-tions where the truth of a proposition depends on a judge (Lasersohn, 2005;Stephenson, 2007). This situation is exemplified in (11a): when A utters (11),she is saying that the party is fun according to her perspective, a reading that canbe made explicit by adding to me, as in (11b):(11) a. This party is fun.b. This party is fun to me.In (11) it is the personal taste predicate that is responsible for this perspective.Note, however, that (11a) might also have a reading where the Speaker is trying toconvey that the party has the property of being fun\u2013not only from her perspective,17but from a general perspective. In this case, the Addressee can refuse to acceptthat the judgement (that the party is fun) is shared, as the dialogue in (12) shows.Speakers can then \u201cagree to disagree\u201d: A thinks the party is fun, while B thinksthat the party is not fun.(12) A: This party is fun.B: It isn\u2019t, I hate it here!A: Well, I\u2019m having a great time!In order to account for this type of example, Farkas and Bruce (2010) proposethat a sentence like (11) can propose to add the proposition p=\u2018This party is fun\u2019to either the CG or to A\u2019s particular discourse commitments, DCA, which includecommitments which only one of the interlocutors has committed to. If the Ad-dressee does not accept an update of the CG, the proposition will not be shared,but it will stay in DCA.I will refer to this type of sentences as evaluative assertions, as opposed tonon-evaluative assertions, which by default propose to add a proposition to theCG. Other terms include subjective versus objective assertions (Rudin and Bel-trama, 2019) or subjective versus factual assertions (Beltrama, 2018). Note that inthese analyses they are all assumed to have assertive force, that is, that the Speakeris still ultimately trying to update the CG\u2013this falls out from the assumption inFarkas and Bruce (2010) that moving propositions from the discourse commit-ments to the CG is one of the main forces that drive conversations.121.3.1.4 Individualized component: To-Do-ListsWhen a Speaker puts forward a command like in (13), the Speaker is adding anoutcome (taking out the garbage) to the Addressee\u2019s To-Do-List (TDL) (Portner,2004).12This is not the only way to \u201ccut the pie\u201d: De\u00b4chaine et al. (2017) argue for a stronger defini-tion of assertion, and the addition of another illocutionary force, presentation, that would be theillocutionary force of perspective-dependent predicates that do not intend to expand the CG. Thiswill be pertinent to our discussion of a possible paradigm of root complementizers in Chapter 6.18(13) Take out the garbage.Each interlocutor has a To-Do-List that contains the outcomes to which eachSpeaker commits. Commands such as the one in (13) propose to update the To-Do-List of the Addressee, but promises like (14) update the To-Do-List of theSpeaker: they say that the Speaker commits to complying with the action of takingout the garbage (Portner, 2004).(14) I promise that I\u2019ll take out the garbage.We therefore need to distinguish the TDL of the Speaker and another TDL forthe Addressee, just as we needed two different sets of discourse commitments.1.3.2 Canonical conversation updatesThe general consensus in recent literature is that speakers propose context updatesand the Addressee decides whether the proposed update is carried out or not (Clarkand Schaefer, 1989; Clark and Brennan, 1991; Ginzburg, 1996, 2012; Farkas andBruce, 2010; Thoma, 2016; Wiltschko, 2016). Figure 1.3 shows a model in whichan assertion consists of A uttering the proposition p, and p being automaticallyaccepted by B.A p BFigure 1.3: The simplest possible model of interaction: A utters p, and p isautomatically accepted by B into the CG.Clark and Schaefer (1989) and Clark and Brennan (1991) criticized a modelsuch as the one illustrated in Figure 1.3 since it does not reflect the collaborativenature of conversations. They highlight the fact that a speaker\u2019s contribution todiscourse is contingent on the actions of the addressee: the main purpose of a con-versation is establishing mutual knowledge, and it therefore needs collective acts19performed by all interlocutors (Clark and Schaefer, 1989, p.259). They thereforepropose that each conversational turn is divided into two phases: the Presenta-tion phase and the Acceptance phase, illustrated in figure 1.4. In the Presentationphase, A presents an utterance u for B to consider. In the Acceptance phase, Baccepts u by showing that they have understood what A means by u. If B acceptsu, both speakers arrive at a mutual belief that A\u2019s contribution is understood (andaccepted).Presentation A u BAcceptance A u BFigure 1.4: Two-step model of a dialogue contribution, based on Clark andSchaefer (1989).Recent formal conversation models have included the idea of presenting anutterance (specifically an assertion) before it is accepted into the CG. This dis-sertation will continue this trend and offer further evidence from Spanish DMs infavour of this idea, expanding it to include speech acts others than assertions andpolar questions (which Clark and Schaefer (1989) mention, but don\u2019t formalizespecifically). This section presents one such formal model, namely the one pro-posed by Farkas and Bruce (2010). This model formalizes the proposal nature ofassertions.The model presented in Farkas and Bruce (2010) (F&B henceforth) is meantto formalize the effects that assertions and polar questions have on the conversa-tion, and how the similarities (and differences) between these effects can explainthe possible reaction to them. F&B formalize context update proposals by addinga Table component to their model\u2013I will refer to their model as the Table Model20henceforth (Rudin, 2018). The Table is specifically conceived to represent the factthat context updates need to be proposed (tabled) before they are accepted into theshared knowledge component\u2013in this case, the CG. Another element F&B incor-porate is the projected set (PS): it represents the possible \u2018future\u2019 CG, that is, whatthe CG would look like if the proposed update were to be taken. The completemodel is represented in Figure 1.5, which includes individualized discourse setsfor each discourse participant (here assuming two discourse participants).13A Table BDCA S DCBCommon Ground cg Projected set psFigure 1.5: The Table Model by Farkas and Bruce (2010).Different utterances propose different updates. F&B introduce first a simpli-fied initial state K1 that holds prior to an utterance, as in Figure 1.6: there isnothing on the Table nor in the discourse commitment sets, and the CG and the psare equal. As we will see in 2.3.3.1, when analyzing the licensing (and effect) ofDM no and DM si we will have to exploit this initial state.A Table BCommon Ground s1 Projected Set ps1={s1}Figure 1.6: Initial state K1.An assertion such \u2018Sam is home.\u2019 proposes to update the CG with the propo-sition p = \u2018Sam is home\u2019. When A utters p, however, the CG is not immediatelyupdated: the utterance is put on the Table (both the proposition it denotes, and13This model assumes a separation between the CG and the individualized commitments foreach discourse participant; in Figure 1.2, this would be the leftmost representation of the nature ofCG.21the sentence itself as a syntactic object with the declarative feature [D]) and theprojected set ps is updated.A Table Bp \u3008 \u2018Sam is home\u2019 [D];{p} \u3009Common Ground s2 =s1 Projected Set ps2 = {s1 \u222a {p}}Figure 1.7: State K2: A utters an assertion.If B accepts this proposal, the next state would be the one in Figure 1.8: theTable is empty, as well as the PS, and the proposition is added to the CG.14A Table BCommon Ground s3 =s2 \u222a {p} Projected Set ps3 = \/0Figure 1.8: State K3: B accepts the move.What we have just seen is one possible effect of assertions on the discourse.F&B also discuss the effect that polar questions have: instead of proposing theaddition of a proposition to the CG, when A utters \u201cIs Sam home?\u201d she proposesto update the CG with either p or \u00acp. This is reflected in the context set byupdating the projected set ps, which now contains both possibilities (p or\u00acp addedto the previous CG s1).Here, [I] represents the interrogative form of the sentence.A Table B\u3008 \u2018Sam is home\u2019 [I]; {p, \u00ac p}\u3009Common Ground s1 Projected Set ps1 = {s1 \u222a {p}, s1 \u222a {\u00acp}}Figure 1.9: State K4: A asks a polar question.14Previous to this, p is added to the DC of both interlocutors, but as it enters the CG it is removedfrom these individualized commitment lists (Farkas and Bruce, 2010, p.99).22The idea that bare assertions and polar questions are update proposals insteadof direct updates is key in the development of analyses of elements that modifyassertions and polar questions, such as intonation and, crucially, DMs (Heim et al.,2014; Malamud and Stephenson, 2015; Thoma, 2016; Heim, 2019a; Jamieson,2019).1.3.3 Complex conversation updates: two casesThe exchanges we saw modelled in section 1.3.2 are a simplification of what goeson in a conversation. Simplified versions of what happens during a conversationare a useful basis for developing a theory of conversation; however, it is also im-portant to look at more complex interactions. This dissertation focuses on twotypes of context that call for a more complex type of update of the conversation:(i) situations where not all discourse participants share the same background in-formation; and (ii) situations where the Speaker requests the contribution of theAddressee to confirm the validity of the contextual update. Behind the use of DMsin these situations there seems to be a drive by language users to achieve maximalalignment, a concept defined in (15):(15) Maximal alignment: Interlocutors strive to eliminate any belief or out-come misalignment.This principle seems to drive language users to either repair a perceived mis-alignment (the situation with DM si), or to confirm that there is, in fact, epistemicalignment between the interlocutors (the situation with DM no).1.3.3.1 Mending epistemic misalignment: Peninsular Spanish DM siMost of the foundational work that has been done on the flow of conversations hasfocused on its additive nature: how interlocutors build up a conversation together,exchanging information and adding to the shared background information. Farkasand Bruce (2010, p.83) assume that \u201cconversational changes that result in addi-tion to commitments are less marked than those that result in the retraction of such23commitments\u201d, and that \u201cdenials are conversationally marked moves because theylead to conversational crisis\u201d (idem, p. 110). However, some circumstances re-quire the use of more marked, less additive conversational changes, as for example(16):(16) Context: A and B have been friends for a while, and had long conversationsabout their families. A never mentioned siblings. One day, B tells A thatthey can\u2019t wait to meet A\u2019s siblings. A says :Sisisoyamhijachildu\u00b4nica.unique\u2018[Si] I\u2019m an only child.\u2019In (16) there is an epistemic conflict: A thinks that it is in the backgroundknowledge that she does not have siblings (since she never mentioned them); how-ever, B has presupposed that A does have siblings, which means that the CG theyare working with contains the proposition p=\u2018A has siblings.\u2019Something is wrong. The proposition that B tables presupposes a propositionthat is in conflict with a proposition that is in the CG (according to A). The par-ticipants\u2019 information states are misaligned. Stalnaker (2002) mentions that whena speaker recognizes that not all discourse participants share the same CG, \u201csomekind of corrective action is called for\u201d (Stalnaker, 2002, p.717). This is exactlywhat DM si in (16) does: it is the corrective step after a speaker has recognized adefective context. But how does it do it? And what does it mean for our model ofconversation?In Chapter 3 I claim that si-clauses, like the one we just saw in example (16),do not represent assertions but superassertions. Unlike assertions, superasser-tions do not propose to update the CG with a new proposition, but they mark thata proposition is already part of the CG. This proposal also reflects previous anal-yses of DM si. Schwenter (2016a) argues that si marks that the proposition it in-troduces is obviously true to the Speaker: this analysis steers away from previous24ones that relied on the notion of contrast or contradiction (such as Rodr\u0131\u00b4guez Ra-malle (2011)), since si-clauses can appear in contexts where there isn\u2019t any con-tradiction.What the example in (16) highlights, though, is that it seems that there is notjust one version of the CG that is accessible to the participants: two interlocutorscan have a different idea of what is contained in the shared CG. What is more: theSpeaker seems to be able to identify this conflict, which means that she must betracking what the Addressee thinks that the CG looks like. In Figure 1.10, this isrepresented by the image on the right, whereas the image on the left represents amodel where there is only one CG:Figure 1.10: One or two CGs?Gunlogson (2008, p.108) points out that:Strictly speaking, each agent should have their own version of theoverall discourse structure, i.e., there should be one structure [...] peragent. I follow common practice in idealizing away from that level ofrepresentation for present purposes, assuming that the agents\u2019 individ-ual representations of the context do not differ substantially enoughto impede the progress of the discourse.The contexts of use of si-clauses do not allow us to make that assumption. Ipropose a model of conversation that includes two versions of the CG. However,both versions of the CG are mediated through the Speaker: no interlocutor hasaccess to the mind of the Addressee and knows exactly what the other personbelieves to be publicly shared. DM si therefore forces us to reflect the fact that25when we represent in a model what the dialogue looks like at a time t during theconversation, we are seeing it though the eyes of the Speaker, and this includesthe mutually shared knowledge. I will propose to formalize this mediation in themodel itself. This highlights the Speaker-centred analysis of DM si in Schwenter(2016a).The felicitous use of DM si forces us to consider much of what came beforethe utterance marked with the particle: we need to highlight the different statesthat precede the use of the si-marked clause if we want to understand the effect ithas on subsequent discourse. In order to fully account for the contexts of use andeffect of these constructions, we have to keep track of various phases before theSpeaker utters the si-marked clause.1.3.3.2 Confirming epistemic alignment: the Peninsular spanish DM noAnother example of a complex update can be seen in (17), where the utterancemade by the Speaker seems to have more than one effect on the conversation:(17) This is red, isn\u2019t it?The example in (17) illustrates an English tag question, consisting of an an-chor and the tag, which here has the form of a negated auxiliary and a subjectpronoun.15 On the one hand, the utterance asserts to some extent the proposition(p = \u2018This is red\u2019), but it also asks the Addressee to confirm whether this proposi-tion is true. It opens up a negotiation in the sense that it requires the involvementof the Addressee in establishing the shared knowledge (the CG): the proposition\u2018This is red\u2019 will not enter the CG until the Addressee accepts it. The differencebetween this and a canonical update proposal is that in (17), the Speaker onlytentatively commits to the proposition, as Malamud and Stephenson (2015) note.They model this distinction by claiming that there is not only one projected set forthe CG, as F&B had included in their Table Model, but that also individual DCs15More precisely, it is a reverse polarity tag.26have their own projected set in this account. Tentatively committing to a proposi-tion translates in this account to adding it to the projected DC of the Speaker (inthe case of reverse polarity (RP) tags) as Figure 1.11 illustrates.DCA DC*A DCB DC*BpTABLEpCG CG*Figure 1.11: How Malamud and Stephenson (2015) analyze RP-tags.Components with an asterisk (*) represent projected sets.A request for confirmation can take many forms\u2013and importantly, it can re-quest the confirmation of things beyond propositions. Wiltschko and Heim (2016)show in (18) and (19) how the DM eh in Canadian English can be used both toinquire about the truth of a proposition p (in (18)) and to request confirmation ofthe Speaker\u2019s assumption of the Addressee\u2019s knowledge of p (in (19)). In the lattercontext, not all confirmationals are felicitous, as the non-acceptability of huh andright shows.(18) Context: John knows that Mary would like to have a new dog. He hasn\u2019tseen her in a long time. And he keeps wondering whether she got a newdog. One day he runs into her while she\u2019s walking a new puppy. John utters:You have a new dog, {eh\/huh\/right}?= Confirm that p is true (Wiltschko and Heim, 2016, 3:5)(19) Context: Mary is walking her new dog when she runs into John. She isexpecting that he would congratulate her on the new dog, but he\u2019s not men-tioning it. She isn\u2019t sure anymore whether he actually realizes that she has27a new dog. So she utters:I have a new dog, {eh\/*huh\/*right}?= Confirm that you know that p is true (Wiltschko and Heim, 2016, 3:6)Wiltschko and Heim (2016) propose that speakers can table not just proposi-tions, but also attitudes towards propositions. In addition, they also propose tomodel the biased nature of tag questions as in Figure 1.12: the use of the confir-mational DM eh tabels A\u2019s bias towards B believing p (marked by boldfaced font)over its polar opposite. This bias is also encoded in the Speaker\u2019s commitmentset. In this framework, the CG is the intersection between the two individualizedgrounds of the interlocutors.Speaker (S) Table Addressee (A)Bel(A, p) \u2228Bel (A, \u00acp)Bel(A, p) \u2228 Bel(A, \u00acp)Figure 1.12: Analysis of Canadian eh in Wiltschko and Heim (2016).Two things are especially interesting for analyzing DM no: on one hand, thefact that we need to table not only propositions, and on the other hand, the biasthat comes with the use of a request for confirmation.The analyses just presented use F&B\u2019s Table Model as a way to representthese complex contexts where interlocutors not only exchange propositional infor-mation, but also convey more meta-conversational information. I also use F&B\u2019sTable Model as a basis for my own model, and include some modifications to it(section 2.3.3). However, there is another advantage to F&B\u2019s Table Model thathas not been exploited to its full potential (neither in the more general discussionabout dialogue nor in the contribution of DMs), and that is the way it representsthe link between speech act types and different types of conversational updates.28This is especially important for the discussion about DM no.DM no can accompany a wide variety of anchors, as we will see in both Chap-ters 2 and 4. This DM does not discriminate in terms of the sentence type it canmodify\u2013but it does discriminate in the type of speech act it anchors to. A sneakpeek of the types of anchors that DM no can attach to is given in Table 1.2:Table 1.2: Felicity of confirmational no with different anchors.Type of speech act ConfirmationalAssertive (affirmative) 4Assertive (negative) 4Assertive (taste) 4Assertive (internal state) 7Polar question 7Wh-question 7Rhetorical Wh-question 4Command 4Expressive 7Commissive 7Declaration 7Malamud and Stephenson (2015) make a point of saying that they focus ondeclaratives and tags, and Farkas and Bruce (2010) only discuss assertions andpolar questions. Wiltschko and Heim (2016) do discuss the possibility of confir-mationals requesting confirmation of something other than the truth of a proposi-tion, and their framework in itself makes a point in stating that some confirmation-als might be sensitive to something beyond sentence type (Yang and Wiltschko,2016; Wiltschko et al., 2018). However, the discussion on declarative sentenceshas typically centred on assertions, and not on other speech act types that are also(typically) expressed with the declarative sentence type, such as commissives. Theinfelicity of using DM no with this specific speech act type will be a key piece ofevidence in the analysis.291.3.4 The link between speech acts and conversational updatesAs I just mentioned, one of the main points in F&B is that different speech acttypes have an effect on the dialogue that is particular to that speech act. We haveseen what happens with assertions and questions, but what about other types ofspeech acts? And is there any other way in which speech act type impacts the useof DM si and DM no?The first important point is to distinguish sentence type and speech act type.I will distinguish four basic sentence types (declaratives, imperatives, interroga-tives, and exclamatives), and six basic speech act types, summarized in Table 1.3.Distinguishing between sentence type (the actual syntactic form of the sentence)and speech act type (the performative use of the sentence) is crucial for the anal-ysis of DMs. On the one hand, DM no is sensitive to speech act types but notsentence type (a cross-linguistics parameter of variation (Wiltschko et al., 2015)),and the commonalities between the speech act types that can occur with it tellus something about the contribution of no. On the other hand, the fact that eachspeech act type has a number of felicity conditions associated with it explains thelicensing of DM si, as we will see in Chapter 3.I will slightly modify the taxonomy in Searle (1979) by making a distinctionbetween commands and questions\u2013for him, these two speech act types form onesuperset called directives. The speech act types that will be discussed in thisdissertation are assertions, commands, questions, expressives, commissives, anddeclarations, and are summarized and described in Table 1.3.30Table 1.3: Speech act types discussed in this dissertation, based on Searle(1979, 1976)Type Characteristics ExampleAssertions Commit the speaker to thetruth of the expressed propo-sition.There\u2019s a pet store on Main.Commands Attempts by the speaker to getthe hearer to do something.Give me the keys.Questions Attempts by the speaker to getthe hearer to provide informa-tion.What\u2019s your name?Expressives Express the speaker\u2019s attitudeabout objects and facts of theworld.I\u2019m sorry.Commissives Commit the speaker to somefuture course of action.I promise I\u2019ll do it.Declarations Change a non-linguistic stateof affairs in the world.You\u2019re fired.There is a \u2018canonical\u2019 link between sentence types and speech act types: as-sertions are typically conveyed with declarative sentences, questions are typicallyconveyed with interrogative sentences, etc. This is not a one-to-one mapping: forinstance, commissives are typically conveyed by declarative sentences, and someinterrogative sentences are said to have the force of an assertion (Han, 2002; Corr,2016)\u2013this is the case of rhetorical questions. This fact will play an important rolein the analysis of DM no in Chapter 4.Different speech act types are also associated with different types of commit-ments (Portner, 2004; Beyssade and Marandin, 2006): an assertion commits theSpeaker to the truth of a proposition, whereas a commissive commits the Speakerto an outcome or action. Another important difference between speech act types is31what can be called the orientation of the commitment: a commissive only requiresthe commitment of the Speaker, whereas a command requests an action from theAddressee. A summary of these characteristics is shown in Table 1.4.Table 1.4: Speech acts, commitment types, and the dialogue elements.Type Commitment to Dialogue componenttargetedAssertion (non-evaluative) proposition CGAssertion (evaluative) proposition DCSpeakerCommand outcome TDLAddresseeQuestion set of propositions CGExpressive N\/A DCSpeakerCommissive outcome TDLSpeakerDeclaration proposition CGThere is another way in which speech act types affect the use of DMs, inparticular DM si: the felicity conditions associated with them. Speech act typeshave a particular set of rules, or felicity conditions, that ensure their success inperforming the act they are supposed to perform. These felicity conditions aredivided into four types: propositional content conditions, preparatory conditions,sincerity conditions, and essential conditions (Searle, 1969; Vanderveken, 1990).Propositional content conditions make sure that the meaning of the sentence isappropriate: for instance, when making a promise, the act that the Speaker ispromising to fulfill has to be in the future. Preparatory conditions describe the\u2018right\u2019 conditions for the speech act to be felicitous: when making a promise,the Addressee should welcome the action that the Speaker is promising to fulfill(otherwise it seems more like a threat than a promise). Sincerity conditions guar-antee that the Speaker has the appropriate thoughts or feelings; for instance, if aSpeaker is commanding an Addressee to perform an action, the Speaker shouldwant the Addressee to perform that action. Essential conditions make sure that32the speech act performed is in fact considered by the interlocutors as performingthe appropriate speech act. These conditions are summarized in Table 1.5.Type Definition Example (request)Propositionalcontentthe meaning of the sentence isappropriateFuture actPreparatory contextual conditions that needto holdAddressee is able todo the actSincerity Speaker\u2019s attitude Speaker wants Ad-dressee to do theactEssential Speaker\u2019s intention to performthe speech actPerformed act countsas a requestTable 1.5: Searle\u2019s felicity condition types (Searle, 1969).Language users have this type of knowledge\u2013the knowledge of felicity condi-tions and of the rules of a conversation in general. Failing to satisfy these con-ditions will also result in epistemic misalignment, and this can trigger the use ofDM si, as we will see in Chapter 3.1.4 The road aheadThe protagonists have been presented, and I have just laid out a very abstractsetting, introducing a general, dynamic idea of how conversations work and thelink between this idea of conversations and speech act types. In the next sectionsI describe briefly how the DMs fit into this setting, and the structure of the presentdissertation.1.4.1 Where we find si and no and what they doChapter 2 focuses on the two DMs themselves: where we find them and whatthey do. Both DM no and si have a wide functional range (Thoma, 2016), which33means that they are used in a wide variety of different contexts. I will describedifferent uses of si and no other than their use as DMs, although the main focuswill be on their use as DMs. I refer to previous analyses of these DMs in theliterature, and present the next steps in the analysis: the tools for modelling thecontribution of DM si and no, as well as a first approximation to their analyses.1.4.2 Analysis of DM si: monitoring the CGChapter 3 proposes an analysis for the semantic and the pragmatic effects of theDM si. One of the main points is that we need to distinguish the contribution ofthe si-clause to the discourse from the contribution of the DM si to the clause itintroduces. Regarding the contribution of the si-clause to the discourse I claim thatit is used to reject the previous discourse contribution based on it not observing afelicity condition. This infelicitous discourse contribution makes apparent thatthere is a mismatch in the CG. In (20), A is rejecting the previous commandbecause it does not meet one of the felicity conditions of commands: namely,that the Addressee should be able to perform the action in the future.(20) CONTEXT: B orders: Bring me a cookie! A replies:Sisiteyoulasthemacabastefinish.PST.2SGayer.yesterday\u2018[Si] you finished them yesterday.\u2019In (20), A\u2019s reply rejects the previous command: A cannot possibly complywith that command (bringing a cookie), since B ate all the cookies the day before.In other words: it is impossible for A to accept the command because she cannotperform the action required by the order. That is, A rejects an order because oneof the felicity conditions (the Addressee needs to be able to perform the actionrequired) is not met.Regarding the contribution of the DM si, I claim that it is used to remind the34Addressee that the proposition denoted by the sentence it introduces is already inthe CG. In other words, it doesn\u2019t propose to update the CG, but reminds the Ad-dressee of the content of the CG. I claim that this is a special type of illocutionaryforce: instead of asserting a proposition p, a si-clause superasserts p.The difference between an assertion and a superassertion is that whereas whensomeone asserts p they propose to update the CG with p, when someone su-perasserts p they mark that the proposition is already in the CG, and hence, noupdate of the CG is needed.1.4.3 Analysis of DM no: monitoring the Speaker-AddresseerelationChapter 4 focuses on DM no. Its core function is to monitor the relationshipbetween the Speaker and the Addressee by requesting the involvement of the Ad-dressee in adding a proposition or an outcome from the individual DCs or TDLsto the shared CG or shared TDL.In the following examples, we see how the Speaker is trying to create align-ment with the Addressee: both in terms of future outcomes (21) and in beliefs (22).In (21), the Speaker wants to know whether the outcome described by the impera-tive (\u2018Let\u2019s go somewhere else\u2019) is something that the interlocutors are committedto; in (22), the use of DM no checks whether the Speaker\u2019s opinion (about themovie being boring) is shared with the Addressee:(21) Venga,come.onvamosgo.PRS.1PLatootroanothersitio,placeno?no\u2018Come on, let\u2019s go somewhere else, [no?]\u2019 (Rodr\u0131\u00b4guez Mun\u02dcoz, 2009)35(22) Context: A and B just started watching a popular show. After 5 minutes Ais bored, and B looks at A and rolls their eyes. A says:Estathisserieshowesisunarollo,dragno?no\u2018This show is boring, no?\u2019Whereas DM si takes a proposition as its argument, DM no takes a wholespeech act, as will be shown in Chapter 4. These different levels at which DM siand DM no operate can also be seen in the different positions these DMs occupyin the syntactic structure.1.4.4 What\u2019s syntax got to do with it?Chapter 5 proposes a syntactic analysis of the DMs si and no, adopting proposalsthat enrich and expand the left periphery of the clausal architecture. Rizzi (1997)proposes that the CP consists of several projections, including Topic, Focus, andother information structure related phenomena. Using data from Romance lan-guages, especially Italian, Rizzi (1997) proposes that we need to distinguish atleast two projections: ForceP, which hosts the complementizer, and FinitenessP,which is a direct link to the finite\/non-finite feature of the verb in IP. Betweenthose two projections there are other projections which are landing sites of topicsand foci. There have been later additions to this \u2018exploded CP\u2019, including an Inter-rogative Phrase (which would host interrogative complementizers) and ModifierPhrase (which hosts discourse-oriented adverbials), represented in Figure 1.13(abstracting away from certain projections that are not relevant for current pur-poses).The CP is the syntactic layer where a sentence radical (IP) gets linked to an-other clause or to the rest of the discourse, which is which is how I have charac-terized the role of DMs in this chapter: it seems intuitive to locate our DMs inthis part of the syntactic structure. However, things are a bit more complicated;in Chapter 5 I will propose that only one of the two DMs (DM si) is located in36ForcePForce\u2019Forceo TopicPhola Topic\u2019Topo IntPInt\u2019Into FocusPFocus\u2019Foco FinPhola Fin\u2019Fino IPholaholaholaFigure 1.13: Exploded CP.the CP layer, and that given its function, its relative order with respect to otherelements of the left periphery and its scope it lexicalizes the highest projection ofa split phrase encoding Force (Coniglio and Zegrean, 2012).Previous research has stated that DMs are outside the scope of traditional syn-tax (Pons Border\u0131\u00b4a, 2006; Travis, 2006; Roggia, 2012; Montan\u02dcez Mesas, 2015).However, recent work within generative syntax has proposed ways in which utterance-level phenomena (such as DMs) can be modelled. I refer to these models asneoperformative analyses following Wiltschko and Heim (2016), referring backto Ross\u2019 Performative Hypothesis (Ross, 1970). He proposed to analyze sentenceslike (23a) as having a covert performative verb (23b): that is, the idea that everyutterance is performative is encoded in the syntax:(23) a. I will give you your money back.37b. I promise you that I will give you your money back.He extends this hypothesis to include declarative clauses. The proposal istherefore that all matrix clauses are in essence bi-clausal in their deep structure,with an element representing the Speaker (the subject in this upper clause), anelement representing the addressee (the object), and a verb of saying linking bothelements (the predicate). In other words: the utterance \u201cAdriana is writing\u201d is infact bi-clausal, with the higher clause \u201cI tell you (that)\u201d being deleted. This isexemplified in the following tree:(24) SNPIVPVtellNPyouSthat Adriana is writingThe Speaker has a central role in assigning meaning to certain linguistic items:indexicals such as first and second person pronouns, demonstratives, and temporallocutions (such as tomorrow or today) need to have a referent in order for theAddressee to compute the meaning. Ross (1970) proposes that this referent is theSpeaker in declarative sentences, and he represents the Speaker in the syntax inorder to account for phenomena like antecedent reflexive pronouns in (25), wherethe reflexive pronoun \u201cmyself\u201d would be bound by the covert Speaker element inthe higher clause.(25) (Ii tell you that) This book was written by Fred and myselfi.Recently, some researchers have re-taken the idea of having a syntactic layerencoding the Speaker-Addressee relationship. These neoperformative analyses,38however, do not use a bi-clausal structure. Instead, they propose a projectionof functional categories that encode pragmatic information. This projection hasbeen given different names: DiscourseP (Beninca`, 2001), Speech ActP (Speasand Tenny, 2003; Haegeman and Hill, 2013; Haegeman, 2014; Corr, 2016, 2018),AttitudeP (Paul, 2014), or GroundingP (Lam, 2014; Heim et al., 2016; Thoma,2016; Wiltschko and Heim, 2016; Wiltschko, 2017; Jamieson, 2019), but they allrevolve around the claim that this layer encodes pragmatic information about thespeech act participants. For now, I will call this layer UtteranceP; its position isillustrated in Figure 1.14.Figure 1.14: Proto-neoperformative analysisRecent analyses of DMs such as Canadian English eh (Wiltschko and Heim,2016), Schwabian German gell (Heim, 2019b), and tag questions in Scots varieties(Jamieson, 2019) have proposed to analyze these items as lexicalizations of theutterance-level syntactic layer, based both on functional and syntactic criteria.I propose a syntactic analysis of the Peninsular Spanish DMs si and no inChapter 5. I will discuss how the data from Spanish DMs contributes to the on-going discussion of utterance-level syntax. Although DM no is relatively inert interms of its syntactic activity, it displays certain properties regarding scope andinteraction with pragmatic roles that have been proposed to be part of the syntaxof the utterance (Speas and Tenny, 2003; Giorgi, 2010; Haegeman and Hill, 2013;39Corr, 2016; Thoma, 2016; Wiltschko and Heim, 2016; Wiltschko, 2017). On theother hand, DM si can be described as an illocutionary complementizer (Corr,2018), and I will use diagnostics used in the literature for similar complementiz-ers to propose a syntactic position (Corr, 2016) still within CP, but in a much morearticulated left periphery.1.5 On dataBefore jumping into the main body of the dissertation, I would like to discusssome points regarding the data used in this dissertation. The data that appearsin this dissertation was collected in four different ways: (i) from previous litera-ture; (ii) from elicitations with other native speakers; (iii) based on my own nativespeaker judgement; and (iv) from open access corpora. Sources have been prop-erly cited when citing data coming from other sources (literature or corpora), andmost of my own native speaker judgements have been confirmed by another nativespeaker.Many of the contexts used in elicitation, especially regarding DM no, comefrom the Syntax of Speech Act Project at UBC, led by Martina Wiltschko.16. Thisproject aims to describe, analyze, and model the way language users interact,mainly by focusing on phenomena like DMs and intonation. In addition, thisproject has developed a repository of contexts which target different parametersof discourse particles. These contexts have been marked with a H.It is important to note that although both DM no and DM si are used (as faras I am aware) in all varieties of Spanish, the felicity of these DMs in particularcontexts might differ between varieties, as it does for other languages (see, for ex-ample, the differences between Canadian English eh (Gold, 2005; Gold and Trem-blay, 2006; Denis, 2013; Heim et al., 2014; Wiltschko and Heim, 2016) and NewZealand English eh (Meyerhoff, 1994; Starks et al., 2008; Schweinberger, 2018)).16The elicitation storyboards are available in the following link:https:\/\/syntaxofspeechacts.linguistics.ubc.ca\/40The felicity judgements reported in this dissertation are constrained therefore tothe Central Peninsular variety of Spanish, spoken in the centre of the IberianPeninsula.As a final note on how the data is represented and talked about, I have decidedto use fixed pronouns for Speaker and Addressee: when referring to the Speaker, Iwill use the pronoun \u2018she\u2019, and when referring to the Addressee, I will use \u2018they\u2019,unless the example states otherwise.41Chapter 2The Peninsular Spanish DMs si andno: where we find them and whatthey doDM si and DM no seem to be polar opposites in terms of function: whereas DM sirejects the previous contribution of the Addressee, DM no actively seeks the con-tribution of the Addressee. But something that both DMs share is the functionalrange of their lexical items: besides the use as a DM that we have seen so far, theparticle si can be used as the head of complement and adverbial clauses, as wellas introducing exclamative sentences. The particle no can be used as a (negative)response marker, a marker of sentential negation, and as a phatic particle, in ad-dition to its use a confirmational DM which is the focus of this dissertation. Inthis chapter I present these \u201cother jobs\u201d of the DMs, as well as previous observa-tions made in the literature about them. Based on these observations I propose tomodify the F&B Table Model discussed in Chapter 1 to account for the distribu-tion and licensing contexts of the DMs, including highlighting speaker mediationin the Dialogue Board and ways to model the effect of felicity conditions on theDialogue Board.422.1 Peninsular Spanish si: where we find it andprevious accountsThe particle si in Peninsular Spanish can be found in dependent and independentclauses, always in the left periphery. Figure 2.1 shows that the first big division isbetween whether si appears in a dependent or independent clause. In independentclauses its role is as a complementizer, which is usually considered to be thecounterpart of English \u2018if\u2019. When it appears in independent clauses, it has twomain functions: as an exclamative marker, and as a DM.SIDependentInterrogative ConditionalIndependentExclamative DiscourseFigure 2.1: Schema of the different functions of si.(26) a. DEPENDENT SI: INTERROGATIVEMemyselfpreguntoask.PRS.1SGsiwhetheresta\u00b4islloviendo.raining\u2018I wonder whether it\u2019s raining.\u2019b. DEPENDENT SI: CONDITIONALNoNEGvoygo.PRS.1SGsiifesta\u00b4islloviendo.raining\u2018I won\u2019t go if it\u2019s raining.\u2019c. INDEPENDENT SI: EXCLAMATIVE43Sisisera\u00b4be.FUTtonto!dumb\u2018Isn\u2019t he dumb!\u2019d. INDEPENDENT SI: DISCOURSEContext: B tells A that they are going out to catch some sun. A says:Sisiesta\u00b4islloviendo!raining\u2018[Si] it is raining!\u2019The question of the relationship between dependent and independent uses ofsi has been framed in the context of insubordination: the use of non-root clausesin root environments (Evans and Watanabe, 2016). The evidence is tricky: on theone hand, there is evidence to suggest that independent si-clauses are indeed rootclauses: in section 2.1.1.3 we will see that there are both prosodic and syntacticreasons to consider them independent clauses (Elvira-Garc\u0131\u00b4a et al., 2017; Schwen-ter, 2016a); however, they seem to require a previous discourse contribution to befelicitous.In this dissertation, I focus on the use of DM si in contexts such as (27), whereit appears in the left periphery of root clauses. I refer to this use as DM si. Inprevious literature it has been described as having a refutational or adversativemeaning. This is illustrated in (27): by using a si-marked clause, Speaker Aconveys why she is against B\u2019s exhortation to bring a raincoat with her (namely,that it is not raining, so it is not helpful):(27) Context: A is about to leave the house, but B tells her to bring a raincoatwith her. A says:44SisinoNEGesta\u00b4islloviendo.raining\u2018[Si] it isn\u2019t raining.\u2019Before zooming in on the contexts of use of DM si, let us describe how wecan distinguish it from the other uses in dependent and root clauses.2.1.1 Where we find siIn what follows, I illustrate the syntactic contexts in which DM si is found, startingwith the clause-typing si that occurs in dependent CPs (section 2.1.1.1), then look-ing at the exclamative si that occurs with root CPs (section 2.1.1.2), and finallyturning my attention to the DM si that occurs with root CPs (section 2.1.1.3).2.1.1.1 Si and dependent CPs: interrogative and conditional siThe schema in Figure 2.1 shows that the first division of the distribution of siis whether it introduces an embedded or a root clause. In the former case, si isanalyzed as a complementizer heading interrogative clauses (as in (28)) or con-ditional clauses, as in (29) and (30). Interrogative si introduces indirect polarinterrogatives, as in (28) and selects a verb in the indicative mood.(28) a. MeREFLpreguntoask.1SGsiifera\/es\/sera\u00b4be.IND.PST\/PRS\/FUTverdad.truth\u2018I wonder whether it was\/is\/will be true.\u2019b. *MeREFLpreguntoask.1SGsiiffuera\/sea\/fuerebe.SBJV.PST\/PRS\/FUTverdad.truthConditional si introduces the protasis or antecedent and can appear with theverb in indicative (29) or subjunctive (30) mood .45(29) Siifhacedo.PRS.INDfr\u0131\u00b4ocoldnoNEGvamosgo.PRS.INDatolathepiscina.swimming-pool\u2018If it is cold we won\u2019t go to the swimming pool.\u2019(30) Siiftuvierahave.PST.SBJVdineromoneymeREFLcomprar\u0131\u00b4abuy.CONDunacoche.car\u2018If I had money, I would buy a car.\u2019There is the possibility of eliding the consequent, yielding what has beencalled oraciones suspendidas (\u2018suspended clauses\u2019), as in (31):(31) Context: B asks whether A is joining them in their roadtrip. A replies:Siiftuvierahave.PST.SBJVdinero. . .money\u2018If I had money. . . \u2019The omission of the consequent is due to pragmatic reasons or the informa-tional status of the content (given information may be omitted): in the exampleabove, the consequent is easily recoverable from B\u2019s question. Crucially, this eli-sion has a prosodic consequence, marked here with \u201c. . . \u201d: a continuation contourat the end of the utterance (which in this case would only consist of the conditionalclause). This contrasts with the contour of independent si-clauses, which do notpossess this contour (Elvira-Garc\u0131\u00b4a et al., 2017).2.1.1.2 Root CPs: distinguishing exclamative si from DM siExclamative independent si-clauses, also called \u2018C root exclamatives\u2019 in Gutie\u00b4rrez-Rexach (1999), are exclamative root clauses with si as their initial element. Theyare exemplified in (32):46(32) CONTEXT: B has made the same mistake two times in a row. B makes thesame mistake yet again, and A exclaims:Sisisera\u00b4sbe.FUT.INDidiota!idiot\u2018Boy, you are such an idiot!\u2019 (Gutie\u00b4rrez-Rexach, 1999, p.169:2a; my con-text)Hernanz (2012, 2015) identifies certain characteristics that distinguish excla-mative independent si-clauses: these include tense and mood restrictions, the or-der of the subject and verb within the clause, and the type of adjective allowedin the predicate. These differences are illustrated below, and summarized in Ta-ble 2.1.Tense and mood are restricted in exclamative si clauses: the future indicative isallowed, as we saw in (32), but not the past or the present, as the ungrammaticalityof (33) shows.(33) *Sisifuiste\/eresbe.PST.SBJV\/PRS.SBJVidiota!idiotEXCLAMATIVE SI(Gutie\u00b4rrez-Rexach, 1999, p.182:47)Both exclamative and DM si need to appear in a clause that bears indicativemood. But whereas exclamative si can only co-occur with the future as shown inexamples (32) and (33), DM si can appear with any tense (past, present, future).(34) Sisiten\u0131\u00b4a\/tengo\/tendre\u00b4have.PST.IND\/PRS.IND\/FUT.INDdinero.moneyDM SI\u2018[Si] I had\/have\/will have money.\u2019Another important distinction is the position of the subject. In exclamative si-clauses, the verb phrase [tendra\u00b4 dinero] has to move to precede the subject ([estefarsante] in the examples below), as (35) shows:47(35) a. *Sisiestethisfarsantephonytendra\u00b4have.FUT.IND.dinero!moneyEXCLAMATIVE SI\u2018This phony has so much money!\u2019b. Si tendra\u00b4 dinero este farsante!(Hernanz, 2012, p.19-21; (60),(65))DM si-clauses do not have this restriction. Note that in (36) I use a differenttense to further mark the differences between the two uses of si:(36) a. Sisiestethisfarsantephonyten\u0131\u00b4ahave.PST.INDdinero!moneyDM SI\u2018[si] this phony had so much money!\u2019b. ?Si ten\u0131\u00b4a dinero este farsante!The final difference is that exclamative si clauses cannot include a non-gradablepredicate, such as being a vegetarian in (37), whereas DM si accepts such predi-cates (38):(37) #SiSIsera\u00b4be.FUT.INDvegetariana!dumbEXCLAMATIVE SI\u2018S\/He is so vegetarian!\u2019(38) SiSIesbe.PRS.INDvegetariana.vegetarianDM SI\u2018[Si] she is a vegetarian.\u201948Table 2.1: Summary of differences between exclamative and discourse siExclamative si-clauses Discourse si-clausesGradable predicate 7 3Verb before subject 3 7Tense restriction 3 7Mood restriction 3 3Although there are both exclamative and discourse independent si-clauses,from now on I will use the term \u201c si-clause\u201d or \u201csi-marked clause\u201d to refer tothe discourse type.12.1.1.3 Si and root CPs: DM siThe status of independent si-clauses as root clauses is motivated by prosodic andsyntactic differences between independent and dependent clauses.Prosody: Elliptical conditional clauses have a continuation rise: although thespecific contour varies, the boundary tone is always H% (Elvira-Garc\u0131\u00b4aet al., 2017). On the other hand, independent si-clauses have a L% boundarytone.Syntax: Independent si-clauses do not license the use of negative polarity items(NPIs), as opposed to elliptical conditional clauses (Schwenter, 2016a). In(39), the NPI cualquier (\u2018ANY\u2019) is only grammatical in the elliptical condi-tional clause (a), and not in the independent si-clause (b):(39) a. Si tienes cualquier duda . . . ELLIPTICAL CONDITIONAL\u2018If you have ANY doubt . . . \u2019b. *Si tienes cualquier duda! INDEPENDENT SI-CLAUSE\u2018[Si] you have any doubt!\u20191I make the distinction between DM si and si-clauses to distinguish the lexical item si from theentire clause it heads, respectively.49In addition, independent si-clauses cannot be embedded, nor conjoined, nornegated, as opposed to dependent si-clauses. These three characteristics will beillustrated and explored further in Chapter 5.2.1.2 What DM si does: previous accountsThe discursive function of si was observed in Spanish grammars as early as the19th century. Schwenter (2016a, p.4) mentions that Bello (1847) \u201cdescribes theconstruction in elliptical terms and notes how the context easily supplies the con-tent of the elided material\u201d, using as an example a dialogue from 1791. Its use asa marker of adversativity or contrast has been the main focus of previous literatureon this particle.2.1.2.1 Si-utterances are adversativeThe function of DM si has been defined as adversative, refutational, corrective,contrastive, and mirative just to name a few terms (Schwenter, 1998; Montol\u0131\u00b4o Dura\u00b4n,1999; Schwenter, 2000b; Rodr\u0131\u00b4guez Ramalle, 2011; Hernanz, 2015; Schwenter,2016b). What the literature seems to agree on is that DM si appears in reactiveutterances: it needs a previous trigger, to which it reacts, as in (40). In this ex-ample, the si-clause is reacting to the previous discourse contribution that statesthat Ottawa is in the United States, and corrects this information introducing theproposition \u2018It is in Canada\u2019:(40) B: Ottawa esta\u00b4 en Estados Unidos.\u2018Ottawa is in the US.\u2019A: Sisiesta\u00b4iseninCanada\u00b4!Canada\u2018[Si] it\u2019s in Canada!\u201950The role of contrast in the core meaning of DM si has been debated: DMsi seems to be licensed in contexts where there is an apparent contradiction (aswe have seen in most examples until now, but maybe most notably in (40)). ForRodr\u0131\u00b4guez Ramalle (2011), the notion of contrast is encoded in DM si, in a waythat links it to the complementizer si when it introduces embedded polar interrog-ative clauses: si (both in its role as complementizer and as a DM) contrasts twoopposing alternatives. In the case of interrogative si, it\u2019s two polar alternatives; inthe case of DM si, the two contrasting alternatives are at the level of discourse. In(41), discourse si contrasts the clause it introduces (Mar\u0131\u00b4a canta, \u2018Mar\u0131\u00b4a sings\u2019)with a previous discourse time where Speaker A did not know that Mar\u0131\u00b4a sang(Rodr\u0131\u00b4guez Ramalle, 2011, 212).(41) SisiMar\u0131\u00b4aMar\u0131\u00b4acanta.sings\u2018[Si] Mar\u0131\u00b4a sings\u2019 (Rodr\u0131\u00b4guez Ramalle, 2011, 212:27a)However, as Schwenter (2016a) pointed out, independent si-clauses can alsobe used as reactions to utterances where Speaker A is not refuting or contradictinganything Speaker B is saying. This is what happens in example (42)\u2013in fact,Speaker A is agreeing with the previous utterance (marked by the use of claro \u2018ofcourse\u2019).(42) B: A Juana la han aceptado en Stanford.\u2018Juana was accepted into Stanford.\u2019A: Claro,of.coursesisiesismuyveryinteligente.intelligent\u2018Of course, [si] she\u2019s very intelligent\u2019 (Schwenter, 2016a, 26: 10)The analysis presented in this dissertation will propose a way in which we canbring together these two insights (that there is some sort of correction but that it51is not necessary), and which will rely on the idea of felicity conditions. Beforepresenting this analysis, I illustrate the contexts of use of independent si-clausesthat have been identified in the literature.2.1.2.1.1 Independent si-clauses refute implicaturesExample (40a) showed that DM si can be used to react to the propositional con-tent of the previous discourse contribution; on the other hand, examples in (43)show that DM si can also be used to react to non-propositional content, in this caseimplicated content. In (43a), Speaker A is reacting to the implicated meaning thatC is a good person to ask advice\u2013this conversational implicature is the result ofthe Maxim of Relevance, since Speaker A assumes that Speaker B is suggestingC given the topic of the conversation. In (43b) we see that conversational impli-catures and scalar implicatures can trigger the use of DM si (Schwenter, 2000b,a,2014). Here, Speaker A understands that Speaker B thinks that C is just a \u2018bad\u2019candidate and nothing worse than that, but she \u201ccorrects\u201d this by saying that C isnot only bad, but the worst candidate:(43) a. TRIGGER: CONVERSATIONAL IMPLICATUREContext: While looking for some advice for their US roadtrip, B sug-gests to ask C. A replies:SisinoNEGhahassalidoget.outdeofEspan\u02dca!Spain\u2018[Si] he hasn\u2019t been outside of Spain!\u2019 (adapted from Schwenter (2000b))b. TRIGGER: SCALAR IMPLICATUREContext: B says C is a pretty bad candidate for the job. A replies:Sisiesiselthepeor!worst\u2018[Si] he is the worst!\u2019522.1.2.1.2 Independent si-clauses refute inferences from contextAnother important observation made in the literature is that independent si-clausescan react to non-verbal cues. In (44), A infers from the physical act of loweringthe bicycle seat to a certain point that the seat cannot be lowered any further. DMsi is used then followed by the word incluso \u2018even\u2019 which makes this inferenceexplicit:(44) CONTEXT: Looking at a bicycle for a 4-year-old son, B says: \u2018Look, youcan lower the seat also.\u2019 while lowering the seat. A replies:SisiinclusoevenseIMPRSpuedecanbajarlowerma\u00b4s.more\u2018[Si] you can even lower it more.\u2019 (Schwenter, 2000b)2.1.2.1.3 Independent si-clauses refute the form of the previous contributionAnother context in which we find DM si is when a Speaker wants to correct herinterlocutor\u2019s pronunciation of a word, as in (45):(45) Context: B says: Man\u02dcana tengo que dar clase, sobre GENEOLOGI\u00b4A \u2018To-morrow I have to teach a class about geneology.\u2019 A replies:SisiesisGENEALOGI\u00b4A.genealogy\u2018[Si] it\u2019s genealogy\u2019 (Schwenter, 2000a)In (45), the Addressee has incorrectly pronounced the word genealog\u0131\u00b4a \u2018ge-nealogy\u2019, and the Speaker uses a si-marked clause to introduce the correct formof the word. We therefore find DM si in contexts where the form of the previousmessage is not felicitous.532.1.2.1.4 Independent si-clauses refute previously held beliefsIn the above sections, independent si-clauses appeared after a previous movefrom the interlocutor, regardless of what specific thing in the move it actuallytargeted (the propositional content, an implicature, the form, etc). But DM sican also be used when there is no Addressee around. This use, exemplified in(46), is what authors have called the mirative use of si (Rodr\u0131\u00b4guez Ramalle, 2011;Torres Bustamante, 2013; Hernanz, 2015):(46) CONTEXT: B has always heard Juan complain about cigarette smoke. Oneday, she sees him smoking a cigarette, and exclaims:SiSIJuanJuanfuma!smokes\u2018[Si] Juan smokes!\u2019In (46), there is no interlocutor to correct, literally speaking: the Speaker issurprised that Juan smokes, given that she thought that he didn\u2019t. I will assumewith (Torres Bustamante, 2013) that she is in fact correcting herself\u2013her interlocu-tor, so to say, is herself in the past.2.1.2.2 DM si marks obvious truthSchwenter (2016b) remarks that the core meaning of DM si is to mark that apiece of information is obviously true to the Speaker\u2013in this analysis, the rejectingflavour of si-clauses is a result of the types of contexts in which one may want tomark this obviousness, but not necessarily a part of the core meaning of DM siitself. One of the pieces of data that he uses is examples such as (47), repeatedfrom (42), where there is no apparent conflict: Speaker A is agreeing with SpeakerB, and is, in fact, just giving more evidence or reasons to believe what A is saying:54(47) Context: B tells A that Juana was accepted into Stanford. A replies:Claro,of.coursesisiesismuyveryinteligente.intelligent\u2018Of course, [si] she\u2019s very intelligent\u2019 (Schwenter, 2016a, 26: 10)2.1.3 SummaryThis first part has introduced one of our protagonists: DM si. Si-clauses behavesyntactically and prosodically like root clauses, and are used in contexts that re-fute a previous contribution to discourse. The literature on this DM has typicallyfocused on its adversative flavour; however, Schwenter (2016b) proposes that thecore meaning of the DM is to mark an obvious truth to the Speaker, a meaning thatyields adversativity when used in certain contexts (such as when the Addressee iswrong). In Chapter 3 I follow this last line of thought and formalize the coremeaning of DM si with respect to the effect it has on the CG. Before we do that,let us turn to our second protagonist: whereas independent si-clauses are used tomark adversativity, no-tagged clauses are used to do the opposite: seek alignmentwith the Addressee.2.2 Peninsular Spanish no: where we find it andprevious accountsThe lexical item no has several functions in Spanish, summarized in Figure 2.2,and exemplified in (48). It encodes sentential negation, it is used as the negativeresponse marker, and it can also be used as a pragmatic marker with two bigfunctions: (i) requesting confirmation, which we have called DM no, and (ii) tokeep the interlocutor engaged (phatic no). The focus of this dissertation is on thefirst function (its confirmational use).55NOPolarity MarkerSentential negation Response MarkerPragmatic MarkerConfirmational PhaticFigure 2.2: Schema of the different functions of no.The word no is used to negate the sentence in (48a), and as a negative responsemarker in (48b). The last two examples are the functions of no as a discoursemarker: in (48c) the Speaker is requesting the Addressee to confirm a proposition(in this case, p=\u2018Adriana\u2019s cat is black\u2019) whereas in (48d) the Speaker is trying tokeep the Addressee\u2019s attention:(48) a. ElthegatocatdeofAdrianaAdriananoNEGesisblanco.whiteSENTENTIAL NEGATION\u2018Adriana\u2019s cat is not white.\u2019b. No,noelthegatocatdeofAdrianaAdrianaesisnegro.blackRESPONSE MARKER\u2018No, Adriana\u2019s cat is black.\u2019c. ElthegatocatdeofAdrianaAdrianaesisnegro,blackno?noCONFIRMATIONAL\u2018Adriana\u2019s cat is black, [no?].\u2019d. Elthegatocatesisnegro,blackno?noyandunapocobitgordito.fat.DIMPHATIC\u2018Adriana\u2019s cat is black, [no], and a little chubby.\u201956Although the focus of this dissertation is on its role as a confirmational maker,in the next sections I will give a quick overview of the other uses of no, startingwith its use as a (negative) response marker and as marker of sentential negation.2.2.1 Where we find noNo is found across a broad range of contexts, which I briefly survey. This includesits use as a negative response marker (section 2.2.1.1) and as a pragmatic marker(section 2.2.1.2). Under pragmatic marker I include its function as a phatic marker(section 2.2.1.2.1 and as a confirmational DM (section 2.2.1.2.2). In the latterfunction, it can anchor to a broad range of sentence types, including declaratives(2.2.1.3), imperatives (2.2.1.4), interrogatives (2.2.1.5) and exclamatives (2.2.1.6).2.2.1.1 No and propositional polarityThe use as of no as a negative response marker is exemplified in (49)-(52). Dif-ferent speech act types will allow or disallow the use of the response marker asa reaction; this distribution is summarized in Table 2.2. Here, \u2018enriched\u2019 meansthat the Speaker needs to add more information to a bare response marker for theanswer to be felicitous\u2013that is, a mere yes or no would be too short of an answer,and a more elaborate answer is typically favoured.57Type of speech act Bare no as a response Enriched no as a responseAssertion (affirmative) 3 3Assertion (negative) 3 3Assertion (evaluative) # #Polar question 3 3Wh-question # #Command 3 3Expressive # # \/ 3Commissive # 3Declaration # 3Table 2.2: Felicity of responding with no to different speech act types.The response marker no can be used to respond to polar questions, assertions(affirmative and negative), and commands.2(49) a. A: Man\u02dcana es jueves? POLAR QUESTION\u2018Is tomorrow Thursday?\u2019B: No.b. A: Man\u02dcana es jueves. ASSERTION (AFFIRMATIVE)\u2018Tomorrow is Thursday.\u2019B: No(, es mie\u00b4rcoles).\u2018No, (it\u2019s Wednesday).\u20193c. A: Man\u02dcana no es jueves. ASSERTION (NEGATIVE)2In the following discussion about no as a response marker, there is not morpheme gloss butdirectly the translation. In these cases, and since the focus is on the content of the examples andnot so much the form, the translation offers sufficient evidence for the point that I am trying tomake.3It seems pragmatically unlikely to only use the response marker and not give the \u201ccorrect\u201dinformation, but it is not infelicitious.58\u2018Tomorrow is not Thursday.\u2019B: No.4d. A: Tra\u00b4eme un cafe\u00b4. COMMAND\u2018Bring me a coffee.\u2019B: No.It is not felicitous to use the response marker no to respond to an expressivespeech act such as thanking, or to respond to an evaluative assertion, as shown in(50), where it is also indicated that a fuller response (one that includes a reasonfor the rejection besides the response particle) is accepted only for utterances thatinclude a predicate of personal taste:(50) a. A: Muchas gracias! EXPRESSIVE (THANKING)\u2018Thank you so much!\u2019B1: #No.B2: # No, no me cuesta nada.B3: No me cuesta nada.\u2018(No), it didn\u2019t cost me a thing.\u2019b. A: Estoy cansada. INTERNAL STATE\u2018I am tired.\u2019B1: #No.B2: ?No, no puedes estar cansada, te acabas de levantar.B3: No puedes estar cansada, te acabas de levantar.\u2018(No), you cannot be tired, you just woke up.\u20194Continuations could be either agreeing with the proposition that it is not the case that it isThursday (No, es cierto, no es jueves. Es mie\u00b4rcoles \u2018No, it\u2019s true, it\u2019s not Thurday. It\u2019s Wednes-day.\u2019) or disagreeing with it (No, no es cierto, s\u0131\u00b4 que es jueves. \u2018No, it\u2019s not true, it IS Thursday.\u201959c. A: Este arroz esta\u00b4 riqu\u0131\u00b4simo. TASTE\u2018This rice is delicious.\u2019B1: ?No.B2: No, se han pasado con la sal.\u2018No, they put too much salt.\u2019It is important to note that there are certain predicates of personal taste thatare less infelicitous to respond to with no, such as the prototypical example \u2018tobe tasty\u2019 (Lasersohn, 2005). Especially if not explicitly perspectivized to expressthe speaker\u2019s opinion, these tasty utterances are more flexible in their acceptanceof a negative response marker. For example in (51), Speaker A states that theyfind a rice dish tasty, and Speaker B replies with no and a reason why Speaker Bdoes not share this opinion (the dish is too salty). Speaker A acknowledges B\u2019sopinion, and settles for an \u201cagree to disagree scenario\u201d where Speaker A thinksthat the rice is delicious, and Speaker B thinks that it\u2019s too salty. This type ofresolution is sometimes referred to as faultless disagreement (Ko\u00a8lbel, 2004).5(51) A: Este arroz esta\u00b4 riqu\u0131\u00b4simo.\u2018This rice is delicious.\u2019B: No, se han pasado con la sal.\u2018No, they put too much salt.\u2019A: Bueno, pues a m\u0131\u00b4 me gusta.\u2018Well, I like it.\u2019What is important to note here is that rejecting an assertion with an internalstate predicate is not possible (shown in (50b)), whereas rejecting an assertion5The behaviour of predicates of personal taste, and specifically their behaviour with respectto disagreement, has been the focus of much work that is unfortunately outside the scope of thisdissertation (among others, Lasersohn (2005); Stephenson (2007); Stojanovic (2007); Lasersohn(2009); S\u00e6b\u00f8 (2009); Moltmann (2010); Crespo and Ferna\u00b4ndez (2011); Schlo\u00a8der (2018)).60with a predicate of personal taste that could be understood as reflecting the tasteof all interlocutors (as in (51)) is possible. This difference between a propositionintrinsically Speaker-oriented and one that can have the potential of being sharedwill be important for the analysis of DM no in Chapter 4.It is also infelicitous to respond to a commissive speech act (such as a promise)with a bare response marker no, as well as to a declaration, as is shown in (52):(52) a. A: Te lo prometo. COMMISSIVE\u2018I promise.\u2019B1: #No.B2: No, no puedes prometerlo porque no puedes hacerlo.\u2018No, you cannot promise it because you cannot do it.\u2019b. A: Os declaro unidos en matrimonio. DECLARATION\u2018I declare you joined in matrimony.\u2019B1:#No.B2: No, no tienes jurisdiccio\u00b4n aqu\u0131\u00b4.\u2018No, you have no jurisdiction here.\u2019To end the section on the relationship between no and negation, Spanish usesthe same item\u2013no\u2013both as a negative response particle, and as sentential negationmarker. This is illustrated in (53), where the first instance of no is as a responsemarker (and hence glossed in English as \u2018no\u2019) and the second is an instance of theuse of no as sentential negation\u2013I will gloss this as NEG.(53) Context: Addressee asks whether today is Thursday.No,nonoNEGesisjueves.thursday\u2018No, it\u2019s not Thursday.\u201961Sentential negation is marked pre-verbally with no. This negative marker li-censes other negative elements such as an n-phrase\u2013nada, nadie, ningu\u00b4n\/ninguno\/ninguna,corresponding to nothing\/anything, nobody\/anybody, noone\/anyone in English.(54) a. No salio\u00b4 nadie.NEG go.out-PST.3SG nobody\u2018Nobody went out.\u2019b. No ocurrio\u00b4 nada.NEG happen-PST.3SG nothing\u2018Nothing happened.\u2019c. No salio\u00b4 ningu\u00b4n estudiante.NEG go.out-PST.3SG no student\u2018No student went out.\u2019 (Mart\u0131\u00b4n-Gonza\u00b4lez, 2003)2.2.1.2 No and discourse markingBeyond its use as a response marker or as a sentential negation marker, no alsohas a discursive role. The literature highlights two main ways in which no helpsin the organization of discourse: as a phatic marker (controlling contact betweeninterlocutors) and as a confirmational marker (requesting confirmation of a fact oran opinion).66I follow Montan\u02dcez Mesas (2008) in assuming that the expletive use of no that Garc\u0131\u00b4a Vizca\u0131\u00b4no(2005) identifies (that is, when it is used as a filler) is just a lack of function, instead of a functionin itself.62(55) PHATICYandquethatsuertelucktienenhaveellostheyde...of...de...of..deofpoderbe.ableelegir,chooseno?noporbecause.oflathenorma...norm...religiosareligious\u2018and how lucky are they of being able to choose [no], because of the reli-gious norm\u2019 (Garc\u0131\u00b4a Vizca\u0131\u00b4no, 2005, p. 95; my translation)(56) CONFIRMATIONALBuenowelltu\u00b4youtieneshaveunabuengoodcoche,carno?no\u2018Well, you do have a good car, [no?]\u2019 (Rodr\u0131\u00b4guez Mun\u02dcoz, 2009, p. 90; mytranslation)The main difference between the phatic and confirmational use is whether ornot there is a request for interaction: confirmational no asks for a response fromthe Addressee\u2013regardless of whether this response actually takes place or not. Onthe other hand, phatic no is mainly used to keep or secure contact with the inter-locutor, and does not request any form of action from the Addressee. In termsof formal differences, DM (confirmational) no occurs at the end of the utterance,whereas phatic no occurs utterance-medially. In terms of intonation, the litera-ture claims that DM no bears rising intonation (Ortega, 1985; Montan\u02dcez Mesas,2008; Go\u00b4mez Gonza\u00b4lez, 2014; Montan\u02dcez Mesas, 2015), but there is usually lit-tle mention of the intonation of the phatic use of no. Montan\u02dcez Mesas (2015,p.223) mentions that for the pragmatic marker eh (which can also appear turn me-dially and turn finally) there are no differences in intonational contour betweenthe different turn positions, although the turn-final one seems to have a higher risethan the medial one. Given the similarities between Spanish DM eh and DM no Iwill assume that the same intonational patterns would apply to DM no compared63to phatic no, marked in Table 2.3 by adding a + sign for DM no7. This is notcrucial for the analysis of DM no proposed here, since I will not focus on non-confirmational uses of the DM.8 For this dissertation, I assume that the fact thatthe Speaker is eliciting a response from the Addressee is encoded in the high risingintonation of the DM no, an analysis proposed for similar confirmational markersand Addressee-oriented speech acts (Gunlogson, 2004; Heim et al., 2014, 2016;Wiltschko and Heim, 2016; Heim, 2019a).Table 2.3: Formal differences between phatic and confirmational functionsof no.Wants response? Position IntonationPhatic no No Medial Speculated: risingConfirmational no Yes Final Rising+Although the focus of this dissertation will be the confirmational use of no, Ibriefly illustrate the phatic use in 2.2.1.2.1, and then focus on the confirmationaluses of DM no in 2.2.1.2.2. In the rest of the dissertation I use \u201cDM no\u201d to referexclusively to the confirmational use of the DM.2.2.1.2.1 Non-confirmationalThe non-confirmational or phatic function of no is used to keep or secure contactwith the interlocutor. It does not require an answer from the Addressee. In (57),Speaker A uses no in the middle of their turn: speaker A does not use no turn-finally to call on the participation of the Addressee, but uses it turn-medially\u2013thefact that they are not done talking can be seen by the quick continuation startingwith y \u2018and\u2019:(57) A: Porque resulta, sabes? que ha habido problemas, no?, y all\u0131\u00b4 y con eso,y todas esas historias, esta\u00b4 [un poco]7This is similar to Canadian eh (Wiltschko, p.c.).8See Cabedo Nebot (2013) for a different claim, where it is pauses, and not pitch contours,which contribute to the meaning of DM no.64\u2018Because it turns out, you know, that there have been problems, [no?], andwith all that, and those stories, he\u2019s a little bit.\u2019A: [S\u0131\u00b4], si ese es el rollo!\u2018Yes, that is the deal!\u2019 (Rodr\u0131\u00b4guez Mun\u02dcoz, 2009, p.91; my translation)Garc\u0131\u00b4a Vizca\u0131\u00b4no (2005) defines the phatic use of no as securing contact withthe Addressee and making sure that they are paying attention to the message. Itdoes not want to elicit a response from the Addressee, but the speaker in (57) isstill making contact with the interlocutor. This type of functions is sometimes re-ferred to as the narrative use of certain discourse particles, for example Canadianeh (Heim et al., 2014; Denis et al., 2016). The narrative use is also characterizedby a lack of rising intonation and a non-peripheral position (Lam et al., 2013; De-nis et al., 2016). Crucially, and in contrast to the confirmational use of DM no,this phatic use does not elicit the partipation of the Addressee (Montan\u02dcez Mesas,2015).2.2.1.2.2 ConfirmationalThe confirmational use of DM no (which from now on will be the function thatI will refer to when I refer to DM no), which is the focus of this dissertation,typically has the following structure, consisting of an anchor and the rising DM noattached to the anchor in final position, as in Figure 2.3 (repeated from Figure 1.1):Figure 2.3: Anchor and DM no.Utterances of this form are typically called tag questions, and the appendedelement at the end is referred to as the tag.9 The tag can take the form of a clause9This brings up the question of whether the term \u2018tag\u2019 only refers to a shortened clause (as in65or, as in our case, a word (Kimps, 2018). In contrast with some tags in otherlanguages, the use of DM no is not restricted by the polarity of the host clause.This is the case for English reverse-polarity tags, where a positive anchor will befollowed by a negative tag (58a), and vice versa (58b):(58) a. Luis lives in Sweden, doesn\u2019t he?b. Luis doesn\u2019t live in Sweden, does he?Similarly, the French tag non? is incompatible with negative sentences:(59) #MarieMarien\u2019estNEG.ispasNEGpartie,leftnon?NON\u2018Marie hasn\u2019t left, [non?]\u2019 (Beyssade, 2012)However, Spanish confirmational no does not seem to be restricted in this way.In the minimal pair presented in (60) polarity is changed, and the tag is felicitouswith both:(60) a. LuisLuisvivelive.PRS.3SGeninSuecia,Swedenno?no\u2018Luis lives in Sweden, [no?]\u2019b. LuisLuisnoNEGvivelive.PRS.3SGeninSuecia,Swedenno?no\u2018Luis does not live in Sweden, [no?]\u2019But what exactly are we confirming? What can be the anchor of the DM? Thisis the focus of the next sections, where we will see that DM no can appear withany sentence type.the English tag questions) or whether it can also refer to appended word, as Kimps (2018) suggests.I take an approach closer to Kimps\u2019, understanding \u2018tag\u2019 as any appended element that is added toan anchor.662.2.1.3 Declaratives can be anchors of DM noThe co-occurrence of no with declaratives is the one that appears most frequently,as reported by Go\u00b4mez Gonza\u00b4lez (2014). Although she doesn\u2019t refer to sentencetypes, Garc\u0131\u00b4a Vizca\u0131\u00b4no (2005) also reports that confirmational no is used morefrequently to confirm \u201cfacts\u201d. Examples of declaratives (both affirmative and neg-ative) followed by no are given in (61):(61) a. Man\u02dcanatomorrowesismartes,Tuesdayno?no\u2018Tomorrow is Tuesday, [no]?\u2019b. Man\u02dcanatomorrownoisesnotmartes,Tuesdayno?no\u2018Tomorrow is not Tuesday, [no]?\u2019In (61), the Speaker wants the Addressee to confirm the fact that the next dayis Tuesday. Note that this is not the same as wanting the Addressee to answerwhether or not the next day is a Tuesday: this would be accomplished by the polarquestion Man\u02dcana es martes? (\u2018Is tomorrow Tuesday?\u2019). Instead, what the Speakeris doing in (61) is eliciting a confirmation from the Addressee about a fact.2.2.1.4 Imperatives can be anchors of DM noDM no can appear with an imperative anchor, like the example in (62) illustrates:(62) E\u00b4chatethrow.IMP.2SG=REFLma\u00b4s,moreno?no\u2018Help yourself to more, [no]?\u2019 (Go\u00b4mez Gonza\u00b4lez, 2014)Not all imperative anchors are felicitous with DM no, however: Garc\u0131\u00b4a Vizca\u0131\u00b4no(2005) notes that there is a gradience in the acceptability of the tag with commands67as a whole (illustrated here in Table 2.4). Warnings cannot be hosts of no, whereaspieces of advice and suggestions can. I have modified, however, her judgments:although I agree that the tag is \u201cpragmatically invalid\u201d with the more imposingdirectives (warnings), the rest seem pragmatically valid and not \u201cdubious\u201d as shejudges them.10Table 2.4: Modified from Garc\u0131\u00b4a Vizca\u0131\u00b4no (2005): gradience in theacceptability of no and directives.Imposition Type of directive Acceptability+ Warning #Te lo advierto: estudia, no?(Advertencia) \u2018I warn you: study, [no?]\u2019Warning #Si no estudias no sales, no?(Aviso) \u2018If you don\u2019t study you don\u2019t go out, [no?]\u2019Order Estudia ma\u00b4s, no?(Mandato) \u2018Study more, [no?]\u2019Petition Por favor, estudia ma\u00b4s, no?(Peticio\u00b4n) \u2018Please, study more, [no?]\u2019- Advice Deber\u0131\u00b4as estudiar ma\u00b4s, no?(Consejo) \u2018You should study more, [no?]\u2019In Table 2.4 we see that the same command anchor Estudia ma\u00b4s, no? \u2018Studymore, [no?]\u2019 has different acceptability depending on whether it expresses astrong imposition (as in a warning) or a weak imposition (as in a petition).2.2.1.5 Interrogatives can be anchors of DM noDM no can have an interrogative clause as a host (Go\u00b4mez Gonza\u00b4lez, 2014). Thisis the case in (63):10I have also changed the symbol she uses to mark that an utterance is pragmatically invalidfrom an asterisk (*) to a pound\/hashtag (#).68(63) Deofque\u00b4whatesta\u00b4ishablando,talkingno?no\u2018What is he talking about, [no]?\u2019However, only a specific type of interrogatives can be hosts of DM no, namelyrhetorical questions. Rhetorical questions differ from canonical questions in thatthey are not information seeking.How do we know that the anchor in (63) is a rhetorical question and not acanonical, information-seeking question? There are two reasons: one is the typeof context that this utterance is licensed in, and another is the possibility of addingexpressive constructions typically associated with rhetorical questions\u2013both areillustrated in (64), taken from (Bertrand et al., 2015):(64) 3 Context 1: A is sitting beside B listening to a TED talk about a topic shedespises. She looks at B and asks:H7 Context 2: A arrives late to a TED talk and has no idea what the speakeris talking about. She asks B:HDeofque\u00b4whatesta\u00b4ishablando,talkingno?no\u2018what is he talking about, [no]?\u2019In Context 1, the Speaker is not really expecting an answer from Speaker B:she is there at the same talk, listening to it. What she is expressing is, in fact, anopinion on the talk itself; specifically, her dislike of the talk. In Context 2, SpeakerA is genuinely asking Speaker B about the content of the talk: she wants to knowwhat the speaker has been saying.Notice that the unmodified anchor is felicitous in both contexts, but cruciallyit is also felicitous in an information-seeking context\u2013whereas the modified utter-ance is not.69(65) 3 Context 1: A is sitting beside B listening to a TED talk about a topic shedespises. She looks at B and asks:3 Context 2: A arrives late to a TED talk and has no idea what the speakeris talking about. She asks B:Deofque\u00b4whatesta\u00b4ishablando?talking\u2018What is he talking about?\u2019In fact, comments from speakers insisted that the sentence in (64) wouldgreatly improve if it was preceded by pero (\u2018but\u2019), or pero que\u00b4 demonios (\u2018butwhat in the hell\u2019) as in Pero de que\u00b4 demonios esta\u00b4 hablando? (\u2018But what in thehell is he talking about?\u2019), further evidencing that it is rhetorical questions, andnot information-seeking questions, that can be hosts of DM no. I assume, there-fore, that the host of the confirmational in (64) is a rhetorical question and not acanonical information seeking question.2.2.1.6 Exclamatives can be anchors of noDM no can modify exclamatives, like in example (66), where the DM follows thewh-exclamative expressing that the Speaker is feeling cold:11(66) Oye,hear.IMPque\u00b4whatfr\u0131\u00b4ocoldhacemakesaqu\u0131\u00b4!,hereno?no\u2018Hey, it\u2019s freezing in here, [no]?\u2019 (Rodr\u0131\u00b4guez Mun\u02dcoz, 2009, p.90;C2:3)It is important to note here that not all exclamatives accept the addition ofDM no. If the exclamative contains an internal state predicate, as in (67), thejudgement changes and it is no longer felicitous:11Although exclamatives are not one of the main sentence types that are found cross-linguistically, I add them here to give a better overview of what type of hosts that DM no cananchor to.70(67) #Que\u00b4howcansadatiredquethatestoy,amno?no\u2018I\u2019m so tired, [no]?\u2019We see that DM no does not seem to be sensitive to the sentence type, as it canfollow exclamatives like (66). However, it is sensitive to the type of exclamative:it cannot modify exclamatives that denote an internal state of the Speaker like(67).2.2.2 What DM no does: previous accounts2.2.2.1 DM no requests confirmation of facts or opinionsThe literature has noted two different functions of DM no as a confirmational:when the Speaker is asking for confirmation of a fact, as in (68), or when theSpeaker is requesting the Addressee\u2019s opinion on an evaluation, as in (69). I willadopt the terms used in Go\u00b4mez Gonza\u00b4lez (2014) to distinguish these types: facticand opinionative. In (68), the Speaker is asking the Addressee to confirm the factthat they have indeed a common group of friends (the Speaker in this context is atarot reader); in (69), the Speaker is not requesting the confirmation of a fact but ofa personal evaluation, signalled here by the use of the verb creer \u2018to think\/believe\u2019:(68) CONFIRMATIONAL (FACTIC)eh,ehmemesalecome.upestothisaqu\u0131\u00b4hereenonlasthecartas,cardsustedesyou.2PL.FORMtienenhaveunagrupogroupdeofamistadesfriendshipsenincomu\u00b4n,commonno?no\u2018Eh, this is on the cards, you have a common group of friends, [no]?\u2019(Rodr\u0131\u00b4guez Mun\u02dcoz, 2009, p. 89)71(69) CONFIRMATIONAL (OPINIONATIVE)Bueno,wellyoIcreothink.PRS.1SGquethatlasthedostwocosas,thingsno?no\u2018Well, I think both, [no]?\u2019 (Garc\u0131\u00b4a Vizca\u0131\u00b4no, 2005, p. 93)2.2.2.2 DM no can be Speaker or Addressee-orientedGo\u00b4mez Gonza\u00b4lez (2014) introduces an even more fined-grained distinction of thefunctions of tag questions, under which she includes the DM no.12 As we can seein Table 2.5, her main distinction is between Addressee-centered and Speaker-centered functions, which roughly correlate with the factic and opionionativefunctions identified by Garc\u0131\u00b4a Vizca\u0131\u00b4no (2005) and Rodr\u0131\u00b4guez Mun\u02dcoz (2009).Within those two groups, Addressee-centered functions can seek to confirm in-formation or an action; Speaker-centered functions are used \u201cas stance markersto express either strong positive or negative affect.\u201d (Go\u00b4mez Gonza\u00b4lez, 2014,p.116)1312I am setting aside the phatic function, which she analyzes as exchange-centered; I include itin the table for completeness.13Go\u00b4mez Gonza\u00b4lez (2014) claims that the Speaker-oriented function is associated with a fallingintonation, but there does not seem to be evidence to suggest it is. Given the several other studiesthat identify DM no as having rising intonation, I set this question aside for further research.72Garc\u0131\u00b4a Vizca\u0131\u00b4no (2005) &Rodr\u0131\u00b4guez Mun\u02dcoz (2009)Go\u00b4mez Gonza\u00b4lez (2014)Factic Addressee-centeredInformation\/Confirmation seekingAction seekingOpinionative Speaker-centeredAttitudinalChallenging(Phatic) (Exchange-centered)(Focusing)(Phatic)(Regulatory\/Delaying)Table 2.5: Previous analyses of the use of DM no? (based onGo\u00b4mez Gonza\u00b4lez (2014, p. 114).However, this does not mean that there are no restrictions on the use of DMno: although there is no restriction on the sentence type that can be the anchor,there is a restriction on the speech act type.2.3 Where to go from hereDMs si and no perform complex conversational updates: they seem to have morethan one effect on the conversation:1. Clauses with DM si react to a previous discourse contribution, refuting somepart of its meaning, while at the same time marking that the propositionthey introduce is uncontroversial. I will take these two characteristics as thetwo main ingredients in a unified analysis of DM si: (i) the trigger makesapparent that there is an epistemic misalignment between the interlocutors,and (ii) the Speaker assumes that the proposition that DM si introduces isuncontroversial (or, in Schwenter (2016b), obvious). The question of how73we can give a unified account of what counts as an epistemic conflict andwhat is an uncontroversial proposition will be addressed in section 2.3.1,and how we can formalize these two characteristics will be explained insection 2.3.3. A more fully fleshed-out account of DM si will be providedin Chapter 3.2. Clauses with DM no also propose a complex conversational update: theyrequest the confirmation of a fact or an opinion from the Addressee, sothey seem to assert that fact or opinion, and at the same time request theAddressee\u2019s input about that fact or opinion. These two functions will beunified as one in section 2.3.2 and in more detail in Chapter 4.2.3.1 A unifying analysis for DM siI propose an analysis of DM si with two main components: (i) the role of thesi-clause with respect to the previous discourse, and (ii) the role of the DM siwith respect to the sentence it appears in. What connects these two sides of theanalysis is the concept of felicity conditions and the CG: the si-marked clause willbe triggered by a violation of a felicity condition, and the DM si will mark thatthe proposition it introduces is already in the CG.Montol\u0131\u00b4o Dura\u00b4n (1999) proposes that independent si-clauses react to the lackof relevance of the previous contribution: for example, in (70), the si-Speaker canreact with a si-clause to an imperative (\u2018Give me more cake\u2019) that is not relevantin the dialogue context because (a) there is no more cake; (b) the cake is not tasty;(c) the Addressee (the one wanting the cake) is on a diet; and (d) the Addresseehas already eaten too much cake:(70) Pa\u00b4samepass.IMP.2SG=meotroanothertrozopiecedeofpastel,cakeporforfavor.favour\u2018Get me another piece of cake, please.\u201974a. SisiyaalreadynoNEGhayisma\u00b4s.more\u2018[Si] there\u2019s no more left.\u2019b. Sisiesta\u00b4ismal\u0131\u00b4simo.bad.SUPER\u2018[Si] it\u2019s awful.\u2019c. Sisiesta\u00b4sareaPREPre\u00b4gimen.diet.\u2018[Si] you\u2019re on a diet.\u2019d. Sisiyaalreadyhashave.2SGcomidoeatenmucho.much.\u2018[Si] you\u2019ve already had a lot\u2019(Rodr\u0131\u00b4guez Ramalle, 2011, p. 211; ex. 24; based on Montol\u0131\u00b4o Dura\u00b4n(1999))These possible independent si-clauses show something more specific: I pro-pose that what they are targeting is a felicity condition that has been violated in theprevious contribution.14 I will use Searle (1969, 1976) definition of felicity con-ditions, in which he distinguishes four types: propositional content conditions,preparatory conditions, sincerity conditions, and essential conditions.15 DM sican, in fact, react to the non-observance of these conditions for all major typesof speech acts, including assertions, commands, questions, commissives, expres-sives, and declarations. This analysis also generalizes to examples where the si-clause is used to react to a failed implicature, a mispronounced word, and also tomirative examples, where there is apparently no Addressee.An important point of this dissertation, and a major point in Schwenter (2016b),14Following the proposal for German doch in Egg (2013).15See Chapter 1 for definitions of each.75is that we need to distinguish what the role of the si-clause is with respect to theprevious discourse (which, as I have just claimed, is to flag that the previous dis-course move is infelicitous) from the contribution of the DM si to the clause itheads. I claim that we can unify all uses of DM si by proposing that it marks thatthe proposition denoted by the sentence it introduces is already in the CG. The no-tion of the CG will be crucial in the analysis of si-clauses and the DM si because itwill help us understand (a) the licensing conditions for the si-clause (that is, whatdoes the discourse context have to look like for an interlocutor to be able to usea si-clause felicitously); and (b) the contribution of the DM si to the sentence itintroduces.The first question that we might want to ask ourselves is how do we model whythe si-Speaker reacts to a violation of felicity conditions. Even before that, howdo we model how such a violation would happen? I will propose that a violationof felicity conditions can be analyzed as what Stalnaker (2002) calls a defectivecontext: a context in which it becomes apparent that not all interlocutors share thesame CG. If we assume that felicity conditions are \u2018stored\u2019 in the CG as generalbackground information (which would also include world knowledge), when aninterlocutor perceives another interlocutor\u2019s contribution as infelicitous, what theyare identifying is a conflict in the CG. We therefore have to model that there areat least two CGs: one representing what the Speaker thinks how the CG lookslike, and one representing what the Speaker thinks the Addressee\u2019s conception ofthe CG is. This is a rather recursive notion, which is simplified in an image inFigure (2.4). Note that the CG \u2018according to the Speaker\u2019 is not marked as suchin this model: one can argue that interlocutors tend to think that their conceptionof what is in the CG is the one that is \u2018right\u2019, and therefore their version of whatshould be in the CG is the one that should prevail.76Figure 2.4: Two CGs.I propose we also need the CG to give a unified account of the contribution ofDM si to the clause it introduces. I propose that DM si marks that the propositiondenoted by the sentence it introduces is already in the Common Ground. In Chap-ter 3 we will see evidence to support this claim, including the impossibility of DMsi to introduce \u2018hot news\u2019 (Schwenter, 2016b) and the infelicity of using DM si tointroduce propositions that are known not to be shared by the Addressee.2.3.2 A unifying analysis for DM noWe have seen that DM no (i) can take any sentence type as its anchor, and (ii) hasbeen mainly analyzed as a marker of contact control (Briz, 1998; Montan\u02dcez Mesas,2015). DM no has been described as having many functions, the main ones sum-marized again in Table 2.6. The question is: is there a core abstract function ofthe DM no that can account for these different functions, that also predicts therestrictions in its anchor that have been identified in the literature (its use onlywith certain types of commands) and in this chapter (its use only with rhetoricalquestions, but not information-seeking questions)?77Garc\u0131\u00b4a Vizca\u0131\u00b4no (2005) &Rodr\u0131\u00b4guez Mun\u02dcoz (2009)Go\u00b4mez Gonza\u00b4lez (2014)Factic Adressee-centeredInformation\/Confirmation seekingAction seekingOpinionative Speaker-centeredAttitudinalChallengingPhatic Exchange-centeredFocusingPhaticRegulatory\/DelayingTable 2.6: Previous analyses of the use of DM no (based onGo\u00b4mez Gonza\u00b4lez (2014)).I propose an analysis of DM no where its core meaning is to request confir-mation of the conversational update proposed by the anchor. In other words: DMno cares about the speech act type of its anchor, but not its sentence type. We justsaw evidence for this in section 2.2.1.5: DM no can appear with an interrogativeanchor only if this interrogative anchor is a rhetorical question. In some analyses,rhetorical questions have interrogative form but assertive illocutionary force (Han,2002; Corr, 2016). This means that the conversational update that they propose isnot that of a question, but of an assertion. In Chapter 1 I mentioned that differ-ent speech act types propose different conversational updates (Farkas and Bruce,2010): an assertion proposes to add a proposition p to the CG; a polar questionproposes to resolve whether p or \u00acp should be added to the CG; a command pro-poses to add an outcome to the TDL of the Addressee (Rudin, 2018). I will extendthis idea to other speech act types, namely expressives (in Searle\u2019s terms, for in-stance thanking and apologizing), commissives, and declarations. As we will seein Chapter 4, this correctly predicts where we find DM no.78The second part of the analysis to be presented in Chapter 4 will center aroundthe role of the no-tagged clause in discourse. Table 2.6 shows, in fact, the rolesthat no-tagged clauses have on discourse: as Denis et al. (2016) mention for Cana-dian eh, this is not a list of the functions of the DM no per se but of the no-taggedclauses. In Chapter 4 I propose that the unifying role of no-tagged clauses is toconfirm a positive bias, whether a bias towards a belief or an outcome. We cansee the role that bias plays in the licensing of no-tagged clauses in (71): in the firstcontext, the Speaker has evidence that the Addressee has the intention of comingto the party since they RSVP\u2019ed\u2013in other words, the Speaker believes that the Ad-dressee has the outcome of going to the party in their TDL (Yang and Wiltschko,2016). On the other hand, in Context 2 the Speaker has no such evidence, andcannot therefore be biased towards the Addressee having the intention of comingto the party:(71) 3 Context 1: Speaker has organized a party the next day, and has receivedan RSVP from the Addressee. She is confirming the guest list, and asks:7 Context 2: Speaker has organized a party the next day, and has not re-ceived an RSVP from the Addressee. She is confirming the guest list, andasks:Vienescome.2SGman\u02dcana,tomorrowno?no\u2018You\u2019re coming tomorrow, [no]?\u2019In order to model the licensing conditions for both DMs, we need to makesome modifications to the F&B Table Model presented in Chapter 1. These mod-ifications are presented in the next section.792.3.3 Tools for modelling Peninsular Spanish si and noIn this section I introduce the modifications to the Table Model proposed by Farkasand Bruce (2010) that I will implement for my analysis of the Spanish DMs. Thesemodifications do not change the nature of the model: they are merely adaptationsneeded in order to better formalize the data that the Spanish DMs offer.2.3.3.1 Phases to describe the contexts of useDMs like no and si are complex to model because they not only contribute some-thing to the clause they appear with, but they also restric and are restricted bythe previous and following discourse. We therefore need to represent these \u201cbe-fore DM\u201d and \u201cafter DM\u201d moments in the discourse to capture the contexts inwhich the DMs are licensed. I will refer to these \u201cbefore\u201d and \u201cafter\u201d moments asphases. Farkas and Bruce (2010) do represent different phases of a conversation(which they call context states) in order to model the effect of assertions and polarquestions on the conversation. However, with the contexts that we have discussedpreviously (misaligned information states) we need to pay closer attention to thephases before the DM is used. In these phases, which similarly to Farkas andBruce (2010) represent the context at different points in the conversation, inter-locutors need to keep a record of who is committing to what (Thoma, 2016; Heim,2019a).Phase 0: AlignmentCertain DMs (such as si) refer back to a symmetrical phase where there isno epistemic misalignment and the interlocutors share the same information(that is, they are aligned).Phase 1: MisalignmentInformation states of interlocutors are not maximally aligned.Phase 2: NegotiationSpeaker A tries to remedy the misalignment.80Phase 3: AlignmentInterlocutors are aligned.The following exchange in (72) illustrates these phases:(72) Context: A and B work together in an office 9-5, Monday to Friday. OnThursday, B tells A how they\u2019re looking forward to the next day becauseit\u2019s the weekend and they\u2019ll be able to sleep in. But the next day is Friday,not Saturday! A replies:Sisiman\u02dcanatomorrowesisviernes.Friday\u2018[Si] tomorrow is Friday\u2019Phase 0: SymmetrySpeaker A and Speaker B both know that tomorrow is Friday (p=\u2018Tomorrowis Friday\u2019). A can make this assumption based on the regularity of theirschedule, and the fact that people with a regular schedule tend to knowwhat day of the week it is.A thinks that . . . p is in the CG.. . . B thinks that p is in the CG.Phase 1: MisalignmentSpeaker B says something that makes it obvious to A that she does not, infact, know that tomorrow is Friday. In this case: saying that she will be ableto sleep in.B utters q, where q contradicts p.A thinks that . . . p is in the CG.. . . B thinks that q is in the CG.Phase 2: NegotiationA superasserts that p is in the CG by using DM si.81Speaker A superasserts p=\u2018Tomorrow is Friday\u2019.A thinks that . . . p is in the CG.. . . B thinks that q is in the CG.Phase 3: AlignmentEpistemic states are symmetrical again, with p being in the CG again.A thinks that . . . p is in the CG.. . . B thinks that p is in the CG.This is a simplified version of the analysis that will be presented in Chapter 4,but it serves as an illustration of the different phases.2.3.3.2 Committing to propositions and outcomesIn Chapter 1 I mentioned that Farkas and Bruce (2010) make a distinction betweenthe speakers\u2019 discourse commitments (individualized sets that contain proposi-tions to which each speaker has individually committed) and the (Stalnakerian)CG. These two sets are doxastic: they deal with beliefs, and contain propositions.However, expressing beliefs in propositions and working towards a shared set ofbeliefs is not the only aim of human interaction and conversation: we also want toact (together and separately). The Table Model and subsequent work has mainlyfocused on how to model beliefs and propositions, but has not focused on mod-elling other types of commitments. Rudin (2018) proposes a bifurcation of theTable Model to account for imperatives (both bare and modified via intonation):to the doxastic part of the Table Model, he adds a teleological part, where theSpeakers negotiate their public preferences.I adopt this modification to the Table Model, since we need a way to accountfor the effect that DM no has on anchors that do not denote propositions or nego-tiate beliefs, as in (73), where the anchor is an imperative:82(73) Pe\u00b4inate,comb.IMP.2SG=REFLno?no\u2018Brush your hair, [no]?\u2019Following Rudin (2018), I mirror the distinction between DCs and CG in the\u201ccommanding\u201d realm: we need both an individualized TDL for each discourseparticipant, and a shared, public TDL. Both components contain outcomes (fol-lowing Beyssade and Marandin (2006)): for instance, if Speaker A utters an im-perative, they would be proposing to add an outcome to Speaker B\u2019s TDL (TDLB),but if Speaker A utters a promise, they would add an outcome to their own TDL(TDLA). In both cases, they propose to add these outcomes to a public, sharedcomponent. I do not assume that these components form a separate half of the Ta-ble Model, however: I include all of them in my own model of discourse, which Iwill call the Dialogue Board, represented in Figure 2.5. This is an adaptation andfurther development of F&B\u2019s Table Model.The last ingredient that we will need for the Dialogue Board is how to rep-resent speaker mediation. I propose to visually mark that the conversations arealways mediated by the Speaker holding the floor at each turn. Most models al-ready assume this mediation, since it is impossible to know with absolute certaintywhat the Addressee knows. This allows for situations where there is an epistemicconflict in the CG (with DM si), and for situations where we want to refer tothe overall effect of the discourse move (with DM no). Regarding the first situ-ation: it is possible to imagine a context where it becomes obvious to a speakerthat their interlocutor does not share the same information, even if this informa-tion had been assumed to be shared. As we will see in Chapter 4, this is exactlywhere DM si is licensed: it encodes a meaning that can sometimes be paraphrasedas \u201cyou should have known this\u201d. Having two information states for each inter-locutor but having both mediated by the speaker holding the floor allows us tocapture this type of meaning: the fact that a speaker can make assumptions aboutwhat their interlocutor knows. With each utterance the Dialogue Board (KA) will83change; this change is a representation of the effect of the previous conversationalmove. I follow F&B\u2019s nomenclature, but in this case this K denotes the DialogueBoard without the Table, which encompasses the individual information states ofSpeaker and Addressee. I will only look at KA Dialogue Boards, which meansthat these Dialogue Boards are mediated by Speaker A. This will allow us to ex-plain the complex update that DM no makes on the conversation. The DialogueBoard is summarized in Figure 2.5.A (A\u2019s commitments)CURRENT PROJECTEDDC propositions propositionsTDLA outcomes outcomesCG propositions propositionsTDL outcomes outcomesB (B\u2019s commitments)CURRENT PROJECTEDDC propositions propositionsTDLB outcomes outcomesCG propositions propositionsTDL outcomes outcomesKAt= Speaker A\u2019s information state at time tTABLEFigure 2.5: The Dialogue Board.2.4 SummaryThis chapter, in addition to introducing the two protagonists (DM si and DM no)also introduced the friendly amendments that must be made to our formal dis-course model in order to analyze the contributions of these two DMs. The mostimportant finding is that each DM makes two types of contributions:i the contribution of DM si or no relative to the clause they introduce, andii the contribution of the si-clause or no-clause relative to the rest of the dis-course.84Chapter 3Monitoring the Common Ground:Peninsular Spanish si3.1 Ingredients of the analysisDMs are not obligatory, and this generalization also applies to DM si: the dialoguein (74) conveys the same propositional content with or without the use of DMsi\u2013namely, that contrary to what it seems that B believes, Toronto is in Canadaand not in the US:(74) Context: B is holding a US guide. They then say \u201cI can\u2019t wait to visitToronto!\u201d A replies:(Si)siTorontoTorontoesta\u00b4iseninCanada\u00b4,Canadaloitdiscutimosdiscuss.PST.1PLayer.yesterday\u2018([Si]) Toronto is in Canada, we talked about it yesterday.\u2019Why would a speaker, then, choose to use DM si in this context? One pos-sible reason, and the one I will propose, is that DM si restricts the possible linkbetween the proposition it introduces and the previous discourse. In particular,DM si specifies that the proposition it introduces is already in the CG. Without85the particle, this discourse-anaphoric link is left implicit. This can be seen frompossible continuations to (74): a bare assertion, as in (75), would merely proposean update of the CG, as we have seen in previous chapters. Therefore (75) couldbe used to inform the Addressee that Toronto is in Canada, without assuming thatthey already knew this\u2013and one can add me temo \u201cI\u2019m afraid\u201d, as one may do tointroduce \u201cunwelcome\u201d new information. However, it would be odd to continue asi-clause with this same expression:(75) TorontoTorontoesta\u00b4iseninCanada\u00b4,Canadamemetemo.fear.PRS.1SG\u2018Toronto is in Canada, I\u2019m afraid.\u2019(76) SisiTorontoTorontoesta\u00b4iseninCanada\u00b4,Canada??memetemo.fear.PRS.1SG\u2018[Si] Toronto is in Canada, I\u2019m afraid.\u2019There may be more than one way for a bare assertion to be linked to the pre-vious discourse: (75) may be understood as introducing new information, or asre-introducing old information that is meant to correct a previous discourse move(as in (74)). However, the si-clause in (74) and (76) can only be understood as thelatter.In this chapter I flesh out this proposal, distinguishing two contributions of DMsi: (i) the contribution of the si-clause to the conversation, and (ii) the contributionof the particle itself to the sentence. With respect to what si-clauses contribute toa conversational exchange, I propose that what unifies the different contexts ofuse of si-marked clause is that they are all used to reject a previous infelicitousspeech act. With respect to what the particle si contributes to the sentence, DMsi marks that the proposition it introduces is already in the CG. I argue that a si-clause is a stronger type of assertion\u2013as I will call it, a superassertion, since it86does not propose to add a proposition to the CG but reminds the Addressee that aproposition is already in the CG.(77) a. THE si-CLAUSE AS A UNIT (REJECTION)si-p rejects a previous infelicitous discourse move.b. THE PARTICLE si (UNCONTROVERSIALITY):DM si marks that the proposition p is already in the CG.1The uncontroversiality part of the meaning of DM si rejects the previous dis-course move. This part explains not only the \u201ccorrective\u201d flavour of DM si, butalso its reminding effect: it will depend on the context (whether the Addressee hasforgotten as opposed to having \u201cwrong\u201d beliefs) what effect the particle has.But why would a speaker want to mark that a proposition is already known? Ipropose that si-clauses are licensed when it becomes apparent that the interlocu-tors do not share the same information in the CG. More specifically, I claim thatthis misalignment stems from the violation of a felicity condition in the immedi-ately preceding move. This felicity condition violation makes apparent that thereis a misalignment with respect to the content of the CG: while A believes p, Bbelieves q, and q pragmatically contradicts p; this is illustrated in Figure 3.1. AsI will demonstrate at length, this analysis correctly predicts the use of DM si incontexts where various felicity conditions of different speech acts are violated.1The term \u2018uncontroversiality\u2019 is taken from Grosz (2010), who uses it to describe the Germanparticle ja. This is the same idea behind (Schwenter, 2016b)\u2019s analysis of si as marking that theinformation is obvious to the Speaker.87ACURRENT PROJECTEDDCATDLACG pTDLBCURRENT PROJECTEDDCBTDLBCG qTDLKA1TABLEFigure 3.1: The licensing context for DM si: Two versions of the CG.Crucially, p and q pragmatically contradict each other.The fact that the proposition introduced by DM si has to be known by theinterlocutors was highlighted as one of the main differences between DM si andother particles, such as the complementizer que (Schwenter, 2016b). In particular,it has been noted that DM si cannot be used when the proposition it introducesconveys new information, as the \u201chot news\u201d scenario in (78) shows:(78) Sabes que\u00b4? Que\/#si\/ \/0 mi hermana esta\u00b4 embarazada!\u2018You know what? {QUE\/SI}my sister is pregnant!\u2019 (adapted from Schwen-ter (2016b, p.26; 8a))In (78), the Speaker is announcing some news (their sister\u2019s pregnancy). Pre-sumably, the Addressee does not know about this information (hence the openingline Sabes que\u00b4? \u2018You know what?\u2019).2 DM si is not allowed in this type of context.Si-clauses therefore convey a \u201cyou should have known\u201d meaning that is notrepresented in the misalignment phases illustrated in Figure 3.1. We need a wayto reflect that, according to the Speaker, there was a previous stage in the discoursewhere the Addressee knew p. I therefore propose the addition of another phase to2In this same context the other Spanish complementizer que is allowed (see Chapter 6 for acomparison between si and que), as well as not having any complementizer present.88explain the use of si-clauses: a phase that represents an original aligned CG. Thisis a way to formalize the \u2018you should have known\u2019 flavour of DM si.3 This phaseis illustrated in Figure 3.2, and comes before Phase 1 which depicts the momentthe Speaker realizes that the CG is not symmetrical.ACURRENT PROJECTEDDCATDLACG pTDLBCURRENT PROJECTEDDCBTDLBCG pTDLKA0TABLEFigure 3.2: Phase 0: original alignment in the CG.With these two adjustments we can successfully account for the contexts ofuse of si-marked clauses, as we will see in section 3.3. But as I mentioned, thisis only one part of the contribution of DM si to the discourse: we still need toaccount for the contribution of the actual particle to the sentence it appears in. AsI have mentioned before, I propose to formalize previous insights in the literature(Rodr\u0131\u00b4guez Ramalle, 2011; Schwenter, 2016b) and propose that DM si marks thatthe proposition denoted by the sentence it appears in is already part of the CG.3.2 How to monitor the CG: reminding theAddressee of its contentIn this section I focus on showing how language users monitor the alignment ofthe CG. In section 3.2.1 I show how DM si reminds the Addressee about the status3Since we are dealing with a dialogue between two people,\u2018you should have known x\u2019 and\u2018everyone should know x\u2019 are the same situation in this case.89of a proposition as being already part of the CG, and what this means for the statusof the CG in the interlocutors\u2019 information states (in section 3.2.2).3.2.1 Reminding the Addressee that the proposition is alreadyin the CGWe have just seen that DM si cannot be used in a \u2018hot news\u2019 context, that is, itcannot introduce information that is known to be new for the Addressee. An-other piece of evidence that shows that DM si can only refer to propositions thatare already known to the Addressee\u2013i.e. in the CG\u2013comes from the infelicity ofusing a si-marked clause as possible response to the question in (79), where theproposition expressed in the sentence introduced by DM si cannot be known bythe Addressee:4(79) Context: A and B are in a speed dating event. As an ice-breaker, B asks:B: Prefieres carne o pescado?\u2018Do you prefer meat or fish?\u2019A: (#Si)(si)soyamvegetariana.vegetarian\u2018([Si]) I am vegetarian.\u2019In (79), B cannot be expected to know that A is vegetarian since they only justmet. It is therefore not felicitous to use DM si here, since the proposition p=\u2018A is avegetarian\u2019 is not a shared proposition in the CG. Note that a bare assertion wouldbe felicitous: in that case, A would be proposing the update of the CG with thenew information that she is vegetarian. I will therefore refer to what DM si doeswith respect to the proposition as superasserting it: it does more than just proposea CG update (as an assertion would do), and seems more imposing than assertions.4Always mediated by the Speaker: the Speaker believes that the Addressee knows p, but itmight not be the case.90In fact, Schwenter (2016b) mentions that si-clauses \u2018often act as \u201cblockades\u201d tothe topical thread\u2019 (Schwenter, 2016b, p.28): I claim that this is an effect of thesuper-assertion force that DM si brings.It is important to note that the Speaker may make the wrong assumption that aproposition is in the CG\u2013that is, the Speaker may think her interlocutors share thesame propositions. This may not always be the case: in the (pragmatically odd)situation where the same B in example (79) did use DM si (in the same context, aspeed dating event), A could easily reply that they had no way to have known thatfact, as in (228). This reply would not be felicitous if B had merely asserted thatshe is vegetarian:(80) B: Si soy vegetariana.si am vegetarian\u2018[Si] I am vegetarian.\u2019A: Yandco\u00b4mohowloitibago.IMPER.1SGatosaber?know\u2018And how was I supposed to know?\u2019Further evidence that DM si marks that a proposition is in the CG comes froma frequent reply to si-clauses: the use of ya (lit. \u2018already\u2019, which can often betranslated as \u2018I know\u2019 or \u2018yeah\u2019), as exemplified in (81):(81) Examples from Schwenter (2016b, p.28, (15) (16)):a. R: Ya los echo de menos!\u2018I already miss them!\u2019A: Si todav\u0131\u00b4a no se han ido a ninguna parte!\u2018[Si] they haven\u2019t gone anywhere yet!\u201991R: Ya, pero no los vamos a ver mucho ahora que se han graduado.\u2018Yeah, but we are not going to see them much now that they have grad-uated.\u2019b. G05: que a Guerri en la fa\u00b4brica de su padre le llamaban el gordo\u2018In your father\u2019s factory they called Guerri the fat guy\u2019G01: si no esta\u00b4 tan gordo\u2018[Si] he\u2019s not that fat\u2019G05: ya pero antes pesaba ma\u00b4s de cien kilos\u2018true but before he weighed more than one hundred kilos\u2019In both examples, the Addressee (R and G05) literally marks that the proposi-tion introduces by si (p=\u2018they haven\u2019t gone anywhere yet\u2019 in (a) and p=\u2018he\u2019s notthat fat\u2019 in (b)) is already known.Saying that a proposition is already part of the CG is a marked discoursemove: under \u201cnormal\u201d circumstances, one of the main goals of a conversation isto add new information and to add to the CG. However, there are certain contextsthat license this redundant use of information: one such context is when there isan epistemic conflict. This has been formalized in the Principle of Economy byRomero and Han (2004):(82) Principle of Economy: (Romero and Han, 2004, p. 629)Do not use a metaconversational move unless necessary (to resolve epis-temic conflict or to ensure Quality).This Economy constraint states that only certain discourse contexts allow theuse of metaconversational moves, i.e., moves that refer to the conversation itself.These are also the contexts where DM si is used. As we have seen in the previous92section, when B utters the antecedent utterance there ensues an epistemic con-flict: a proposition is being introduced to the CG that is in conflict with previouspropositions that are already in the CG.3.2.2 The status of the Common GroundAn important take-away of the use of DM si is that CG needs to be an independentelement in our discourse model that represents public, shared knowledge. In chap-ter 1 we introduced two ways of modelling the CG: it can either be an independentelement, separate from individual discourse commitments by each discourse par-ticipant (as I propose), or it can be seen as the intersection of the belief set of theSpeaker and the Addressee. These two possibilities are schematized in Figure 3.3,repeated from Figure 1.2:Figure 3.3: Two possibilities for the nature of CG: as an independentcomponent (left) or as the intersection of the interlocutor\u2019s belief sets(right).The fact that we use DMs like si can be seen as evidence for an independentCG in our model: if we did not have such an element where shared, public knowl-edge was stored, it would be odd not to tolerate the types of contexts were DM siis used.5 Furthermore: the fact that DM si seems to react to general knowledgetype of information (felicity conditions, which are known by hypothesis to fluentusers of a given language) also indicates that there needs to be a repository of5One could still model such a situation in the rightmost possibility in Figure 3.3, where theintersection between A and B is empty (assuming A and B are sets of possible worlds). However,this possibility does not explain the \u2018public\u2019 part of it, that is, the conflict between propositionsshould be public.93this type of information in our conversations, that we can refer back to when aninterlocutor seems to be violating the rules of the conversation.63.3 Why monitor the CG: rejecting infelicitousspeech actsWe have seen in Chapter 2 that DM si can react to (almost) anything that marksa defective context: the propositional content of the previous utterance, the pre-supposition and implicatures it introduces, and its (violated) felicity conditions.This patterns with the observation in Egg (2013) about the contexts of use of Ger-man doch, which can also react to defective speech acts. In his analysis, dochrelates two propositions: the one it introduces, and an antecedent proposition. Inthe cases where it responds to a non-declarative, this antecedent proposition isa felicity condition of the antecedent utterance, which are modelled as presup-positions (Searle and Vanderveken, 1985; Vanderveken, 1990). Another way inwhich Spanish DM si patterns similarly to German doch is in its reminding func-tion: Rojas-Esponda (2014) proposes that Germam doch is used to re-answer analready closed Question-Under-Discussion, which is a similar analysis to the oneoffered in this dissertation.7The unified schema for the use of DM si is given in (83), and will be appliedto the contexts of use of DM si discussed in Chapter 2.(83) PHASE 0: (ALIGNMENT)p is in CGPHASE 1: (MISALIGNMENT)B makes move \u00b56There is a debate about the cultural and linguistic specificity of felicity conditions; I assumehere that even if these are not the specific conditions that are known by language users, there aresome pragmatic rules that users of a specific language in a specific community know.7Both German doch and ja share similarities with Spanish DM si; a comparison of these parti-cles is unfortunately outside the scope of this dissertation.94\u00b5 is subject to a felicity condition qp pragmatically contradicts qPHASE 2: (NEGOTIATION)A utters si-p, which super-asserts pPHASE 3: (ALIGNMENT)8p is in CGThis analysis is reminiscent of the insights in Schwenter (1998) and Rodr\u0131\u00b4guez Ra-malle (2011) who paraphrase si-clauses as what others have called \u2018biscuit condi-tionals\u2019 (DeRose and Grandy, 1999) or \u2018premise conditionals\u2019 (Haegeman, 2003):conditionals that refer to the conditions of the illocutionary act. In this sense, si-clauses could be paraphrased as \u201cHow can you {assert, command, express, ask. . .} u IF p is in the CG.\u201dLooking at some of the examples of the use of DM si in the literature, a gen-eral thread is that it seems to be rejecting the previous speech act since a felicitycondition has not been met, as in (84):(84) A: \u00bfDo\u00b4nde esta\u00b4 tu hermano?\u2018Where is your brother?\u2019B:siSifinishPRS.1SGacaboPREPdearrivellegar.\u2018[si] I just arrived.\u2019 (Rodr\u0131\u00b4guez Ramalle, 2011))In this example, what A is basically conveying is that she cannot possiblyknow the answer to the question since she has only just arrived. In a way, thiscould be paraphrased as \u201cWhy are you asking where my brother is, if I have onlyjust arrived\u201d.8This is under the assumption that B accepts p.95In the next few sections, we explore whether DM si can in fact react to any typeof felicity condition (propositional content, preparatory, sincerity, and essentialconditions), and whether it can react to any type of speech act.3.3.1 Si-clauses reject infelicitous assertionsWe start with assertions; their felicity conditions are listed in (85):(85) Felicity conditions for assertions (Searle, 1969)a. PROPOSITIONAL CONTENT: Any proposition p.b. PREPARATORY:(i) S has evidence for p.(ii) It is not obvious to both S and H that H knows p.c. SINCERITY: S believes p.d. ESSENTIAL: Counts as an undertaking to the effect that p representsthe actual state of affairs.Reacting to the violation of the propositional content rule of assertions is rel-atively hard, since it stipulates that the semantic meaning of the utterance shouldbe any proposition p. I will assume that there cannot be any real violation of thiscondition, since essentially any proposition would satisfy this condition.In (86), we see that a si-clause can reject a failed preparatory condition (85bi):B asserts that John is Canadian, but there is no way for B to know this since theyhave never talked with John. B having evidence for the truth of what they areasserting is a preparatory condition that A thinks is not being met, and A uses DMsi when she points this out:(86) Epistemic misalignment: violation of preparatory condition for assertion(Condition: B has evidence for p)96Context: A, B are in a group of international students, and someone has justintroduced John by name to the group. While talking to another member ofthe group, B says:B: John es canadiense.\u2018John is Canadian.\u2019A: Sisinuncaneverhashave.2SGhabladotalkedconwithe\u00b4l!him\u2018[Si] you\u2019ve never talked with him!\u2019DM si can also react to the breach of the second preparatory felicity condition,namely that it should not be obvious to both the Speaker and the Addressee thatthe Addressee knows p. In (87), both A and B know John well and know that heis Canadian, so B\u2019s utterance is obvious to A:(87) Epistemic misalignment: violation of preparatory condition for assertion(Condition: It is not obvious to both A and B that A knows p)Context: A, B, and John have been friends for a while. At a party for John,B brings a bunch of Tim Horton\u2019s cups, and B tells A:B: Es que John es canadiense.\u2018It\u2019s just that John is Canadian.\u2019A: Sisiyaalreadyloitse\u00b4!know.PRS.1SG\u2018[Si] I already know!\u2019The sincerity condition of assertions rules that a Speaker who asserts a propo-sition has to believe in its truth. In (88), B\u2019s reply to A shows that A cannot97possibly believe that John is Canadian because he has just seen his US passport.Again, this utterance can be preceded by DM si:(88) Epistemic misalignment: violation of sincerity condition for assertion (Con-dition: B believes p)Context: A, B, and John have been friends for a while, and have gone outto pubs together. Since John looks pretty young, he has been asked to showhis ID many times, and both A and B have seen his US passport. Whiletalking to a mutual friend with A, B says:B: John es canadiense.\u2018John is Canadian\u2019A: Sisiteyouhahasensen\u02dcadoshownsuhispasaportepassportestadounidense!American\u2018[Si] he has shown you his US passport!\u2019The final essential condition stipulates that an assertion should count as pre-senting the actual state of affairs: in (89), we see that a si-clause can be used toreject an assertion that does not count as a representation of the actual state ofaffairs\u2013in this example, B is not giving an accurate representation of the context,according to A:(89) Epistemic misalignment: violation of essential condition for assertion (Con-dition: p does not represent the actual state of affairs)Context: A and B are waiting at the bus stop after a failed picnic.98B: Hace un d\u0131\u00b4a magn\u0131\u00b4fico.\u2018It\u2019s a gorgeous day.\u2019A: SisinoNEGhahasparadostoppeddeofllover!raining\u2018[Si] it hasn\u2019t stopped raining!\u2019We have just seen that a si-clause can be used to mark the violation of propo-sitional content, preparatory, sincerity, and essential conditions. But how can wemodel this step by step in our Dialogue Board? I propose that felicity conditionsare stored in the CG as propositions, and that B\u2019s discourse move introduces aproposition to the Dialogue Board that conflicts with one of them. The schemafor the example (87) is in (90):(90) PHASE 0: ALIGNMENTp is in CG (p=\u2018A knows that John is Canadian\u2019.)PHASE 1: MISALIGNMENTB makes move \u00b5 (\u00b5=\u2018John is Canadian.\u2019)\u00b5 is subject to a felicity condition q (q=\u2018A doesn\u2019t know that J is Canadian\u2019)q pragmatically contradicts pPHASE 2: NEGOTIATIONA utters si-p (p=\u2018A knows that John is Canadian.\u2019)PHASE 3: ALIGNMENTp is in CG (p=\u2018A knows that John is Canadian.\u2019)Given the context in (87), it is fair to assume that it is obvious that A knowsJohn\u2019s nationality. More importantly, given the context it should be obvious to B99that this is the case. Therefore, p=\u2018A knows that John is Canadian\u2019 is in the CG.This state of affairs is represented in Figure 3.4:9ACURRENT *DCATDLACG pp=\u2018A knows that J is Canadian.\u2019TDLBCURRENT *DCBTDLBCG pp=\u2018A knows that J is Canadian.\u2019TDLKA0TABLEFigure 3.4: Phase 0 (Alignment): both A and B know p according to A.As soon as B makes move \u00b5 (when B asserts \u2018John is Canadian\u2019), it becomesapparent to A that there is a problem: by asserting the proposition \u2018John is Cana-dian\u2019, B signals that B assumes that A does not know that John is Canadian. Aswe have seen, this is a violation of one of the preparatory conditions of assertions.This condition is formalized in Figure 3.5 as a proposition q in the Addressee\u2019s(B\u2019s) information state. Most importantly, since this is a felicity condition, this isassumed to be shared in the CG. The result is that there is a mismatch between thetwo CGs (the one representing what the Speaker thinks CG looks like, and the onerepresenting what the Speaker thinks that the Addressee\u2019s thinks). In fact, B\u2019s ideaof the CG contains a proposition q that pragmatically contradicts the propositionp in A\u2019s CG.9The projected set is represented here with an asterisk (*) due to space issues.100ACURRENT *DCATDLACG pp=\u2018A knows that J is Canadian\u2019TDLBCURRENT *DCBTDLBCG qq=\u2018A doesn\u2019t know J is Canadian\u2019TDLKA1TABLE\u00b5(\u00b5=ASSERT(\u2018John is Canadian.\u2019))Figure 3.5: Phase 1 (Misalignment).Note that this points at the need to distinguish between the CG as conceivedby the Speaker, and the CG that the Speaker thinks the Addressee has in mind.Since there is a mismatch between the two versions of the CG, A uses DM si: inFigure 3.7, she re-introduces into the conversation something that is already in theCG, namely p. We therefore find si-p on the Table, as well as the previous move\u00b5 since it has not been resolved.ACURRENT *DCATDLACG pp=\u2018A knows that J is Canadian\u2019TDLBCURRENT *DCBTDLBCG qq=\u2018A doesn\u2019t know J is Canadian\u2019TDLKA2TABLEsi-p;\u00b5Figure 3.6: Phase 2 (Negotiation): A uses a si-clause.101The desired outcome is to achieve an aligned epistemic configuration like theone in Figure 3.4, repeated in Figure 3.7. The Table is empty now, since all the is-sues have been resolved. B has revised their CG, and p is again in the CG\u2013withoutany conflicting proposition.ACURRENT *DCATDLACG pp=\u2018A knows that J is Canadian\u2019TDLBCURRENT *DCBTDLBCG pp=\u2018A knows J is Canadian\u2019TDLKA2TABLEFigure 3.7: Phase 3 (Alignment): both A and B know p.Figure 3.7 represents a symmetrical, harmonious conversation: both interlocu-tors are aligned.3.3.2 Si-utterances reject infelicitous commandsMoving on to commands, Searle (1969) identifies the following conditions for thistype of speech act:(91) Felicity conditions for commands (Searle, 1969)a. PROPOSITIONAL CONTENT: Future act.b. PREPARATORY:(i) H is able to do the action.(ii) It is not obvious to S and H that H will do the action in the normalcourse of events.c. SINCERITY: S wants H to do the action.102d. ESSENTIAL: Counts as an attempt to get H to do the action.The first preparatory condition (91bi) is that the Addressee has to be able to dothe action that the Speaker wants them to do. In example (92), this is not the case:B is ordering A to translate a bossanova song (which are typically in Portuguese),and B uses a si-marked clause to reply that she cannot comply with the commandsince she does not speak Portuguese:(92) Epistemic misalignment: violation of preparatory condition for commands(Condition: A is able to do the action)B: Tradu\u00b4ceme esta cancio\u00b4n de bossanova.\u2018Translate for me this bossanova song.\u2019A: SisinoNEGhablospeak.PRS.1SGportugue\u00b4s.Portuguese\u2018[Si] I don\u2019t speak Portuguese.\u2019 (modified from Egg (2013))A\u2019s reply is not correcting any factual information in B\u2019s utterance, but isnonetheless repairing the previous speech act by marking that one of the felic-ity conditions is not met.In (93) we see that A can in fact reject different felicity conditions of thesame utterance. In this case, A is marking that the second preparatory conditionis violated (by which the interlocutors don\u2019t know whether the Addressee willperform the action in the \u201cnormal course of events\u201d):(93) Epistemic misalignment: violation of preparatory condition of commands(Condition: It is ot obvious to B that A will perform the action)103B: Tradu\u00b4ceme esta cancio\u00b4n de bossanova.\u2018Translate for me this bossanova song.\u2019A: Sisiyaalreadyloitibago.IMPER.1SGatohacerdoestathistarde.afternoon\u2018[Si] I was going to do it this afternoon.\u2019In (93) A\u2019s utterance implicates that she has time allotted for the task, andtherefore it is obvious that she will do the action she is required to do\u2013translatingthe bossanova song. Therefore, B\u2019s command is not felicitous: it does not fulfillthe felicity condition in (91bii).The sincerity rule for commands stipulates that the Speaker wants the Ad-dressee to perform the action conveyed by the command. In (94), A says that theaction she is being ordered to do\u2013again, translating a song\u2013is not something thatB really wants her to do:(94) Epistemic misalignment: violation of sincerity condition of commands (Con-dition: B wants A to perform the action)B: Tradu\u00b4ceme esta cancio\u00b4n de bossanova.\u2018Translate for me this bossanova song.\u2019A: Sisiodiashate.PRS.2SGlathebossanova.bossanova\u2018[Si] you hate bossanova.\u2019Let us see in more detail how this works with the example in (92). The overallschema of the conversation is the following:(95) PHASE 0: ALIGNMENTp is in CG (p=\u2018A doesn\u2019t speak Portuguese.\u2019)104PHASE 1: MISALIGNMENTB makes move \u00b5 (\u00b5=\u2018COMMAND (Translate this bossanova song).\u2019)\u00b5 introduces a felicity condition q (q=\u2018A speaks Portuguese\u2019)q contradicts pPHASE 2: NEGOTIATIONA utters si-p (p=\u2018A doesn\u2019t speak Portuguese.\u2019)PHASE 3: ALIGNMENTp is in CG (p=\u2018A does not speak Portuguese.\u2019)The context in (92) specifies that both A and B know each other. This meansthat A can make estimates about what B already knows more readily than if theydid not know each other.10 In this case, A and B are both aware of the fact that Adoes not speak Portuguese, since they know each other well. This is illustrated infigure 3.8, where p=\u2018A does not speak Portuguese\u2019.ACURRENT *DCATDLACG pp=\u2018A doesn\u2019t speak Port.\u2019TDLBCURRENT *DCBTDLBCG pp=\u2018A doesn\u2019t speak Port.\u2019TDLKA0TABLEFigure 3.8: Phase 0 (Alignment): Both interlocutors know that A does notspeak Portuguese.When B utters the command Tradu\u00b4ceme esta cancio\u00b4n de bossanova (\u2018Trans-10This does not mean that it is true that A knows everything B thinks.105late this bossanova song for me.\u2019), they are assuming that A can perform the ac-tion: this is one of the felicity conditions of commands. In this case, being ableto perform the action denoted by the command means that A must be able tospeak Portuguese. This condition is again represented by q (q=\u2018A can speak Por-tuguese\u2019) in Figure 3.9, where q is in B\u2019s version of the CG.ACURRENT *DCATDLACG pp=\u2018A doesn\u2019t speak Port.\u2019TDLBCURRENT *DCBTDLBCG qq=\u2018A speaks Port.\u2019TDLKA1TABLE\u00b5(\u00b5=COMMAND (Translate this bossanova song))Figure 3.9: Phase 1 (Misalignment): It becomes apparent that A and B donot share the same CG.Figure 3.10 shows the part of the exchange when A uses the si-clause, markedhere again as si-p. As the model shows, DM si is re-introducing the propositionp, which was already in the CG (in both representations in Phase 0, only in theSpeaker\u2019s representation of the CG in Phase 1). The Table is pretty busy: si-p isthere as it was just introduced by A (hence the boldfaced A), and \u00b5 (\u00b5=Tradu\u00b4cemeesta cancio\u00b4n de bossanova. \u2018Translate this bossanova song for me.\u2019) is still on theTable.1111Note that I abstract away from the update proposed by the command itself, which would beto add an outcome to the TDL of the Addressee.106ACURRENT *DCATDLACG pp=\u2018A doesn\u2019t speak Port.\u2019TDLBCURRENT *DCBTDLBCG qq=\u2018A speaks Port.\u2019TDLKA2TABLEsi-p;\u00b5Figure 3.10: Phase 2: A utters a si-clause.The desired outcome is to return to an aligned epistemic state\u2013to go back toPhase 0. This is captured in Figure 3.11: everything has been removed from theTable, and B has revised their knowledge (and CG) to the previous state whereonly p holds.ACURRENT *DCATDLACG pp=\u2018A doesn\u2019t speak Port.\u2019TDLBCURRENT *DCBTDLBCG pp=\u2018A doesn\u2019t speak Port.\u2019TDLKA3 = KA0TABLEFigure 3.11: Phase 3: Both interlocutors know that A does not speakPortuguese.1073.3.3 Si-clauses reject infelicitous questionsNow we turn to questions:(96) Felicity conditions for questions (Searle, 1969):a. PROPOSITIONAL CONTENT: Any proposition or propositional func-tion.b. PREPARATORY:(i) S does not know the answer.(ii) It is not obvious to both S and H that H will provide the informationat that time without being asked.c. SINCERITY: S wants this information.d. ESSENTIAL: Counts as an attempt to elicit this information from H.The first preparatory condition for the felicity of a question is that the Speakerdoes not know the answer to the question. This condition is violated in (98). Inthe context, all the friends have decided who the designated driver is, and thedesignated driver will not drink during the party. B, therefore, should know thatthe answer to their question is no.12(97) Epistemic misalignment: violation of preparatory condition for questions(Condition: B does not know the answer)Context: A group of friends are partying, and they have nominated A as thedesignated driver. During the party, B approaches A and asks:12Searle (1969) himself notes that there are certain questions that do not meet this conditionsuch as exam questions. I refer here only to non-exam, non-rhetorical questions (which wouldalso not meet the sincerity condition).108B: Quieres una copa?\u2018Do you want a drink?\u2019A: Sisisoyameltheconductor.driver\u2018[Si] I am the designated driver.\u2019The second preparatory condition says that it shouldn\u2019t be obvious to bothinterlocutors that the Addressee will provide information without being asked.(98) Epistemic misalignment: violation of preparatory condition for questions(Condition: it is not obvious to both A and B that A will provide the infor-mation at that time without being asked)Context: A group of friends are partying, and they have nominated A asthe designated driver. During the party, a bartender is going around askingpeople what they want to drink. When they get to A, B asks:B: Quieres una copa?\u2018Do you want a drink?\u2019A: Sisisoyameltheconductor.driver\u2018[Si] I am the designated driver.\u2019The sincerity condition for questions stipulates that the Speaker wants to knowthe truth about what she is asking. Again, this is not the case in (99): A denies thefact that B wants to know the answer to the question they are asking, and uses asi-clause to do so:109(99) Epistemic misalignment: violation of sincerity condition for questions (Con-dition: B wants this information)Context: A has been dieting for a month and complaining to A about it. Heis eating a chocolate bar when he turns to B and says:B: Cua\u00b4ntas calor\u0131\u00b4as tiene esta chocolatina?\u201cHow many calories are in this chocolate bar?\u201dA: SisieninrealidadrealitynoNEGloitquiereswant.PRS.2SGsaber.know\u2018[Si] you don\u2019t really want to know.\u2019Finally, the essential condition of questions is that they count as an attemptto elicit information from the Addressee. This rule is violated in (100): whensomeone asks a question, they believe that the Addressee is able to supply theinformation needed. But in (100) A cannot possibly know where her brother isbecause she just got home. In a sort of protest, she uses a si-clause telling B of theimpossibility of her knowing the whereabouts of her brother:(100) Epistemic misalignment: violation of essential condition of questions (Con-dition: Counts as an attempt to elicit this information from the Addressee.)Context: A has just gotten home from work. Her father (B) asks:B: \u00bfDo\u00b4nde esta\u00b4 tu hermano?\u2018Where is your brother?\u2019A: Sisiacabofinish.PRS.1SGdePREPllegar.arrive\u2018[si] I just arrived.\u2019 (Rodr\u0131\u00b4guez Ramalle, 2011)110Let us see what the exchange in (98) would look like:(101) PHASE 0: ALIGNMENTp is in CG (p=\u2018B knows that A is the designated driver.\u2019)PHASE 1: MISALIGNMENTB makes move \u00b5 (\u00b5=\u2018Do you want a drink?\u2019)\u00b5 is subject to a felicity condition q (q=\u2018B doesn\u2019t know if A wants a drink.\u2019)q contradicts pPHASE 2: NEGOTIATIONA utters si-p (p=\u2018A is the designated driver.\u2019)PHASE 3: ALIGNMENTp is in CG (p=\u2018A is the designated driver.\u2019)ACURRENT *DCATDLACG pp=\u2018A is the designated driver\u2019.TDLBCURRENT *DCBTDLBCG pp=\u2018A is the designated driver.\u2019TDLKA0TABLEFigure 3.12: Phase 0 (Alignment): A thinks that both interlocutors knowthat A is the designated driver.111ACURRENT *DCATDLACG pp=\u2018A is the designated driver.\u2019TDLBCURRENT *DCBTDLBCG qq=\u2018B doesn\u2019t know if A wants a drink.\u2019TDLKA1TABLE\u00b5(\u00b5=Quieres una copa?)\u2018Do you want a drink?\u2019Figure 3.13: Phase 1: It becomes apparent that A and B do not share thesame CG.ADCTDLOGCG p(p=\u2018A is the designated driver.\u2019)BDCTDLOGCG q(q=\u2018B doesn\u2019t know whether A wants a drink.\u2019)KA2TABLEsi-p;\u00b5Figure 3.14: Phase 2: A utters a si-clause.The final Phase 3 is a return to epistemic alignment: assuming that B doesnot object to A\u2019s si-clause, the conversation would return to a state where bothinterlocutors believe in p (p=\u2018A is the designated driver.\u2019), and where B\u2019s utterancehas had no effect on the CG.112ACURRENT *DCTDLACG pp=A is the designated driverTDLBCURRENT *DCTDLBCG pp=A is the designated driverTDLKA3=KA0TABLEFigure 3.15: Phase 3: A thinks that both interlocutors know that A is thedesignated drived.3.3.4 Si-clauses reject infelicitous promisesWe turn now to promises, a type of commissives. Commissives have a long list ofconditions (listed in (102)):(102) Conditions for commissives (Searle, 1969)a. PROPOSITIONAL CONTENT: Future act.b. PREPARATORY:(i) H would prefer S\u2019s doing A to their not doing A.(ii) S believes H would prefer S\u2019s doing A to not doing A.(iii) It is not obvious to both S and H that S will do A in the normalcourse of events.c. SINCERITY: S intends to do the action.d. ESSENTIAL: Counts as an undertaking to do the action.The propositional content rule of commissives says that a promise needs tocommit to an act to be performed in the future. In (103), B is promising to do113something that has already taken place in the past, and A brings this up using DMsi:(103) Epistemic misalignment: violation of propositional content condition ofcommissives (Condition: the promised action should be a future act)B: Te prometo que te llevo al estreno de Wonder Woman.\u2018I promise that I will take you the premiere of Wonder Woman\u2019A: Sisilaitestrenaronrelease.PST.3PLayer.yesterday\u2018[Si] the premiere was yesterday.\u2019There are three preparatory conditions for promises, the first one being thatthe person \u201creceiving\u201d the promise should prefer that the promised action takesplace to it not taking place. This is not the case in (104), where B is promising Asomething that would not be for their benefit, namely taking them skiing, whichthey do not enjoy.(104) Epistemic misalignment: violation of preparatory condition of commis-sives (Condition: A would prefer B\u2019s doing the action to not doing theaction)Context: B is talking to A, who hates the snow and all snow-related sports.B: Te prometo que man\u02dcana te llevo a esquiar.\u2018I promise that I will take you skiing tomorrow.\u2019A: Sisiatom\u0131\u00b4menoNEGmemegustaplease.PRS.3SGesquiar.ski\u2018[Si] I don\u2019t like skiing.\u2019114The second preparatory condition is similar to the first one: in this case, theSpeaker (the one making the promise) has to believe that the first preparatorycondition holds. This is not the case in (105), which uses the same context andsame promise as (104): A can react with DM si to the B\u2019s lack of awareness ofthe promised act not being preferred by A:(105) Epistemic misalignment: violation of preparatory condition of commis-sives (Condition: B believes A would prefer B\u2019s doing the action to notdoing the action)Context: B is talking to A, who hates the snow and all snow-related sports.B: Te prometo que man\u02dcana te llevo a esquiar.\u2018I promise that I will take you skiing tomorrow.\u2019A: Sisisabesknow.2SGquethatatom\u0131\u00b4menoNEGmemegustaplease.PRS.3SGesquiar.ski\u2018[Si] you know that I don\u2019t like skiing.\u2019The third preparatory condition rules that it should not be obvious to both in-terlocutors that the Speaker will perform the promised action in the normal courseof events. In (106) we see that a si-clause can react to the violation of this rule,since it is not obvious to A that B will be able to perform the action in the normalcourse of events:13(106) Epistemic misalignment: violation of preparatory condition of commis-sives (Condition: It is not obvious to both A and B that B will do theaction in the normal course of events)13I thank Lisa Matthewson (p.c.) for providing this context.115B has brought A a cup of tea to every morning for the past 10 years. Onemorning, B says:B: Te prometo que te traigo una taza de te\u00b4.\u2018I promise that I will bring you a cup of tea.\u2019A: Sisiloithacesdo.PRS.2SGsiempre.always\u2018[Si] you always do.\u2019The sincerity condition states that the speaker who is making a promise needsto intend to do the promised action. This is what (107) shows: given the context,it is probably not true that B will carry out the promise (or, at least, it is not clearthat they actually intend to do so), and A reacts to this breach of sincerity using asi-clause:(107) Epistemic misalignment: violation of sincerity condition of commissives(Condition: B intends to do the action)Context: B is passionate about animal welfare and animal shelters. Theirmother believes that this is counterproductive for them because they aretoo sensitive, so B promises that they will stop checking shelter websites:B: Te prometo que ya no miro ma\u00b4s protectoras.\u2018I promise that I won\u2019t check out more shelter websites.\u2019A: Sisise\u00b4know.1SGquethatnoNEGloitdicessay.2SGeninserio.seriousness\u2018[Si] I know you don\u2019t mean it.\u2019Finally, the essential condition states that the utterance needs to count as anundertaking to do the promised action. In (108) we see that a si-clause can also116react to a violation of this rule: in this example, the person making the promise ismaking a gesture that \u201ccancels\u201d the sincerity of their promise:(108) Epistemic misalignment: violation of essential condition of commissive(Condition: Counts as an undertaking to do the action)Context: B is promising A to tell the whole truth. However, A can see thatthey have their fingers crossed (a sign that they intend to lie):B: Te lo prometo.\u2018I promise.\u2019A: Sisitieneshave.2SGlosthededosfingercruzados.crossed\u2018[Si] your fingers are crossed.\u2019Si-clauses are able to react to the propositional, preparatory, sincerity, andessential conditions of promises. The schema for (104) is given in (109):(109) PHASE 0: ALIGNMENTp is in CG (p=\u2018A doesn\u2019t like skiing.\u2019)PHASE 1: MISALIGNMENTB makes move \u00b5 (\u00b5=\u2018PROMISE (take A skiing).\u2019)\u00b5 introduces a felicity condition q (q=\u2018Taking A skiing benefits A.\u2019)q contradicts pPHASE 2: NEGOTIATIONA utters si-p (p=\u2018A doesn\u2019t like skiing.\u2019)PHASE 3: ALIGNMENTp is in CG (p=\u2018A does not like skiing.\u2019)1173.3.5 Si-clauses reject infelicitous thank-yousNext we turn to a type of expressive speech act, namely thanking. In Chapter 1we saw that expressive speech acts are, according to Searle (1969), speech actsthat showcase the mental state or feelings of the Speaker. It should be noted, how-ever, that one can say thank you without truly feeling grateful, and in fact the termexpressive is also used to refer to utterances like \u201couch\u201d that do express the feel-ings of the Speaker (Kaplan, 1999; Kratzer, 1999; Potts, 2007); acts like thankingwould be more of a social act than an expressive act per se. In fact, Searle men-tions that for thanking speech acts, the sincerity rule (which stipulates the mentalstate of the Speaker) and the essential rule (which stipulates that the utteranceshould count for all interlocutors as an actual performance of the intended act)overlap:(110) Conditions for thanking speech acts (Searle, 1969)a. PROPOSITIONAL CONTENT: Past act A done by H.b. PREPARATORY: A benefits S and S believes A benefits S.c. SINCERITY: S feels grateful or appreciative for A.d. ESSENTIAL: Counts as an expression of gratitude or appreciation.Si-clauses can be used to react to a violation of the propositional content con-dition in (110): in (111), B is thanking A for an invitation to a party, but the partyhas not even been planned yet and it is not certain that it will take place. This isin conflict with the first felicity condition listed above in (110), that states that theaction that is being thanked for has to be in the past. A uses DM si to react to this:(111) Epistemic misalignment: violation of propositional content condition ofthanking (Condition: The thanked action must be in the past)118B: Muchas gracias por la invitacio\u00b4n!\u2018Thank you so much for the invitation!\u2019A: Sisiau\u00b4nyetnoNEGsabemosknow.PRS.1PLsiifvamosgo.PRS.1PLatohacermakelathefiesta!party\u2018[Si] we don\u2019t know yet whether we\u2019re throwing the party!\u2019The preparatory condition for thanking speech acts states that the action thatis being thanked for has to have been beneficial for the Speaker. In (112), A hasgiven B a recommendation for a job position that gives B more work in exchangefor a little more pay. B thanks A, but A points out that the new job would notnecessarily be beneficial for B given the workload. A can bring this up using DMsi:(112) Epistemic misalignment: violation of preparatory condition of thanking(Condition: Action benefits B and B believes the action benefits B)Context: A has given B a recommendation for a job position that entailsmore workload for a little more pay.B: Muchas gracias por la recomendacio\u00b4n!\u2018Thank you so much for the recommendation!\u2019A: Sisiahoranowtendra\u00b4shave.FUT.2SGma\u00b4smoretrabajo!work\u2018[Si] now you\u2019ll have more work!\u2019As I mentioned at the beginning of this section, the sincerity and the essentialcondition overlap in this case. Example (113) shows that si-clauses can be used toreact to these conditions: B thanks C for flowers that B doesn\u2019t really like, and A119knows this. Out of C\u2019s earshot, A points out that B cannot really be thankful sincethey hate this particular type of flower:(113) Epistemic misalignment: violation of sincerity\/essential condition of thank-ing (Condition: B feels grateful or appreciative for the action)B: Muchas gracias por las flores!\u2018Thank you so much for the flowers!\u2019A: Sisiodiashate.PRS.2SGlastherosasrosesrojas!red\u2018[Si] you hate red roses!\u20193.3.6 Si-utterances reject infelicitous greetingsAnother example of how si-marked clauses can flag an infelicitous speech actis their use in canonical adjacency pairs such as greeting-greeting pairs. This isshown in (114), where B can either reply with the same greeting (\u2018Buenos d\u0131\u00b4as\u2019)or with a si-marked clause that marks that the previous greeting is infelicitoussince it is not the appropriate time of day to use it:(114) Epistemic misalignment: A does not believe that the greeting uttered byB is time-appropriate.B: Buenos d\u0131\u00b4as!\u2018Good day!\u2019 (typically said in the morning)A: Sisiyaalreadysonarelasthecincofivedeoflathetarde!evening\u2018[Si] it\u2019s already five in the evening!\u2019120The same analysis would explain the use of DM si in (114): A assumes thatthere is a publicly shared proposition p in the CG, namely the felicity conditionthat the greeting needs to be appropriate for the time of day. B then says Buenosd\u0131\u00b4as (\u2018Good day\u2019), a greeting that tends to be used in the morning and not in theafternoon, when this exchange is happening. A then uses a si-marked utterance toflag that the previous speech act is infelicitous, and reminds B of the time it is.3.4 How this analysis explains other observationsThe analysis presented here for the licensing of DM si revolved around the ideathat it is the violation of a felicity condition that licenses the use of the DM. How-ever, in Chapter 2 we saw that si-clauses can also reject implicatures, inferencesdrawn from the immediate context (sometimes a gesture), and previously held be-liefs by the Speaker. The analysis presented can, in fact, also account for theseother observations.3.4.1 Si-utterances reject failed implicatures andpresuppositionsIn order to give a unified account of the role of si-clauses, I assume that there isa felicity condition that states all implicatures and presuppositions contained in asentence have to be true, defined in (115) and (116).(115) PRESUPPOSITION CONDITION: All presuppositions triggered by the ut-terance of a sentence s have to be true at time of utterance.(116) IMPLICATURE CONDITION: All implicatures triggered by the utterance ofa sentence s have to be true at time of utterance.With these added felicity conditions, we can use the same analysis for sen-tences like (117), where B triggers an implicature with the use of the particlehasta \u2018even\u2019: even Adriana, the least knowledgeable, knows about the topic they121are discussing.14 A then rejects B\u2019s contribution given that it triggers an implica-ture (Adriana is the least knowledgeable) that is not true. A then reminds B thatAdriana wrote a book about the topic, which contradicts the implicature carriedby B\u2019s sentence:(117) B: Eso hasta Adriana lo sabe.\u2018Even Adriana knows that.\u2019A: Sisihahasescritowrittenunalibrobooksobreaboutello!it\u2018[si] she wrote a book about it.\u2019(118) PHASE 0: ALIGNMENTp is in CG (p=\u2018Adriana wrote a book about x.\u2019)PHASE 1: MISALIGNMENTB makes move \u00b5 (\u00b5=\u2018ASSERT (Even Adriana knows x).\u2019)\u00b5 is subject to a felicity condition q (q=\u2018All implicatures in \u00b5 are true.\u2019)\u00b5 triggers an implicature \u03b9 (\u03b9=\u2018Adriana doesn\u2019t know much about x.\u2019)\u03b9 pragmatically contradicts pPHASE 2: NEGOTIATIONA utters si-p (p=\u2018Adriana wrote a book about x.\u2019)PHASE 3: ALIGNMENTp is in CG (p=\u2018Adriana wrote a book about x\u2019.)The definition in (115) covers presuppositions because, as (119) illustrates,DM si can also be used to flag an unsuccessful presupposition. In (119), B\u2019s14Under certain analyses, this type of implicature (conventional) is closer to presuppositionsthan to a conversational implicature. The same mechanics would apply to an example of a differenttype of implicature, however.122utterance presupposes that Pedro has a brother, but in this example this presup-position cannot be accommodated: Pedro is an only child. This is something thatA thought it was in the CG (since they are at a party, it might be safe to assumethat the party guests know the host), but B\u2019s utterance signals that the proposition\u2018Pedro is an only child\u2019 is not in B\u2019s representation of the CG.(119) Context: A and B are at a mutual friend\u2019s party, Pedro, who is an onlychild. They are discussing Pedro with another guest, and B says:B: El hermano de Pedro es arquitecto.\u2018Pedro\u2019s brother is an architect.\u2019A: SisiPedroPedroesishijosonu\u00b4nico!unique\u2018[Si] Pedro is an only child!\u2019We can use the same analysis and the general felicity condition introduced in(115) to explain this, as is schematized in (120):(120) PHASE 0: ALIGNMENTp is in CG (p=\u2018Pedro is an only child.\u2019)PHASE 1: MISALIGNMENTB makes move \u00b5 (\u00b5=\u2018ASSERT (P\u2019s brother is an architect).\u2019)\u00b5 is subject to a felicity condition q (q=\u2018All presuppositions in \u00b5 are true.\u2019)\u00b5 triggers a presupposition \u03b9 (\u03b9=\u2018Pedro has a brother.\u2019)\u03b9 pragmatically contradicts pPHASE 2: NEGOTIATIONA utters si-p (p=\u2018Pedro is an only child\u2019.)123PHASE 3: ALIGNMENTp is in CG (p=\u2018Pedro is an only child.\u2019)3.4.2 Si-clauses do not need to convey contrastIn Chapter 2 I mentioned that although many accounts of DM si and si-clausesreferred to contrast and adversativity, there are examples like (121) where thereseems to be no apparent contradiction or opposition with the truth-conditionalcontent of B\u2019s utterance. Under the analysis presented, examples like this wouldexplained: B is violating one of the preparatory conditions of assertions, namelythat it shouldn\u2019t be obvious to both Speaker and Addressee that the Addresseeknows the information conveyed by the proposition. In this case, one can arguethat A is acting as if they already knew the information, or as if it was not surpris-ing to them to hear the \u201cnews\u201d:(121) B: A Juana la han aceptado en Stanford.\u2018Juana got accepted into Stanford.\u2019A: Claro,of.coursesisiesismuyveryinteligente.intelligent\u2018Of course, [si] she\u2019s very intelligent.\u2019 (Schwenter, 2016b, p. 26; 10)Another example of this can be seen in the following corpus example:(122) B: Por Navidad regalaron: un Nenuco baby\u2018For Christmas they gave her: a Nenuco baby (perfume for children)\u2019A: S\u0131\u00b4,yessisiyaalreadymemeloithashascontado,told,ca\u00b4llate,shut.up[. . . ]\u2018Yes, [si] you\u2019ve already told me, shut up.\u2019 (from CREA Corpus (CREA,2020))124In (122), A literally says that she already knew the information in the previousmove: it does not bring new information, hence the update proposed by B is notsuccessful.3.4.3 Si-clauses reject defective inferences from contextSi-clauses don\u2019t need a verbal antecedent: a Speaker can use the interlocutors\u2019actions to infer what they know.(123) Context: A, B, and C have been friends for a while. During a BBQ, Bhands C a beef burger. A reacts:SiSIesisvegetariana!vegetarian\u2018[Si] she is vegetarian!\u2019In this case, a felicity condition of the action in (123) is that C eats meat.This contradicts the proposition p=\u2018C is vegetarian\u2019 that A thought was in the CG(given that they have been friends for a while). This is schematized in (124):(124) PHASE 0: ALIGNMENTp is in CG (p=\u2018C is vegetarian.\u2019)PHASE 1: MISALIGNMENTB makes move \u00b5 (\u00b5=B hands beef burger to C.)\u00b5 is subject to a felicity condition q (q=\u2018All implicatures in \u00b5 are true.\u2019)\u00b5 implies \u03b9 (\u03b9=\u2018C eats meat.\u2019)q contradicts pPHASE 2: NEGOTIATIONA utters si-p (p=\u2018C is vegetarian.\u2019)PHASE 3: ALIGNMENTp is in CG (p=\u2018C is vegetarian.\u2019)1253.4.4 Si-clauses reject previously held beliefs by SpeakerA challenge arises when we look at examples where there is no interlocutor as in(125):(125) Context: A had heard on the weather forecast that it would rain all week-end long. When she wakes up, it is sunny.A: Sisihacedoessol!sun\u2018[Si] it\u2019s sunny!\u2019This context of use is, however, still in line with the proposed meaning of DMsi. There is still rejection, but in this case of the previous information state of theSpeaker, instead of the Addressee\u2019s version of CG. DM si still introduces a propo-sition that is uncontroversial. The proposal that the contradiction is between thecurrent context and a past version of it is not completely new. A similar analysisis given by Torres Bustamante (2013) for Andean Spanish mirative constructionscontaining the pluperfect; she mentions that the surprise arises \u201cas a consequenceof the clash between the Speaker\u2019s previous beliefs and the current state of affairswhich is discovered at speech time.\u201d (Torres Bustamante, 2013, p. 39). So (125)still conveys the uncontroversiality ascribed to si, but the proposition it contradictsis a belief previously held by the Speaker before the speech time.We can apply the same analysis to what has been called \u2018causative\u2019 si, asexemplified in (126), where the sentence introduced by si conveys the reason whythe previous utterance was made by the same interlocutor.126(126) Co\u00b4mohownoNEGvasgo.PRS.2SGatosaberknowbailar?!dance.INFSisiloitllevastakeeninlathesangre!blood\u2018How can you not know how to dance?! [Si] it\u2019s in your blood!\u2019 (from acomic by Moderna de Pueblo (@modernadepueblo), published Dec. 14,2019 on Instagram)We can still apply the same analysis proposed to dialogues here: the si-markedclause still rejects the previous move due to it not being felicitous, but in this casethe previous move was introduced by the same interlocutor. In the example (126)above, the Speaker highlights how absurd their question is with the propositionintroduced by DM si.3.5 SummaryIn this chapter I have analyzed DM si as a marker that indicates that a propositionis already in the CG\u2013in this way, it expresses superassertive force. We saw thatthis explains the contexts in which we find the particle: it cannot be used to con-vey \u201chot news\u201d, or in case the Speaker knows for sure that the Addressee cannotalready know the information presented.The contribution of the whole si-clause to the dialogue is to reject or objectto the conversational update proposed by the interlocutor on the grounds that thatcontribution violates a felicity condition of the speech act it intends to perform.This shows that interlocutors keep score of the Addressee\u2019s information state, andthat resolving an epistemic misalignment is one way in which language users dealwith misaligned information states.In the next chapter, we turn to the DM no. I will argue that, rather than resolv-ing epistemic misalignment, DM no seeks on the contrary to maintain epistemicalignment with the Addressee, and does so by soliciting the Addressee\u2019s endorse-ment to proceed to the next step of the conversational exchange.127Chapter 4Monitoring the Speaker-Addresseerelation: Peninsular Spanish noLet us make a special effort to stop communicating with each other, sowe can have some conversation. \u2014 Mark TwainDM no has been analyzed in the earliest studies that focused on this DM as amarker that manages interaction (Ortega, 1985). In particular, San Mart\u0131\u00b4n Nu\u00b4n\u02dcez(2011) categorizes DM no as an \u2018interrogative control of contact marker\u2019. One ofthe effects of this interactional side of no is its use as a mitigating device, whichhas been the focus of much work (Ortega, 1985; Garc\u0131\u00b4a Vizca\u0131\u00b4no, 2005; Landone,2009; Rodr\u0131\u00b4guez Mun\u02dcoz, 2009; Montan\u02dcez Mesas, 2015; Ucle\u00b4s Ramada, 2018,2020). But we should distinguish the core meaning of the DM from the discourseeffect of the no-tagged clause has on a particular context of use. The first ingre-dient of my analysis will be the core function of DM no, which is to request theAddressee\u2019s confirmation of the conversational update proposed by the anchor.This update involves sharing a public commitment; that is, the Speaker is request-ing that the Addressee confirm the validity of making an individual commitmenta shared commitment. Different speech act types propose different conversationalupdates, and it is this difference, and not a difference in sentence type, that allows128us to capture the distribution of DM no. The second ingredient of the analysisrefers to why we may want to monitor the Speaker-Addressee relation: I showthat we use no-tagged clauses because we want to confirm a bias, be it towards abelief or towards an outcome.4.1 Ingredients of the analysisI propose that DM no requests confirmation from the Addressee that the contextualupdate proposed by the anchor is publicly shared. In Chapter 2 we have seen twopieces of evidence to support this proposal: (i) the distribution of the DM withrespect to only a subtype of interrogative sentences (rhetorical questions) and asubtype of commands (weak commands or suggestions) hints at the fact that theDM is sensitive to the type of contextual update proposed; and (ii) commentsfrom speakers that paraphrase the effect of the DM as including the Addressee inotherwise Speaker-oriented speech acts (such as exclamations) hint at the fact thatthe Speaker is trying to call on the Addressee to commit to the contextual update.A rough schema of the contribution of DM no is presented in Figure 4.1. Oneof the key aspects of this analysis is that DM no is requesting confirmation of atransition from one information state to another\u2013from one moment where a com-mitment is private and individualized, to another where a commitment is publicand shared.129Figure 4.1: Schema of the effect of DM no on the dialogue.Although this in-a-nutshell analysis is very broad, it predicts that certain speechact types will not accept DM no: the types of speech act that cannot be the an-chor of this DM are those that do not require the participation of the Addressee inorder to be fully performed, namely commissives, declarations, and expressivesof the thanking and apologizing type. Such speech acts, I argue, require an im-mediate context update by the Speaker in order to be felicitous. A declarationlike \u2018I declare you joined in matrimony\u2019 does not require the participation of theAddressee: a declaration is meant to change a state of affairs in the world as itis spoken by the appropriate Speaker, and therefore if the Speaker were to askfor the Addressee\u2019s cooperation in the acceptance of the speech act the Speakerwould be failing at performing the speech act of marrying two people. Table 4.1summarizes the effects of different speech act types, and their co-occurrence withDM no.130Table 4.1: Types of speech act and their predicted co-occurrence with DMno.Speech Act Type Co-occurrence with DM noASSERTION 3COMMAND (subset) 3RHETORICAL QUESTION 3INFORMATION-SEEKING QUESTION 7EXPRESSIVE 7COMMISSIVE 7DECLARATIONS 7I model this function of DM no on the conversation by proposing that it tablesthe update proposal of its anchor. I formalize this by putting on the Table theSpeaker\u2019s information state KA at the time of utterance of the no-tagged clause,which I will referred to as KAt. A in this case represents Speaker A (since the prag-matic roles switch). 1 This information state is the representation of the Speaker\u2019sunderstanding of the conversation at a given time t. In a sense, A\u2019s informationstate at a particular time is a snapshot of the conversation: the difference betweenKA1 and KA2 will capture the conversational update, that is, it will capture thetransition from KA1 to KA2.Note that this way of modelling the idea of tabelling a conversational updateby tabelling the information state of the Speaker might run into a problem: if theAddressee rejects this tabelled information state, how can we know for sure thatthey are just rejecting the update and not something else in the information state?I tried to avoid this issue by specifying the moment in time that each informationstate represents, so that for an assertion uttered at t2 the only difference from KA1to KA2 would be the addition of a proposition p to the projected CG. Therefore, if1This is in a sense similar to the LATEST MOVE component in (Ginzburg, 2012)): a componentthat stores the very last conversational move made by any interlocutor.131we table KA2, we are in essence tabelling the addition of p to the projected CG.There is another part of the proposed analysis that bears discussing: the use ofDM no is that of a confirmational, that is, it requests the Addressee\u2019s corroborationof a move, not merely to check whether a move is valid. This means that theSpeaker is biased towards the Addressee accepting the transition, in the sameway that biased questions convey an expectation on the part of the Speaker ofwhat the answer to the question is (Ladd, 1981; Van Rooy and S\u02c7afa\u00b4r\u02c7ova\u00b4, 2003;Han, 2002; Reese and Asher, 2007, 2010; Sudo, 2013; Krifka, 2015; Domaneschiet al., 2017). In fact, utterances modified by DM no are usually analyzed as theequivalent (in terms of function) to tag questions (Go\u00b4mez Gonza\u00b4lez, 2014), whichare usually analyzed as biased questions. Whereas a standard polar question asksthe Speaker to choose between p and \u00acp, a biased question expresses that theSpeaker has some previous belief in favour of (or against) p. This is illustrated forEnglish in (127):(127) Context: A and B are good friends, and are catching up at a cafe\u00b4. B saysthat they will order the cheesecake. A knows that B is lactose-intolerant,so she replies:Aren\u2019t you lactose-intolerant?In (127), Speaker A is biased towards the answer being positive (\u2018Yes, I amlactose intolerant\u2019). The literature on biased questions differentiates between twomain types of possible sources of bias: contextual evidence for or against theproposition (contextual bias) and the Speaker\u2019s beliefs about the proposition (epis-temic bias) (Bu\u00a8ring and Gunlogson, 2000; Van Rooy and S\u02c7afa\u00b4r\u02c7ova\u00b4, 2003; Romeroand Han, 2004; Reese and Asher, 2007, 2010; Sudo, 2013).In the same way, when a Speaker uses DM no they are biased towards theAddressee accepting the transition from KA1 to KA2. We saw this in exampleswhere DM no was only allowed in contexts where the Speaker believed that the132Addressee would agree with the anchor, as in (128) (repeated from (71)):(128) 3 Context 1: The Speaker has organized a party the next day, and hasreceived an RSVP from the Addressee. She is confirming the guest list,and asks:7 Context 2: The Speaker has organized a party the next day, and has notreceived an RSVP from the Addressee. She is confirming the guest list,and asks:Vienescome.2PRS.2SGman\u02dcana,tomorrowno?no\u2018You\u2019re coming tomorrow, [no]?\u2019In section 4.4.1 I will discuss how these biases affect the use of DM no, andwhether the use of the DM distinguishes between different types of bias. I willuse the notion of outcome bias: when the Speaker does not express a bias towardsthe truth of a proposition, but towards a preferred action or outcome (Van Rooyand S\u02c7afa\u00b4r\u02c7ova\u00b4, 2003; Reese and Asher, 2007; Yang and Wiltschko, 2016).2 SinceDM no can modify commands, we need this type of bias to explain the contextsof use in which the tag appears with this type of speech act.4.2 How to monitor the S-A relation: whatspeakers can do with DM noIn Chapter 2 we saw that DM no can follow declaratives, interrogatives, imper-atives, and also exclamatives. But we have also gotten a glimpse of certain con-straints on the use of DM no, for instance although it can modify interrogative2This is similar to bouletic bias, which is discussed as a type of epistemic bias. According tothis, there are three modal flavours when it comes to epistemic bias: expectations based on rules(deontic), based on beliefs (epistemic), and based on desires (bouletic) (Reese and Asher, 2007;Sudo, 2013).133clauses, it can only do so if this interrogative is a rhetorical question and not acanonical information-seeking question.The analysis presented predicts that only those speech acts whose contextualupdates include a publicly shareable commitment will be felicitous anchors ofDM no. As we will see in the following sections, this prediction is borne out.In sections 4.2.1 through 4.2.4 I will show that DM no can modify assertions,exclamations, a subset of commands, and rhetorical questions. In section 4.3 Iturn to the types of speech acts that are predicted to not co-occur with DM no,namely information-seeking questions, thanking and apologizing, commissives,and declarations. In section 4.4 I describe the role of bias in the licensing of DMno.4.2.1 DM no negotiates the update proposed by assertionsDM no co-occurs with assertions, both affirmative and negative, as illustrated in(129) and (130) where it is shown that both the positive and negative assertion(No) Esta\u00b4 lloviendo \u2018It (isn\u2019t) raining\u2019 can be the anchor of the DM:(129) Context: A has checked the forecast: it\u2019ll be raining all week where B is.Later, A and B are talking on the phone long-distance. A says:Esta\u00b4islloviendo,rainingno?no\u2018It\u2019s raining, [no]?\u2019(130) Context : A has checked the forecast: After a week of rain, it has stoppedraining where B is. Later, A and B are talking on the phone long-distance.A says:134NoNEGesta\u00b4islloviendo,rainingno?no\u2018It isn\u2019t raining, [no]?\u2019Let us first illustrate the effects of a bare non-evaluative assertion on the Di-alogue Board. For expository purposes, I assume a simplified version of the Di-alogue Board where there are no other elements present (that is, no other con-versational moves have been made). I will be showing three \u201csnapshots\u201d of theinteraction between A and B: KA1, which represents the state of the dialogue justbefore the target utterance (the focus will be on utterances modified by DM no,but I will also show the effects of unmodified (or bare) utterances); KA2, whichrepresents the tabelling of the target utterance and the effect it has on the Dia-logue Board; and KA3, which represents the desired outcome (typically involvinga resolution).The (simplified) context prior to the utterance of a bare assertion is modelledin Figure 4.2:3 A has the proposition p (p=\u2018It is raining.\u2019) in her discourse com-mitments (DCA).4ACURRENT PROJECTEDDCA pp=\u2018It is raining.\u2019TDLACGTDLBCURRENT PROJECTEDDCBTDLBCGTDLKA1TABLEFigure 4.2: Phase 1: Before bare assertion of p=\u2018It is raining\u2019.3Unlike DM si, DM no does not convey a previous aligned state (Phase 0).4In Chapter 1 I defined discourse commitments as including both public and private commit-ments by the interlocutors; the key aspect is that they are not shared, as opposed to the CG.135A wants B to add p to the CG. This is the function of an assertion. A asserts p,thus tabelling the proposition. This has a particular effect on the Dialogue Board:as in Farkas and Bruce (2010), p is added to the projected CG. This is a reflectionof the proposal nature of assertions. This update also means that the informationstate KA will change: this dynamic change is reflected in the incremental index ofKA (KA2). The effect of the utterance of a bare assertion is modelled in Figure 4.3:ACURRENT PROJECTEDDCA pp=\u2018It is raining\u2019TDLACG pp=\u2018It is raining.\u2019TDLBCURRENT PROJECTEDDCBTDLBCG pp=\u2018It is raining.\u2019TDLKA2TABLEpp=\u2018It is raining.\u2019Figure 4.3: Phase 2: Bare assertion of p = \u2018It is raining\u2019.After the assertion different outcomes are possible: Figure 4.4 represents theoutcome desired by A, where p is accepted by B, and can therefore move fromthe projected CG to the current CG. This also means that p is removed fromindividualized discourse commitments (Farkas and Bruce, 2010; Malamud andStephenson, 2015). It is also removed from the Table, since it has been resolved:136ACURRENT PROJECTEDDCATDLACG pp=\u2018It is raining.\u2019TDLBCURRENT PROJECTEDDCBTDLBCG pp=\u2018It is raining.\u2019TDLKA3TABLEFigure 4.4: Phase 3: After bare assertion of p = \u2018It is raining.\u2019 assumingSpeaker B accepts p.Now that we have modelled the effect of a bare assertion on the DialogueBoard, let us turn to modelling the effect of adding DM no.4.2.1.1 How DM no modifies non-evaluative assertionsLet us focus first on the effect of DM no on non-evaluative assertive anchors. Idefined such assertions as being non-judge dependent, that is, all interlocutors cancommit to its truth.5Previous analyses of DM no agree on its confirmational nature: when a lan-guage user adds DM no to a clause they are requesting confirmation of a fact or anopinion. This means that the use of DM no would not be acceptable in contextswhere the Speaker knows that the Addressee is not in a position to confirm a fact.This is in fact the case, as the examples in (131) show. In this example, DM nois only licensed in the context where the Speaker might expect the Addressee toknow whether there is saffron in paella:(131) 3 Context 1: Talking to my father, who has been making paellas for ages.7 Context 2: Talking to Donald Trump.5For this definition, see Chapter 1 section 1.3.1.137Lathepaellapaellallevatakesazafra\u00b4n,saffronno?no\u2018Paella has saffron, [no]?\u2019We need to represent the fact that one of the licensing conditions of DM nothat we need to model is that the Speaker thinks that the Addressee can confirmthe information conveyed by the anchor. I model this contextual restriction byadding p (in this case, p=\u2018Paella has saffron.\u2019) to the projected DCs of the in-terlocutors. Note that this is not a misalignment per se, but the interlocutors arenonetheless not maximally aligned: given the assumptions we laid out about howone of the main driving forces of a conversation is to share commitments pub-licly, this configuration could be improved by trying to add the proposition p tothe shared CG.So how does adding DM no to a non-evaluative assertion affect the effect theutterance has on the Dialogue Board? Let us go back to the example used at thebeginning of section 4.2.1:(132) Context: A has checked the forecast: it\u2019ll be raining all week where B is.Later, A and B are talking on the phone long-distance. A says:Esta\u00b4islloviendo,rainingno?no\u2018It\u2019s raining, [no]?\u2019138ACURRENT PROJECTEDDCA pp=\u2018It is raining\u2019TDLACGTDLBCURRENT PROJECTEDDCB pp=\u2018It is raining\u2019TDLBCGTDLKA1TABLEFigure 4.5: Phase 1: Before a modified assertion with DM no.Phase 2 represents the dialogue immediately after the no-tagged non-evaluativeassertion is uttered. The update proposed by an assertion is represented in the Di-alogue Board by adding a proposition p to the projected CG. The difference witha bare assertion is that the Speaker (A, bold-faced in Figure 4.6) puts on the Tablethe update proposed by the anchor, represented here as KA2, and not the proposi-tion itself. This is a way to model that the Speaker is requesting confirmation ofthe transition from KA1 to KA2.ACURRENT PROJECTEDDCA pp=\u2018It is raining\u2019TDLACG pp=\u2018It is raining\u2019TDLBCURRENT PROJECTEDDCB pp=\u2018It is raining\u2019TDLBCG pp=\u2018It is raining\u2019TDLKA2TABLEKA2Figure 4.6: Phase 2: Speaker A utters a modified assertion with DM no.139This accounts for the description of DM no as a \u2018control check marker\u2019 (mar-cador de control del contacto (Briz, 1998)): it makes explicit that the Speaker(A) wants the contribution of the Addressee (B) in updating the conversation.6 A\u2018canonical\u2019 update proposed by an assertion would not require the active partic-ipation of the Addressee\u2013however, here we see that the fulfillment of the updatedepends on the Addressee. This engagement of the Addressee is marked by therising intonation of the DM (Montan\u02dcez Mesas, 2015; Wiltschko and Heim, 2016;Heim, 2019a).7Assuming that A\u2019s desires come true and B does not reject the update proposedby the anchor, the resulting state will look like Figure 4.7: p moves to the CG, andis removed from other sets.ACURRENT PROJECTEDDCATDLACG pp=\u2018It is raining\u2019TDLBCURRENT PROJECTEDDCBTDLBCG pp=\u2018It is raining\u2019TDLKA3TABLEFigure 4.7: Phase 3: Acceptance of modified assertion with DM no.6Another way in which this contribution has been formalized is as a tentative attribution to theAddressee of an illocutionary act (Kiss, 2018).7Their sources are recent studies of rising intonation and sentence-final DMs similar to DMno. There is a vast literature about rising intonation and engagement of the Addressee (see amongothers Ward and Hirschberg (1985); Geluykens (1987); Pierrehumbert and Hirschberg (1990);Bartels (1997); Gunlogson (2004); S\u02c7afa\u00b4r\u02c7ova\u00b4 (2005); Truckenbrodt and Maienborn (2012); Heim(2019a)).1404.2.1.2 How DM no modifies evaluative assertionsEvaluative assertions behave differently with respect to DM no. In the case ofsentences that contain predicates of personal taste, the felicitous use of the DMdepends on whether the Addressee is able to confirm the evaluation or not, as in(133):(133) 4 Context 1: A and B are eating a paella. A says to B:7 Context 2 : A and B are at a restaurant. A is tasting a plate of paella,and B is eating pasta:Lathepaellapaellaesta\u00b4isriqu\u0131\u00b4sima,tasty.SUPERno?no\u2018The paella is very delicious, [no?]\u2019The sentence La paella esta\u00b4 riqu\u0131\u00b4sima (\u2018The paella is delicious\u2019) can be un-derstood to mean that the paella is delicious for everyone in the context or thatthis judgement applies only to the Speaker.8 This ambiguity can be modelled inthe Dialogue Board by adding the proposition p (p=\u2018The paella is delicious\u201d) todifferent components. If the Speaker wants to state that the paella is tasty as ageneral attribute of the dish, she would propose to add the proposition p=\u2018Thepaella is delicious\u2019 to the CG, as in Figure 4.8:8Not all analyses of this type of predicate assume this ambiguity, however. In some analyses,sentences containing predicates of taste are always subject to a specific perspective (De\u00b4chaineet al., 2017).141ACURRENT PROJECTEDDCA pp=\u2018The paella is delicious\u2019TDLACG pp=\u2018The paella is delicious\u2019TDLBCURRENT PROJECTEDDCBTDLBCG pp=\u2018The paella is delicious\u2019TDLKA2TABLEpp=\u2018The paella is delicious\u2019Figure 4.8: A utters a bare evaluative, judge-independent assertion.If the Speaker wants to state that the paella is tasty to her, then the propositionis added to the Speaker\u2019s Discourse Commitments, but not to the shared CG asrepresented in Figure 4.9. This means that only she commits to the truth of p.9ACURRENT PROJECTEDDCA pp=\u2018The paella is delicious\u2019TDLACGTDLBCURRENT PROJECTEDDCBTDLBCGTDLKA2TABLEpp=\u2018The paella is delicious\u2019Figure 4.9: A utters a bare evaluative, judge-dependent assertion.9Note that although p is not added to the CG, the fact that S uttered p will be registered. Anotherway in which this difference could be modelled is by saying that the difference lies on whether p orBel(p) is tabelled (Wiltschko and Heim, 2016). Another option is to have p in another component,the Origo Ground, the set of propositions that are always perspectivized (De\u00b4chaine et al., 2017).142What is the effect of DM no on sentences that contain these predicates ofpersonal taste? Under the analysis presented, using DM no would in effect disam-biguate between the general reading and the to-me reading: by using DM no, theSpeaker (A) is trying to update the CG, and not to make her own DCs public. Howdo we know this? One piece of evidence comes from the difference in acceptabil-ity in various agreeing scenarios. In (134), we see that the sentence La paella esta\u00b4riqu\u0131\u00b4sima (\u2018The paella is delicious)\u2019 would be felicitous when the Speaker thinksthat the Addressee will agree\u2013but also when she thinks they will not:10(134) 4 Context 1: A and B are eating a paella. A likes the paella. B seems tolike it too. A says:4 Context 2: A and B are eating a paella. A likes the paella. B seemsdisgusted. A says:Lathepaellapaellaesta\u00b4isriqu\u0131\u00b4sima.tasty.SUPER\u2018The paella is delicious.\u2019However, when the Speaker uses DM no, the second context (when she seesthat the Addressee does not share her enthusiasm) is not a licensing context, asillustrated in (135):(135) 4 Context 1: A and B are eating a paella. A likes the paella. B seems tolike it too. A says:7 Context 2: A and B are eating a paella. A likes the paella. B seemsdisgusted. A says:10It is true that it seems socially odd (or at least confrontational) to utter this sentence in Context2, but the exchange would improve if the Speaker added Que\u00b4 tiquismiquis eres! (\u2018You are sopicky!\u2019).143Lathepaellapaellaesta\u00b4isriqu\u0131\u00b4sima,tasty.SUPERno?no\u2018The paella is delicious, [no]?\u2019This shows that the use of DM no signals that the Speaker is trying to makethe commitment shared between the two interlocutors, as the analysis predicts.The fact that the Speaker needs to think that the Addressee can confirm thetruth of the proposition is represented in the Dialogue Board by having the propo-sition p=\u2018The paella is delicious\u2019 in both interlocutors\u2019 DC, but there is a differ-ence with what we saw section 4.2.1.1: in this case, p is in the current DC of theSpeaker (A), but in the projected DC of the Addressee (B). This explains whythe rejection of a no-tagged evaluative would not remove the commitment to pfrom the DC of Speaker A. In a no-tagged non-evaluative the commitment of Ais contingent on the contribution of the Addressee (since the proposition was inthe projected DC of both interlocutors), so its rejection would also involve theremoval of p from A\u2019s DC. However, as we can see in (136), the rejection of ano-tagged evaluative by B does not remove A\u2019s commitment: she still thinks thatthe rice dish is delicious.(136) A: Este arroz esta\u00b4 riqu\u0131\u00b4simo.\u2018This rice is delicious.\u2019B: No, se han pasado con la sal.\u2018No, they put too much salt.\u2019A: Bueno, pues a m\u0131\u00b4 me gusta.\u2018Well, I like it.\u2019Phase 1 is represented in Figure 4.10:144ACURRENT PROJECTEDDCA pp=\u2018The paella is delicious\u2019TDLACGTDLBCURRENT PROJECTEDDCB pp=\u2018The paella is delicious\u2019TDLBCGTDLKA2TABLEFigure 4.10: Phase 1: Before the no-modified assertion of an evaluativeproposition.DM no is used to ask whether moving the proposition p from the individualto the shared component is felicitous. This is shown in Figure 4.11, again bytabelling KA2 as a placeholder for the transition from KA1 to KA2:ACURRENT PROJECTEDDCA pp=\u2018The paella is deli.\u2019TDLACG pp=\u2018The paella is deli.\u2019TDLBCURRENT PROJECTEDDCB pp=\u2018The paella is deli.\u2019TDLBCG pp=\u2018The paella is deli.\u2019TDLKA2TABLEKA2Figure 4.11: Phase 2: Speaker A puts forward a no-modified assertion of anevaluative proposition.If there is agreement, the Table will be emptied (since all the issues would havebeen resolved) and the proposition will be part of the CG for both interlocutors A145and B, as Figure 4.12 shows.ACURRENT PROJECTEDDCATDLACG pp=\u2018The paella is delicious\u2019TDLBCURRENT PROJECTEDDCBTDLBCG pp=\u2018The paella is delicious\u2019TDLKA3TABLEFigure 4.12: Phase 3: Acceptance of modified assertion of an evaluativeassertion.The analysis proposed for DM no accounts for the co-occurrence of the DMwith an anchor containing a predicate of personal taste, but it also accounts forthe infelicity of having an anchor containing an internal-state predicate in the firstperson, as (137) shows. This specific type of evaluative assertion can only beSpeaker-dependent, and the proposition they denote does not enter the CG.(137) After a day at uni, A comes home and yawns. A says to B:Estoyamcansada,tired(#no?)no\u2018I am tired, [no?]\u2019Note, however, that there might be a context in which (137) can be uttered: theSpeaker (A) has a condition that makes her not be aware of her own exhaustion,yet she still displays all signs of being tired. Since she is not capable of realizingwhen she is tired, she needs the cooperation of her Addressee to know whether ornot she is tired:146(138) Context: A suffers from a condition that makes her unaware of her ownpain and exhaustion. She comes back from work looking visibly tired,and B looks at her with pity and asks if she wants to relax on the couch.A says:Estoyamcansada,tiredno?no\u2018I am tired, [no?]\u2019In this case, the use of the DM would be felicitous. Although this contextis not common, it is a welcome effect that the analysis allows for this type ofoccurrence, especially since it only allows this if the Addressee can contribute tothe acceptance of the conversational update.4.2.2 DM no negotiates the update proposed by exclamationsGiven the analysis proposed for the core meaning of DM no, we wouldn\u2019t ex-pect to find it in co-occurrence with exclamations: exclamations are typicallySpeaker-oriented, conveying the feelings of the Speaker (Gutie\u00b4rrez-Rexach, 1996;Beyssade and Marandin, 2006; Castroviejo Miro\u00b4, 2006; Rett, 2008; Gutie\u00b4rrez-Rexach, 2008; Rett, 2011). This means that the component of the dialogue thatthis type of speech act targets is an individual component, in this case the Speaker\u2019sDC. Given our analysis, we would expect that this type of speech act would notbe an expected anchor to DM no. And yet, exclamations can be modified by DMno, as we saw in 2.2.1.6 and as illustrated in (139):(139) Context: A and B are married, and A has organized a surprise birthdayparty for B, with the help of a mutual friend C. C lures B into the home,and A and some friends surprise B. A then turns to B and says:H147Que\u00b4whatsorpresa,surpriseno?no\u2018What a surprise, [no]?\u2019Part of the analysis proposed for DM no was inspired by a comment made byone of the consultants, who judged this sentence infelicitous in the given context.When asked why, the explanation was that \u201cit seems that it is saying \u2018You haveto be surprised\u201d\u2019. The judgement was therefore in terms of a social faux-pas: ifyou organize a surprise party, you shouldn\u2019t impose the success of the party onthe birthday person. What is key here, is that the content of the exclamation (thatsomething is surprising) is not Speaker-oriented. This reading contrasts with thedefault reading given to bare exclamatives as Speaker-oriented, bringing up theAddressee into the conversation. It is the use of DM no that calls on the Addressee,as seen in (140): a bare exclamation is not felicitous in the same context.(140) Context: A and B are married, and A has organized a surprise birthdayparty for B, with the help of mutual friend C. C lures B into the home, andA and some friends surprise B. A then turns to B and says:H#Que\u00b4whatsorpresa!surprise\u2018What a surprise!\u2019The Speaker in (139) is essentially asking whether it can be made a sharedcommitment that the party is a surprise (or that the Addressee feels surprise). Inother words: it is asking whether a proposition can make it from the DCs to theCG.I model the effect of DM no on exclamations the same way I analyzed andmodelled the effect of DM no on sentences containing predicates of personal taste.The addition of the DM would add the proposition (which is assumed to be in148both DCs) to the CG: it is asking to make an individual commitment into a sharedcommitment (Figure 4.13).ACURRENT PROJECTEDDCA pp=\u2018It is a surprise\u2019TDLACG pp=\u2018It is a surprise\u2019TDLBCURRENT PROJECTEDDCB pp=\u2018It is a surprise\u2019TDLBCG pp=\u2018It is a surprise\u2019TDLKA2TABLEKA2Figure 4.13: Phase 2: A utters an exclamation with the DM no.The acceptance of this move would add the proposition p=\u2018It\u2019s a surprise\u2019 tothe (current) CG, and remove it from all other components (Figure 4.14):ACURRENT PROJECTEDDCATDLACG pp=\u2018It is a surprise\u2019TDLBCURRENT PROJECTEDDCBTDLBCG pp=\u2018It is a surprise\u2019TDLKA3TABLEFigure 4.14: Phase 3: Acceptance of no-tagged exclamation.Just as with evaluative assertions, if the move is rejected, then the propositionwill be removed from all sets except for the Speaker\u2019s DC (Figure 4.15):149ACURRENT PROJECTEDDCA pp=\u2018It is a surprise\u2019TDLACGTDLBCURRENT PROJECTEDDCBTDLBCGTDLKA3TABLEFigure 4.15: Phase 3(b): Rejection of modified exclamation. p=\u2018It\u2019s asurprise.\u2019.4.2.3 DM no negotiates the update proposed by commandsDM no can co-occur with imperative anchors; however, we also saw in Chapter 2that there seems to be a gradience of what type of command is more felicitouswith the DM. According to Garc\u0131\u00b4a Vizca\u0131\u00b4no (2005), it is only the less imposingcommands (like suggestions, or advice) that can take the DM. I claim that it isnot the anchor in itself that disallows the use of the particle, but the context ofuse: in (141), the use of DM no is disallowed in a context (Context 2) wherethere is a power imbalance (that is, the Speaker has institutional authority over theAddressee), where there is no such power imbalance (Context 1):(141) 4 Context 1: A spouse to another spouse, seeing a full trash bin:7 Context 2: A superior officer to a private, seeing a full trash bin:Sacatake-out.IMP.2SGlathebasura,garbageno?no\u2018Take out the garbage, [no]?\u2019150Essentially, using DM no with a command as an anchor turns the commandinto a suggestion. This is in fact an effect of the analysis of DM no presented: theSpeaker does not unilaterally impose the conversational update, but allows for theparticipation of the Addressee in its acceptance. In Context 2, this effect is notappropriate: a superior officer does not need the acceptance of the private.Commands propose to update the To-Do-List of the Addressee with an out-come o (Beyssade and Marandin, 2006). When Speaker A uses the command in(142) the effect it has on the Dialogue Board is to update the projected TDL of theAddressee, in this case B (TDLB).(142) Estudiastudy.IMP.2SGma\u00b4s.more\u2018Study more.\u2019At the same time, the outcome is added to the shared TDL that includes thepublicly shared outcomes that each interlocutor has committed to see through. Imark whose commitment it is (in this case, who is supposed to do the studying)with an index (in this case, oB). This is illustrated in Figure 4.16.ACURRENT PROJECTEDDCATDLACGTDL oBo=\u2018Study more.\u2019BCURRENT PROJECTEDDCBTDLB oo=\u2018Study more.\u2019CGTDL oBo=\u2018Study more.\u2019KA2TABLEoo=\u2018Study more.\u2019Figure 4.16: Speaker A utters a bare command..151Figure 4.17 represents the desired resolution (desired by Speaker A): SpeakerB accepts the update. The Table is emptied.ACURRENT PROJECTEDDCATDLACGTDL oBo=\u2018Study more\u2019BCURRENT PROJECTEDDCATDLB oo=\u2018Study more\u2019CGTDL oBo=\u2018Study more\u2019KA3TABLEFigure 4.17: Acceptance of a bare command by Speaker B. o=\u2018Studymore.\u2019.What is different, then, when a command is modified with DM no? Yangand Wiltschko (2016) in their study of Northern Mandarin ha mention that theeffect of this confirmational particle on commands is to request confirmation ofthe intention of the Addressee to perform the action requested by the command.Rudin (2018) offers a similar analysis of rising imperatives in English, framingit in terms of preferences. This is precisely what DM no does: when a Speakeradds the DM to a command, she is requesting confirmation of the Addressee\u2019sintention to commit to the outcome o. This can be seen in the different accept-ability between the contexts in (143): when the Speaker (A) sees that it is possiblethat the Addressee might perform the action (bringing a beer), the use of the DMis accepted. The use of the DM when there is no such indication is dispreferred(although not entirely infelicitous):(143) 3 Context 1: A and B are sitting on the couch. B gets up and goes to thekitchen. A tells B:H152? Context 2: A and B are sitting on the couch. A tells B:HTra\u00b4emebringIMP.2SG=meunaacervecita,beer.DIMno?no\u2018Bring me a beer, [no?]\u2019This contrasts with the different acceptability of a bare command in this case,illustrated in (144), where both contexts are equally felicitous:11(144) 3 Context 1: A and B are sitting on the couch. B gets up and goes to thekitchen. A tells B:H3 Context 2: A and B are sitting on the couch. A tells B:HTra\u00b4emebring.IMP.2SG=meunaacervecita.beer.DIM\u2018Bring me a beer.\u2019I represent this in the model by adding the outcome o=\u2018Bring me a beer\u2019 tothe TDLB in Phase one: that is, even before uttering the no-modified command,A thinks that the Addressee might commit to the outcome (Figure 4.18).11Note that in both cases, Context 2 seems pragmatically odd because it is perceived as rude.153ACURRENT PROJECTEDDCATDLACGTDLBCURRENT PROJECTEDDCBTDLB oo=\u2018Bring me a beer\u2019CGTDLKA1TABLEFigure 4.18: Phase 1: Before a command modified by DM no.When A puts forward a modified command like in (143), what she is doingis requesting confirmation from the Addressee of the conversational update of thecommanding anchor. This is represented in Figure 4.19: the outcome o=\u2018Bringme a beer\u2019 is placed in the projected shared TDL. The Speaker (A) requests con-firmation of this move by tabelling the conversational update:ACURRENT PROJECTEDDCATDLACGTDL oBo=\u2018Bring me a beer.\u2019BCURRENT PROJECTEDDCBTDLB oo=\u2018Bring me a beer.\u2019CGTDL oBo=\u2018Bring me a beer.\u2019KA2TABLEKA2Figure 4.19: Phase 2: Speaker utters command modified with DM no.The desired outcome is represented in Figure 4.20: if the Addressee accepts154to comply, the outcome is added to the current TDLs.ACURRENT PROJECTEDDCATDLACGTDL oBo=\u2018Study more\u2019BCURRENT PROJECTEDDCBTDLB oo=\u2018Study more\u2019CGTDL oBo=\u2018Study more\u2019KA3TABLEFigure 4.20: Phase 3: Acceptance of command modified with DM no.In section 4.5 we will come back to the effect of DM no on commands in termsof mitigation, an effect that is expected given the analysis proposed here.4.2.4 DM no negotiates the update proposed by (rhetorical)questionsWe have seen in section 2.2.1.5 that although DM no can co-occur with interroga-tives, these interrogatives are not information-seeking questions. In (145) we seethis distribution again: in Context 2, the interrogative anchor is an informationseeking question (A needs to know the answer in order to prepare a gift), andcannot be followed by DM no. However, in Context 1 the anchor is a rhetoricalquestion: B cannot possibly know the answer to it.(145) 3 Context 1: A and B are talking about possible names for a friend\u2019s baby.This friend is notorious for choosing unexpected names for his children,and he is very secretive about them. A says to B:1557 Context 2: A is talking B about a friend\u2019s future child, who is due in acouple of weeks. She wants to know what the name of the child will be sothat she can get an embroidered blanket. A says to B:Co\u00b4mohowseREFLllamara\u00b4,call.FUT.3SGno?no\u2018What will s\/he be called, [no]?\u2019In Context 1, A is asking B whether they are wondering the same thing (whatthe baby\u2019s name will be). In a sense the meaning could be paraphrased as \u201cIwonder what their name will be\u201d. This paraphrase makes sense given the analysisthat I will assume for rhetorical questions: that although they have interrogativeform, they have assertive illocutionary force (Han, 2002; Corr, 2016).In Context 1 of (146), the Speaker does not really want to know the answer to\u201cwhat is he talking about?\u201d, but is upset about being stuck in a talk she does notwant to hear, and wonders whether the Addressee feels the same way.(146) 3 Context 1: A is sitting beside B listening to a TED talk about a topicshe despises. She looks at B and asks:H7 Context 2: A arrives late to a TED talk and has no idea what the speakeris talking about. She asks B:HDeofque\u00b4whatesta\u00b4ishablando,talkingno?no\u2018what is he talking about, [no]?\u2019Given that I assume that rhetorical questions have assertive illocutionary force,this means that they will propose the same type of conversational update as anassertion: to propose to add a proposition to the CG. The presence of the DM noworks as follows: in the case of (146), the proposition that is being added to the156projected DCB, and the projected CG is p=\u2018He doesn\u2019t know what he is talkingabout.\u2019ACURRENT PROJECTEDDCA pp=\u2018He doesn\u2019t knowwhat he\u2019s talking about\u2019TDLACG pp=\u2018He doesn\u2019t knowwhat he\u2019s talking about\u2019TDLBCURRENT PROJECTEDDCB pp=\u2018He doesn\u2019t knowwhat he\u2019s talking about\u2019TDLBCG pp=\u2018He doesn\u2019t knowwhat he\u2019s talking about\u2019TDLKA2TABLEKA2Figure 4.21: Phase 2: Speaker A modified rhetorical question with DM no.4.3 What speakers cannot do with DM noGiven that DM no is here analyzed as requesting from the Addressee the validityof making the update proposed by the anchor public and shared, there are twotypes of speech act that are predicted not to allow the use of DM no:Prediction 1: Information-seeking questions do not assert any proposition, un-like the anchor of a no-modified utterance. There is therefore no belief oroutcome to confirm, and hence it will be infelicitous to use DM no with thistype of questions. This is discussed in section 4.3.1.Prediction 2: DM no invites a contribution from the Addressee, so it is predictedto be incompatible with Speaker-centered Speech Acts such as commis-sives, expressives, and declarations. This is discussed in sections 4.3.2through 4.3.4.In the following sections we see that both predictions are borne out.1574.3.1 DM no is infelicitous with (information-seeking)questionsDM no can modify rhetorical questions, but it cannot modify information-seekingquestions, as (147) illustrates:(147) (repeated from 64)3 Context 1: A is sitting beside B listening to a TED talk about a topicshe despises. She looks at B and asks:H7 Context 2: A arrives late to a TED talk and has no idea what the speakeris talking about. She asks B:HDeofque\u00b4whatesta\u00b4ishablando,talkingno?no\u2018What is he talking about, [no]?\u2019As Yang and Wiltschko (2016) point out, \u201ca question is itself a request forinformation and hence by using a question as the host clause, [the] S[peaker] hasnothing to confirm from [the] A[ddressee]\u201d (Yang and Wiltschko, 2016, p.75).The same logic applies here: using DM no in (147) conveys that the Speaker isasking whether the Addressee shares that same question, but not asking about thecontent of the question directly. This meta-level of asking, in a sense, forces arhetorical reading of the question when using the DM no, which is felicitous.4.3.2 DM no is infelicitous with thanking and apologizingspeech actsActs like apologizing (148) and thanking (149) cannot be modified with DM no:158(148) #Loitsiento,feel.PRS.1SGno?no\u2018I\u2019m sorry, [no]?\u2019(149) Context: B had promised to cook dinner for A. At 8pm, they hand A aplate of pasta:#Muchasmanygracias,thanksno?no\u2018Thank you very much, [no]?\u2019But again, it is possible in certain contexts to have the utterances in (149)and (148). Specifically, if it is not the person thanking but another discourseparticipant, as in (150), where it is the mother of the child who utters the thanksfollowed by no:(150) Context: B had promised to cook dinner for A and C. At 8pm, they handA and C a plate of pasta. A says to C, pointing at B:Muchasmanygracias,thanksno?no\u2018Thank you very much, [no]?\u2019In these cases, the utterance Muchas gracias (\u2018Thank you very much\u2019) is notan expressive in the sense that the Speaker is not performing the act, but is puttingwords into the Addressee\u2019s mouth, so to speak: (150) asks the Addressee to con-firm that they are indeed thanking the person who cooked.1594.3.3 DM no is infelicitous with commissivesDM no cannot follow commissive speech acts, which include promises and oaths.Commissives are, like assertions, usually formally declaratives. But crucially,they introduce a different commitment to the dialogue: while assertions committhe Speaker to the truth of a proposition by proposing the addition of said propo-sition to the CG, commissives commit the Speaker to an outcome by adding thisoutcome to her To-Do-List (TDL) (Portner, 2004; Beyssade and Marandin, 2006).(151) #Teyouloitjuro,swear.PRS.1SGno?no\u2018I swear, [no]?\u2019In (151), the Speaker cannot ask the Addressee to confirm whether an oathholds: this would defeat the purpose of making a oath in the first place. Com-missives only require the commitment of the Speaker. In fact, if we \u201cinvert\u201d theroles (if the utterance is requiring the Addressee to promise something, as op-posed to the Speaker) the addition of DM no is felicitous. But note: this changealso changes the type of speech act of (152): this is no longer a commissive, sinceit doesn\u2019t follow the felicity conditions for a successful commissive\u2013namely thatit is the Speaker who commits to an action.(152) Memeloitprometes,promise.PRS.2SGno?no\u2018You promise, [no]?\u2019Commissives typically require a first person subject, so by changing the personof the subject we are also changing the type of speech act from a commissive toan assertion. This also happens with the declarations, as we will see in the nextsection. 1212A reason for this is the fact that both commissives and declarations are declaratives in their1604.3.4 DM no is infelicitous with declarationsDeclarations are speech acts that effectively change a state of affairs in the worldwhen they are performed; some examples include firing someone, or joining twopeople in matrimony. The words uttered when an employer says to an employee\u2018You are fired\u2019 effectively terminate the employment.Related to the infelicity of commissives as anchors of DM no is its infelicitywith declarations, as the example in (153) illustrates:(153) 7 Context 1: A boss tells one of the managers to fire an employee. Themanager walks up to the employee and says:3 Context 2: An employee walks out of the boss\u2019 office and starts packingtheir things. The manager walks over to the employee, and says:Esta\u00b4saredespedido,firedno?no\u2018You\u2019re fired, [no]?\u2019The example in (153) shows that there is nothing about the sentence in itselfthat is infelicitous with the tag\u2013in the right context, it is felicitous to use the tagwith the host Esta\u00b4s despedido (\u2018You\u2019re fired\u2019). Importantly, DM no cannot followthe declarative when it is used as a declaration, that is, when it is used to actuallychange the world, in this case, fire someone. However, it is felicitous if the declar-ative is used as an assertion: the Speaker is not firing the Addressee, but wants toconfirm their suspicions of them having been fired. This is further evidence thatit is the type of speech act, and not the sentence type, that restricts the use of theDM.One of the main characteristics of both commissives and declarations is thatthey are not proposals: they do not propose to update the context with a change insentence type; when they do not fulfill the conditions required for commissives\/declarations, theymight be understood as assertions by the Addressee.161the world, they effectively perform such a change as they are being uttered. DMno is not permitted in these cases, as is illustrated in (154):(154) #Osyoudeclarodeclare.1SGunidosjoinedeninmatrimonio,matrimonyno?no\u2018I declare you joined in matrimony, [no]?\u2019This restriction follows from the analysis presented here: it is not possible fora felicitous declaration to request a confirmation from the Addressee, since thiswould not result in a change in the world like declarations impose.4.4 Why monitor the S-A relation: confirming abiasThis section focuses on why the Speaker uses DM no: I claim that the reason isto confirm a bias. The literature on bias has distinguished between at least twotypes: contextual bias, discussed in section 4.4.1, and epistemic bias, discussed insection 4.4.2. To these two types we also have to add an outcome bias (discussedin section 4.4.3) to describe the instances in which DM no modifies a command.4.4.1 DM no conveys a contextual biasContextual bias is defined as dependent on the evidence available in the currentconversational context.13 This is distinct from epistemic bias, which focuses onthe previous beliefs held by the Speaker for or against p. Contextual bias relies oncontextual evidence, defined as in (155).13In the literature, this type of bias is often referred to as evidential. However, one could arguethat epistemic bias is in itself a sort of evidence. It would be more accurate to refer to evidentialbias as \u201ccontextual\u201d or \u201cexperiental\u201d bias. See Waldie (2012) for a related and more in-depthdiscussion.162(155) Contextual Evidence (Bu\u00a8ring and Gunlogson, 2000):Evidence that has just become mutually available to the participants in thecurrent discourse situation.According to Sudo (2013) an evidential context can be [positive] or [nega-tive]: depending on whether the context has positive or negative evidence for p.14A neutral contextual setting would be one where there is no evidence mutuallyavailable to both participants for or against p (e.g. during a long-distance phonecall, as in (159)). Context 1 in (156) is [positive] since there is evidence in favourof p (p= \u2018It is raining\u2019), whereas Context 2 is [negative] since there is evidenceagainst p (similar contexts are used in Bu\u00a8ring and Gunlogson (2000); Romero andHan (2004); Sudo (2013); Heim (2019a):(156) 3 Context 1 Contextual bias: [positive]B enters A\u2019s windowless computer room wearing a dripping wet coat. Asays:7 Context 2 Contextual bias: [negative]B enters A\u2019s windowless computer room wearing a dry coat and sun-glasses. A says:Esta\u00b4islloviendo,rainingno?no\u2018It\u2019s raining, [no]?\u2019Restrictions apply, however: the contextual evidence has to be indirect. InContext 1 (158), the Speaker has only indirect evidence for p (p = \u2018It is raining\u2019),namely the dripping wet coat: in this context, the use of the DM is felicitous. In14In fact, for Sudo (2013) it is possible to have [-positive] and [-negative] contexts. Here Iabstract away from these types. These contexts describe questions that are incompatible withpositive or negative evidence, respectively.163Context 2, the contextual evidence is direct and the Speaker can actually see therain: in this context, the use of the DM is not felicitous.15(158) 3 Context 1 Contextual bias [positive, indirect]B enters A\u2019s windowless computer room wearing a dripping wet coat. Asays:7 Context 2 Contextual bias [positive, direct]A enters B\u2019s room, looks out of the window and it is raining. A says:Esta\u00b4islloviendo,rainingno?no\u2018It\u2019s raining, [no]?\u2019Given what we have just seen, we could conclude that in order to use DM nothe Speaker needs positive, indirect contextual evidence for p. However, the useof no is licensed in contexts where there is no contextual evidence available, as in(159) where the interlocutors do not share an immediate physical context:(159) Context Contextual bias: [neutral]A and B are talking long-distance. A has checked the forecast at B\u2019slocation, and it said that it would be raining all day. A asks:15DM no cannot be used when the positive evidence comes from a presupposition in the previousdiscourse move:(157) B tells A that their sister is coming to town. A says:Tieneshaveunaahermana,sisterno?no\u2018You have a sister, [no]?\u2019I thank Floris Roelofsen (p.c.) for this observation. Note that (157) does not improve even witha surprise marker like \u201cOh\u201d, as the same example with DM huh would in English (Oh, so you havea sister, huh?).(Wiltschko, p.c.), although this \u201coh\u201d is not needed with the DM huh in this context(Matthewson, p.c.).164Esta\u00b4islloviendo,rainingno?no\u2018It\u2019s raining, [no]?\u2019The interlocutors in (159) may not share the same context, but the Speaker ex-pects a positive answer not based on what she perceived but on what she knows\u2013theSpeaker has an epistemic bias. This is the focus of the next section.4.4.2 DM no conveys an epistemic biasContextual evidence alone cannot be the only factor licensing the use of DM no:the Speaker\u2019s previous beliefs are also to be taken into account. If we modifythe context in (156), we see that we cannot use DM no unless there is a previousepistemic bias in favour of the proposition expressed by the anchor p (p = \u2018Itis raining\u2019). This is shown in the different felicity judgements in (160), whereDM no can only be felicitously used if there is a positive bias for the truth of theproposition expressed by the anchor (in this case, the fact that it is raining):(160) 3 Context 1 Contextual bias: [positive]hola Epistemic bias: for p (p=\u2018It is raining.\u2019)A has checked the forecast and it said that it\u2019ll be raining all week. Later,B enters A\u2019s windowless computer room wearing a dripping wet coat. Asays:7 Context 2 Contextual bias: [positive]hola Epistemic bias: against p (p=\u2018It is raining.\u2019)A has checked the forecast and it said that it\u2019ll be sunny all week. Later,B enters A\u2019s windowless computer room wearing a dripping wet coat. Asays:165Esta\u00b4islloviendo,rainingno?no\u2018It\u2019s raining, [no]?\u2019Example (160) shows that the Speaker cannot have prior epistemic evidenceagainst p.16 However, an epistemic bias for p alone does not account for thefelicitous use of DM no: even if the Speaker believes p, if there is direct, negativeevidence for p, DM no is not felicitous. The example in (161) shows that DM nocannot be used if there is contextual evidence against p: in this case, the sunglassesare indirect evidence against the Speaker\u2019s belief that it is raining:17(161) 3 Context 1 Contextual bias: [positive]hola Epistemic bias: for p (p=\u2018It is raining.\u2019)A has checked the forecast and it said that it\u2019ll be raining all week. Later,B enters A\u2019s windowless computer room wearing a dripping wet coat. Asays:7 Context 2 Contextual bias: [negative]hola Epistemic bias: for p (p=\u2018It is raining.\u2019)A has checked the forecast and it said that it\u2019ll be raining all week. Later,B enters A\u2019s windowless computer room wearing sunglasses. A says:Esta\u00b4islloviendo,rainingno?no\u2018It\u2019s raining, [no]?\u201916This includes affirmative and negative anchors: in the case of negative anchors, such as Noesta\u00b4 lloviendo (\u2018It is not raining\u2019), the Speaker will have a previous belief about truth of \u00acp.17According to Sudo (2013), this means that it is [-negative]: it cannot be used when there isnegative evidence.166However, it is possible to use the DM if the evidential context is neutral. Thisis shown in example (162), where the interlocutors do not share the same eviden-tial context. As long as there is an epistemic bias in favour of p and no contradic-tory evidence, DM no is accepted:(162) (partly repeated from (159))3 Context 1 Contextual bias: neutralhola Epistemic bias: for p (p=\u2018It is raining.\u2019)A and B are talking long-distance. A has checked the forecast at B\u2019s loca-tion, and it said that it would be raining all day. A says:7 Context 2 Contextual bias: neutralhola Epistemic bias: neutralA and B are talking long-distance. A says:Esta\u00b4islloviendo,rainingno?no\u2018It\u2019s raining, [no]?\u2019To sum up, in order to felicitously use DM no there cannot be a negative bias,neither contextual nor epistemic.4.4.3 Bias and anchors beyond assertionsThe literature on biased questions has mainly focused on the biases derived fromexpectations regarding the truth of a proposition, given the type of data discussed(polar questions, and in the case of tags, combinations of assertions and polarquestions). However, we have discussed in Chapter 2 how DM no is used tomitigate the use of commands, turning them into suggestions. If we want to extendthe discussion of bias to these examples, we have to go beyond expectations of thetruth of a proposition.167For current purposes, I will call the bias linked to no-modified commandsoutcome bias: the Speaker has an expectation of the Addressee agreeing to theoutcome denoted by the imperative anchor. Interestingly, the evidential contextsin which modified commands are felicitous are almost the opposite of the oneswe discussed in section 4.4.2: DM no is only felicitous when there is no positivecontextual evidence for the outcome, as example (163) shows:(163) 3 Context 1 Contextual bias: negativehola Outcome bias: for o (o=\u2018Study longer.\u2019)A and B are studying together for a hard final exam. After 30min, B packstheir backpack and starts to get up. A says:7 Context 2 Contextual bias: positivehola Outcome bias: for o (o=\u2018Study longer.\u2019)A and B are studying together for a hard final exam. After 30min, B picksup one of the required readings for the course. A says:Estudiastudy.IMP.2SGunapocobitma\u00b4s,moreno?no\u2018Study a bit longer, [no]?\u2019In (163), the only difference between contexts is whether the Speaker has con-textual evidence to think that the Addressee will not study longer (the outcomeexpressed by the imperative). If the Addressee behaves as if they are studyinglonger, then the use of the DM is not felicitous. But if there is contextual evidenceagainst the outcome, then the DM can be used.It is not the case that only negative contextual evidence (against the outcome)is necessary: it might also be the case that the evidential context is neutral, asin (164), where the Speaker does not know if the Addressee is for or against theoutcome:168(164) Context: A rather large group of friends are chatting in a small pub. Aarrives and greets everyone, and says:Venga,come.onvamosgo.1platootroanothersitio,placeno?no\u2018Come on, let\u2019s go somewhere else, [no]?\u2019 (modified fromRodr\u0131\u00b4guez Mun\u02dcoz (2009))In conclusion, no-tagged commands cannot be used when the Speaker doesnot have a bias towards the outcome. The evidential contexts needs to be eitherneutral or negative (that is, there has to be some sort of evidence in the contextagainst the outcome).4.5 How this analysis explains other observations:mitigationThe use of DM no has been analyzed in the past as a mitigating device. Theanalysis presented above is not at odds with analyses that propose that the DM nohas a mitigating effect: however, it does claim that mitigation is a byproduct ofthe actual effect of the DM on the anchor.Mitigation is achieved by decreasing speaker assertiveness and increasing co-operation and negotiation (Fe\u00b4lix-Brasdefer, 2004a,b; Czerwionka, 2012; Ucle\u00b4s Ra-mada, 2020). Mitigation is usually defined as the reduction of unwelcome effectsthat a speech act may have on the Addressee (Fraser, 1980), which also alignswith the contexts described in Chapter 2 and in the present chapter. Although Iagree that DM no has the effect of mitigating the illocutionary force of the anchor,I propose that this is a side-effect of a more complicated combination of the con-tribution of no to the utterance\u2013as Yang and Wiltschko (2016) propose, this effectis a byproduct of the analysis. It is a result of the pragmatic effect of DM no onits anchor, and its turn-yielding position.In order to formalize mitigation, Holmes (1984); Sbisa` (2001); Thaler (2012)169propose that it is an operation over speech acts (similar to the ones describedby Searle and Vanderveken (1985) and Vanderveken (1990)): starting from anunmodified, default illocutionary point strength, certain operators can decreasethis strength.18 I propose that the pragmatic effect of adding confirmational noto an anchor utterance is to decrease the degree of strength of the illocutionarypoint of the anchor, in a similar fashion to the analysis of evidentials in CuzcoQuechua in (Faller, 2002). The baseline is the strength that the canonical speechact type of the anchor would have. Vanderveken (1990) assigns a value of 0 to theunmodified, canonical degree of force of a given speech act type:(165) a. Las aran\u02dcas son ara\u00b4cnidos. ASSERTION\u2018Spiders are aracnids.\u2019ILL=ASSERT (SpeakerA, p)STRENGTH: 0b. Ca\u00b4llate ya. DIRECTIVE\u2018Shut up already.\u2019ILL=COMMAND (SpeakerB, p)STRENGTH: 0The clearest example of the mitigating effect of no comes from its occurrencewith commands as in (166): asking the Addressee to confirm whether they wantto update their To-Do-List with an outcome makes the command less imposing,in a way transforming the command into a suggestion.(166) Differences in degree of strength: directivesa. E\u00b4chateput.IMP.2SG=youma\u00b4s.more\u2018Help yourself to more.\u201918Reinforcement is also possible, in which case, the illocutionary operator would increase theillocutionary strength of the act. This is what I propose confirmational eh does, in Chapter 6.170ILL= COMMAND (sB, p)STRENGTH: 0b. E\u00b4chateput.IMP.2SG=youma\u00b4s,moreno?\u2018Help yourself to more, [no?]\u2019 (Go\u00b4mez Gonza\u00b4lez, 2014)ILL= COMMAND (sB, p)STRENGTH: -1This decrease in the strength of the illocutionary point can be explained giventhe analysis offered for DM no: a strong command does not require the participa-tion of the Addressee, except for compliance. The decrease in strength is achievedby allowing the Addressee to confirm that they intend to perform the action ex-pressed by the imperative.4.6 SummaryThis chapter presented the analysis of the pragmatics of DM no: this DM re-quests confirmation from the Addressee that the contextual update proposed bythe Speaker is shareable and public. It does so by conveying the tabelling of theupdate, which I modelled as tabelling the information state of the Speaker (whichincludes the commitment state of the Addressee, as conceived by the Speaker)at that particular time in the conversation. This is meant to capture the transitionfrom one information state (where the commitment to a proposition or an outcomeis in the individual components of the Dialogue Board) to another informationstate (where the commitment is in the shared components).171Chapter 5Modelling the pragmatics-syntaxinterfaceDMs have been the focus of recent formal syntactic literature that proposes toexpand the structure of the clause in the left periphery. The left periphery (specifi-cally, the Complementizer Phrase) is the part of the structure that links a sentenceradical with the rest of the discourse, be it within the same sentence or the cross-sentential discourse (Rizzi, 1997). In Chapters 3 and 4 we have seen that thisis what Peninsular Spanish DM si and no do: DM si establishes a relationshipbetween a proposition and the CG, whereas DM no establishes a relationship be-tween a speech act and the speech act participants.In this chapter I propose a syntactic analysis of DM si and DM no, based onrecent work by Speas and Tenny (2003), Coniglio and Zegrean (2012), Haegemanand Hill (2013), Corr (2016), Thoma (2016), Wiltschko and Heim (2016) andWiltschko (2017). DM si gives us evidence for an articulated highest layer inCP. This analysis will be based on the relative order of the DM with respect toother elements in the left periphery, as well as the prosodic integration of the DMin the sentence. Interactional DMs like no, which as we have seen in Chapter 4recruit the participation of the Addressee, have been used as evidence to includethe speech act participants (Speaker and Addressee) in the syntactic structure. I172will follow this type of analysis for DM no, placing this DM in a syntactic layerabove CP (Wiltschko and Heim, 2016; Jamieson, 2019).Giorgi (2010) proposes that there is a syntactic layer within the CP in charge ofencoding the temporal coordinates of the Speaker. This layer is the left-most posi-tion within the CP, and she calls it the C-speaker. She motivates this proposal withcross-linguistic differences in tense ordering: she proposes that different com-plementizers (with or without this C-speaker projection) introduce indicative andsubjunctive clauses. The indicative complementizer encodes the speaker\u2019s tem-poral location, and the embedded clause is therefore evaluated with respect to theSpeaker\u2019s here and now. This is not the case for subjunctive embedded clauses(Giorgi, 2010, p.43).The syntactic representation of the Addressee has also been proposed in thepast (as we have seen, as early as Ross (1970)) in order to account for vocatives(Moro, 2003; Hill, 2007), imperatives (Zanuttini, 2008) and discourse particlesthat show allocutive agreement. This type of agreement occurs between speech actroles, and not syntactic roles. In Souletin Basque, for example, there are four waysto utter the sentence \u201cPeter worked\u201d depending on who the speaker is talking to: afemale friend (168), a male friend (167), to someone higher in status (169) or to aplural addressee (170) (data from Miyagawa (2013), originally from Oyharc\u00b8abal(1993)):(167) To a female friend:PettekPeter.ERGlanwork.ABSegindo.PRFdin.AUX-3.S.ABS-2.S.C.SM.ALLOC-3.S.ERG\u2018Peter worked.\u2019(168) To a male friend:PettekPeter.ERGlanwork.ABSegindo.PRFdik.AUX-3.S.ABS-2.S.C.MSC.ALLOC-3.S.ERG\u2018Peter worked.\u2019173(169) To someone higher in status:PettekPeter.ERGlanwork.ABSegindo.PRFdizu\u00a8.AUX-3.S.ABS-2.S.F.ALLOC-3.S.ERG\u2018Peter worked.\u2019(170) To plural addressee:PettekPeter.ERGlanwork.ABSegindo.PRFdu.AUX-3.S.ABS-3.S.ERG\u2018Peter worked.\u2019We see here that there is agreement not with the participants of the event de-scribed in the utterance, but with the participants in the utterance (or speech act)itself. In the same way that the auxiliary in Basque agrees with the Addressee,in Upper Austrian German, discourse particles like goi agree with the Adresseein the speech act (Wiltschko and Heim, 2016). In the following examples, wesee that the form of the DM in the right periphery changes depending on whetherthe speaker is talking to someone informally (171), formally (172) or to a pluralaudience (173):(171) EahehothasananeichnnewHund,doggoiCONF.2.INF.\u2018You have a new dog, [goi]?\u2019(172) EahehothasananeichnnewHund,doggoinsCONF.2.F.\u2018You have a new dog, [goi]?\u2019174(173) EahehothasananeichnnewHund,doggoitsCONF.2.PL.\u2018You have a new dog, [goi]?\u2019The same pattern can be found in Spanish for the DM sabes? (\u2018you know?\u2019),whose form varies depending on who the Addressee is: in (174) the speakeris talking to someone close and uses the second person singular informal tu\u00b4,whereas in (175) the speaker addresses her interlocutor using the formal ustedform (marked here as 2SG.formal for easy comparison, but formally coincideswith the 3SG form):(174) To a friend\/non-formal interlocutor:YomeyaalreadynoNEGse\u00b4knowque\u00b4whatpensar,thinksabes?know.2SG.informal\u2018I don\u2019t know what to think, you know?\u2019(175) To someone higher in status:YomeyaalreadynoNEGse\u00b4knowque\u00b4whatpensar,thinksabe?know.2SG.formal\u2018I don\u2019t know what to think, you know?\u2019This type of evidence has been used to propose an expansion of the clausalstructure in the left periphery to include discourse and speech act-level concepts,mainly the speech act participant roles of the Speaker and the Addressee\u2013in thecases above of allocutive agreement, the sentence final particles would then agreewith the Addressee. These type of proposals are ways of modelling the pragmatics-syntax interface. In the next section I introduce the main proposals that I will adoptfor the syntactic analysis of DM si and DM no.1755.1 Tools for modelling the pragmatics-syntaxinterfaceThe left periphery is considered to be the interface between a clause and the rest ofthe intra-sentential, as well as inter-sentential discourse. Some analyses proposea relatively simple structure in the left periphery: Emonds (2004) proposes an a-categorial projection dominating IP, which he calls \u2018Discourse Shell\u2019 to explainthe transformations that only happen in root clauses (such as topicalization). Onthe other hand, other analyses propose to enrich and extend the syntactic struc-ture to explain the same type of phenomena. Much of the literature on DMs hasadopted this second tradition, and so will I in this chapter.5.1.1 Exploded CP: CForce>CFiniteThe Complementizer Phrase is the projection where the sentence connects to theoutside. This \u2018outside\u2019 can be another clause, or the rest of the discourse. A simplestructure of the CP is given in the tree 5.1:CPC\u2019C IPpropositionFigure 5.1: The Complementizer Phrase.Rizzi (1997) proposes that the CP has a highly articulated structure, with sev-eral functional projections; of these projections, at least two are obligatory: For-ceP, which hosts the complementizer, and FinitenessP, which is a direct link to thefinite\/non-finite feature of the verb in IP. Between those two there are other projec-tions that act as landing sites of topics and foci. Later additions to this structure,176include an Interrogative Phrase (which would host interrogative complementiz-ers) and Modifier Phrase (which hosts discourse-oriented adverbials, not shownin the following structure) (Rizzi, 2001). This articulated structure is representedin (5.2).ForcePForce\u2019Forceo TopicPhola Topic\u2019Topo IntPInt\u2019Into FocusPFocus\u2019Foco FinPhola Fin\u2019Fino IPholaholaholaFigure 5.2: Rizzi\u2019s split CP.Originally, ForceP was meant to host the complementizer, but recent stud-ies have argued that this part of the structure is more complex than originallythought (Coniglio and Zegrean, 2012; Lam, 2014; Corr, 2016; Thoma, 2016;Heim, 2019a). Coniglio and Zegrean (2012) propose to divide ForceP into twoprojections (ForceILLOCUTIONARY and ForceCLAUSE-TYPE): the higher projectionof Force would encode the illocutionary force of the clause, whereas the lower177projection could encode clause typing information. This proposal was originallymeant to explain indirect speech acts such as commands expressed by means ofinterrogatives. Other analyses, to which we turn now, go even further and splitForceP into syntactic projections representing pragmatic roles.5.1.2 Beyond CP:ResponseP>GroundAddresseeP>GroundSpeakerPRecent proposals in the syntactic literature extend the CP structure\u2013in particular,there are proposals that extend the structure of the clause beyond the CP, takinginspiration from Ross\u2019 Performative Hypothesis (Ross, 1970). These proposalsclaim that there is syntactic evidence to motivate the inclusion of at least pragmaticroles (Speaker, Addressee) in the syntactic structure (Speas and Tenny, 2003;Haegeman and Hill, 2013; Wiltschko and Heim, 2016). In addition, Wiltschko(2016); Wiltschko and Heim (2016); Wiltschko (2017) propose to expand thestructure further by adding a layer that encodes what the Speaker wants the Ad-dressee to do with the utterance\u2013this layer instantiates the idea of Call-on-the-Addressee proposed by Beyssade and Marandin (2006). This extended structureis represented in Figure 5.3.178ResponsePResponse\u2019Response GroundAddresseePGroundAddressee\u2019GroundAddressee GroundSpeakerPGroundSpeaker\u2019GroundSpeaker CPpropositionFigure 5.3: The Response and Ground layers (Wiltschko and Heim, 2016).The crucial idea behind this proposal is that we can incorporate the Speakerand the Addressee in the syntactic structure indirectly via their grounds: thesegrounds represent the belief sets of each interlocutor in the Speaker\/Hearer role.As we have seen in Chapter 4, being able to assign a belief or preference is key inthe analysis of the contribution of DM no.5.1.2.1 First steps: The Speech Act PhraseSpeas and Tenny (2003) decompose illocutionary force into different syntacticconfigurations of three elements: the Speaker, the Hearer (in this dissertation: theAddressee), and the Utterance content. The different possible configurations ofthese three components yield different sentence types. In essence, the structurecoding speech act information should mirror the structure of vP. This structure in179called Speech Act Phrase (SAP).In a declarative sentence, the SAP is arranged as in Figure 5.4, with theSPEAKER at the top of the syntactic structure as the specifier of the SAP. TheHEARER is the lower as the complement of the lower SA head.sap(SPEAKER) sasa sa*(UTTERANCE CONTENT) sa*sa* (HEARER)Figure 5.4: Structure of Speech Act Phrase (SAP) in declarative sentencesIn an interrogative sentence the pragmatic roles and the utterance contentwould be in a different configuration, as shown in Figure 5.5. In this situation,the HEARER role moves from the complement to the specifier position. This con-figuration mirrors the promotion of the indirect object in a dative shift operation:180sap(SPEAKER) sasa sa*(HEARER) sa*(UTTERANCE CONTENT) sa*sa* tFigure 5.5: Structure of Speech Act Phrase (SAP) in interrogative sentencesThis account, although inspired by Ross\u2019 Performative Hypothesis, differsfrom it in that the sentence is still mono-clausal (as opposed to bi-clausal).Based on evidence from West Flemish and Romanian particles and vocatives,Haegeman and Hill (2013) and Haegeman (2014) propose to refine the Speech ActPhrase originally proposed by Speas and Tenny (2003). They propose to articulateit in two projections: SAP1 and SAP2. SAP1 is the highest projection, whichis said to activate the speaker-hearer relation and to establish the discourse set-up. On the other hand, SAPlow is speaker-centered and encodes speaker attitude;it does not activate the relationship between the speech participants but insteadconsolidates and qualifies an already established relationship (Haegeman, 2014;Corr, 2016).Why these two projections? Haegeman (2014) focuses on two West FlemishDMs, ne\u00b4 and we`. They differ in terms of their position: whereas ne\u00b4 can occurutterance initially and finally, we` can only occur utterance finally. According to181Haegeman, ne\u00b4 can either focus the attention of the Addressee on the utterancewhen it is in initial position) or \u2018transfer\u2019 the utterance to them (when it is in finalposition). On the other hand, we` \u201cqualifies the speaker-hearer relation, establish-ing that the speaker has the authority (with respect to the hearer as well as withrespect to the content of the utterance) to make the utterance.\u201d (Haegeman, 2014,p. 124). Importantly, ne\u00b4 and we` can co-occur, and when they do, there is a strictorder, shown in (176) (where both DMs are modifying the clause Men artikel isgedoan, \u2018My paper is finished\u2019).(176) a. Ne\u00b4, men artikel is gedoan we`.ne\u00b4 my paper is done we`b. *We`, men artikel is gedoan ne\u00b4.c. Men artikel is gedoan we` ne\u00b4.d. *Men artikel is gedoan ne\u00b4 we`.e. *Ne\u00b4 we`\/* we` ne\u00b4 men artikel is gedoan.Based on these ordering restrictions and the specific selectional properties ofthese DMs (see Haegeman (2014) for detailed argumentation), Haegeman pro-poses that ne\u00b4 heads the higher projection SA1 and that we` heads the lower pro-jection SA2. These two projections are higher than, and independent of, the CPstructure (whose highest projection is Force, as illustrated in (177)):(177) [SA1 ne\u00b4 [SA2 we` [Force . . .I will employ several of the diagnostics that Haegeman (2014) and Corr (2016)use in order to develop a syntactic analysis of the Spanish DMs si and no. Specifi-cally, I will use the relative ordering of the DMs with respect to other lexical items(such as discourse activating particles) to assign a position to DM no within thisinteractional layer.1825.1.2.2 Adding a Call-on-the-Addressee: The Grounding and RespondingLayerThe final framework that I will discuss, and the one that I will adopt to analyze thesyntactic position of DM no, is the one that follows the Universal Spine Hypoth-esis (Wiltschko, 2014; Thoma, 2016; Wiltschko and Heim, 2016). The extensionof this framework also encodes pragmatic information in a functional projectionabove CP (Thoma, 2016); specifically, it decomposes illocutionary force into twosyntactic projections: the grounding layer (GroundP) and the response layer (Re-sponseP) (Wiltschko, 2017). GroundP is based on the idea of grounding (Clarkand Schaefer, 1989; Clark and Brennan, 1991), which is the expression of the in-terlocutors\u2019 attitudes towards the proposition. GroundP is divided into two projec-tions: the Speaker Ground and the Addressee Ground. These grounds are meantto syntactically represent the interlocutors via their belief states, which in a wayrepresent the DCs. The Response layer encodes what Beyssade and Marandin(2006) refer to as the \u2018Call-on-the-Addressee\u2019, which in a nutshell tells the Ad-dressee what the Speaker wants them to do with the proposition. This extendedsyntactic spine is illustrated in Figure 5.6:183ResponsePResponse\u2019Response GroundAPGroundA\u2019GroundA GroundSPGroundS\u2019GroundS CP(propositional structure)Figure 5.6: The Response and Grounding layers (Wiltschko and Heim,2016; Wiltschko, 2017)The Addressee\u2019s Ground projection, and not the Speaker\u2019s, is the highest struc-turally: evidence for this comes from the relative order of sentence final particlesin Cantonese (Lam, 2014). Note that in this framework, rising or falling intonationcan be the head of these projections: Heim (2019a) in fact proposes that intona-tion can encode both commitment (linked to the grounding layer) and engagement(linked to the response layer).1The interactional layer presented in Figure 5.6 forms part of a bigger conceptu-alization of the relationship between syntax and function developed by Wiltschko(2014): the syntactic structure (or spine) is made up of abstract functional cate-gories (\u03ba), which are hierarchically organized. These categories convey different1Commitment and engagement are the two building blocks into which Heim decomposes illo-cutionary force.184abstract grammatical functions, such as grounding, linking, anchoring, introduc-ing a point of view, or classifying an event.Figure 5.7: The extension of the hierarchy of functional categories (\u03ba) inthe Universal Spine proposed by Thoma (2016) based on Wiltschko(2014).What is crucial here (especially for the analysis of DM no) is that if we identifythe function of a lexical item, we can determine its position within this hierarchi-cally organized spine. What is more: different lexical items (Units of Language,or UoLs in Wiltschko (2014); Thoma (2016)) can associate with different cate-gories, allowing us to explain, for instance, the \u201cmany jobs\u201d of the particles si andno beyond their use as DMs.5.1.3 Integrating DM si and DM no into the clause structureI adopt a hybrid analysis, keeping the Force projection as part of CP, as well asadopting the extension of the clause structure proposed by Wiltschko and Heim(2016) to include the grounding and response layers above CP.I analyze DM si as a force operator heading ForceP, as shown in Figure 5.8.As such, it encodes the illocutionary force assigned to the sentence: specifically, it185encodes superassertive force. I have distinguished between asserting p (proposalto update the CG with p) from superasserting p (marking that p is already in theCG).2ForcePForce\u2019ForcesiTopicPhola Topic\u2019Top FocusPFocus\u2019Foc FinPhola Fin\u2019Fin IPholaholaholaFigure 5.8: Position of DM si in ForceP.The difference between a bare declarative (linked to an assertion) and a si-marked declarative (linked to a superassertion) is illustrated in the following ex-amples (178) and (179). The difference between (178) and (179) is that for Con-text 1, A can just assert the proposition p=\u2018Adriana lives in Vancouver\u2019 becausethe proposition introduces new information that the Addressee presumably doesnot know yet; in this situation, A can propose to update the CG with p (=assert p,shown in (178a)). In Context 2, however, p is assumed to be already known, and2I will propose in Chapter 6 that the other Spanish complementizer, que, can also appear inthis position in its reportative function, which also only selects declarative sentences (Demonteand Ferna\u00b4ndez Soriano, 2014).186hence it would be pragmatically odd to propose to update the CG with it; in thiscontext, what A in (179) is doing is super-asserting p, marking that it is alreadyin the CG (as shown in (179b)). The same utterance would not be felicitous ina context like Context 1, where the Speaker cannot make the assumption that theproposition is already mutually shared:(178) Context 1: During a conference, A is introducing Adriana to their ac-quaintance, B, and mentions that she lives in Canada. They talk about theweather, and B mentions concerns about a blizzard in Montreal. A says:a. Adriana vive en Vancouver.\u2018Adriana lives in Vancouver.\u2019b. #Si Adriana vive en Vancouver.\u2018[Si] Adriana lives in Vancouver.\u2019(179) Context 2: While watching the news, Adriana\u2019s family sees that there is ablizzard in Montreal. Adriana\u2019s brothers express concern for her, and herfather says:a. ?Adriana vive en Vancouver.\u2018Adriana lives in Vancouver.\u2019b. Si Adriana vive en Vancouver.\u2018[Si] Adriana lives in Vancouver.\u2019As can be seen in the tree in Figure 5.8, I position DM si still in the CP layer:at the topmost projection, but still part of the CP. One of the reasons for this comesfrom prosody, namely the integration of DM si in the sentence\u2019s intonational con-tour.3 As opposed to other more peripheral elements that may or may not forman intonational unit with the sentence they modify (such as DM no, which I willdiscuss in section 5.4), DM si cannot be prosodically separated from the sentence3I thank Scott Schwenter and other members at the audience of Hispanic Linguistics Sympo-sium 2019 for bringing this to my attention.187it appears in. The prosodic integration of DM si can be seen in Figure 5.9 (fromElvira Garc\u0131\u00b4a (2016)) for the sentence Pero si en Me\u00b4rida se merienda me\u00b4dula \u2018But[si] in Me\u00b4rida one eats bone marrow\u2019. DM si forms one syllable with the follow-ing proposition en \u2018in\u2019, shown by the transcription of this sequence (\u2018si en\u2019) as[sjien].Figure 5.9: Figure from Elvira Garc\u0131\u00b4a (2016, 140; 4.11) shows the syntacticintegration of DM si in the the clause it introduces.DM si cannot be separated from the clause it appears in by an intonationalbreak\u2013as it is shown in (180) it is not possible to divide the sentence Si tengoun gato (\u2018[si] I have a cat\u2019) into two separate intonational phrases (InP). Corr(2016) uses this same diagnostic to distinguish \u2018speech-act heads\u2019 (above CP)from \u2018evaluative\u2019 heads (which she also positions above ForceP):(180) a. (Si tengo un gato)InPb. *(Si)InP (tengo un gato)InP188Crucially, the prosody of DM si differs from that of DM no: although no canbe integrated into the intonational contour of the anchor, it is also possible tohave a break between the anchor and the DM (Cabedo Nebot, 2013). I take thisdifference, and specifically the obligatory prosodic integration of DM si with theclause it introduces, to show that DM si is more syntactically integrated into thestructure of the clause than DMs like no. Thus, while DM si is part of CP, thisdoes not imply that DM no is outside of the clause. Rather, DM no is still part ofthe clause, but it sits at an (even) higher projection.I propose that DM no is located at this interactional layer above CP. Specif-ically, I propose to analyze DM no in a similar fashion to Canadian English eh(Wiltschko and Heim, 2016): DM no takes the whole anchor as its argument, andrequests from the Addressee a confirmation of the validity of the anchor. This isshown in the tree in Figure 5.10.ResponsePResponse\u2019Response GroundAddresseePGroundAddressee\u2019GroundAddresseenoGroundSpeakerPGroundSpeaker\u2019GroundSpeaker CPpropositionFigure 5.10: Proposed syntactic position for DM no.There is syntactic evidence in favour of this structure, such as allocutive agree-189ment (Oyharc\u00b8abal, 1993; Miyagawa, 2013), and ordering restrictions (Haegemanand Hill, 2013). As described in Chapter 3 and in Chapter 4, an important differ-ence between DMs si and no is the role assigned to the Addressee: in the case ofno-tagged clauses, the Addressee is given a pivotal role in the development of theconversation. As we have seen, no-tagged clauses can be used to gauge the in-formation state of the Addressee. This is illustrated with the difference in felicitybetween (181) and (182):(181) Esta\u00b4sarecansada,tired#(no?)no\u2018You are tired, [no?]\u2019(182) Estoyamcansada,tired(#no?)no\u2018I am tired, [no?]\u2019The structure in Figure 5.10 allows us to give a syntactic account of the factthat DM no modifies (and selects) a full speech act. As we have seen in Chapter 4,DM no is not restricted by the type of sentence it modifies (declarative, interrog-ative, imperative, or exclamative), but by the type of speech act (it can modifyassertions, but not commissives, for example).5.2 (Almost) leftmost and CP-internal: DM siWe focus first on DM si. In section 5.2.1 I describe the syntactic distribution ofsi-clauses in terms of the diagnostics developed by Corr (2016) for illocutionaryoperators. In section 5.2.2 I focus on the specific position of si within the leftperiphery, with respect to other elements that precede and follow the DM.1905.2.1 Syntactic distribution: DM si is CP-internal5.2.1.1 DM si selects for sentence typeIn the literature we find claims that DM si can only occur in declarative sentences.This generalization seems to hold true, as illustrated in (183), except for a possibleexception: rhetorical questions. The acceptance of DM si followed by a rhetoricalquestion seems to be marginal, and although examples can be found in naturallyoccurring data, this question will be have to left for further research. For therest of this dissertation, I will assume that DM si introduces mostly, if not only,declarative clauses.(183) a. DECLARATIVESisiPedroPedroesishijosonu\u00b4nico.unique\u2018[Si] Pedro is an only child.\u2019b. EXCLAMATIVE*Sisique\u00b4howbonitoprettyes!is\u2018[Si] how pretty it is!\u2019c. INTERROGATIVE*Sisido\u00b4ndewhereesta\u00b4?is\u2018[Si] where is it?\u2019d. IMPERATIVE*Sisiven!come\u2018[Si] come!\u2019191e. RHETORICAL WH Q(?)SiSisinceramentesincerelyaPREPquie\u00b4nwholethemalcanzareachconwith$301,000?$301,000\u2018[Si] honestly who can live on $301,000?\u2019 (Twitter)f. RHETORICAL POLAR-Q?Sisique\u00b4whatteyoucrees,thinkquethattengohavetodoallelthed\u0131\u00b4a?day\u2018[Si] what do you think, that I have all day?\u2019The judgements in (183) are summarized in Table 5.1. A direct link betweenDM si and sentence type is a diagnostic in favour of si being a C head, accordingto Corr (2016).Table 5.1: Summary of DM si clause-typing properties.Sentence type DM siDeclarative 3Polar Interrogative 7Wh- Interrogative 7Rhetorical polar Q ?Rhetorical wh Q ?Imperative 7Exclamative 75.2.1.2 Si-clauses are root CPsDM si-marked clauses are root CPs, and cannot be embedded. This follows fromDM si being a speech act operator, which are claimed to not be embeddable:although the embedding of full speech acts is a logical possibility, Krifka (2014)argues that it rarely happens. In (184) we see that si-clauses cannot be embedded:192(184) *HehavedichosaidquethatsisiPedroPedroesishijosonu\u00b4nico.unique\u2018I said that [si] Pedro is an only child.\u2019But (184) becomes well-formed if we change the prosody: if the intonationalcontour is changed to a continuation contour, the sentence is grammatical\u2013this ismarked in (186) with suspension dots. This change, however, marks the differencebetween a si-clause and a suspended conditional (Elvira-Garc\u0131\u00b4a et al., 2017):(185) HehavedichosaidquethatsiifPedroPedroesishijosonu\u00b4nicounique. . .\u2018I said that if Pedro is an only child . . . \u2019In this case, [CPsi Pedro es hijo u\u00b4nico] can be embedded because it alreadyis a dependent clause with an elided main clause: this si introduces a dependentconditional clause, as illustrated by the following possible continuation:(186) HehavedichosaidquethatsiifPedroPedroesishijosonu\u00b4nico...unique...noNEGtendra\u00b4will.havesobrinosnephews\u2018I said that if Pedro is an only child... he will never have nephews. \u2019The impossibility of embedding a si-clause is consistent with treating DM sias an illocutionary operator, sitting high up in the left periphery.45.2.1.3 Si-clauses cannot be negatedIllocutionary operators cannot be negated. As the literature notes, the idea of anegated speech act is hard to conceive, and as Corr (2016, p.122) summarizes:4In Chapter 6 I mention the possibility of analyzing si-clauses as elliptical clauses, but as ellip-tical adverbial clauses and not as elliptical complement clauses. It is possible that the differencein prosody could indicate different types of anaphora that support ellipsis (De\u00b4chaine, p.c.).193\u201cone may either PERFORM an act (do, or say, something), or fail to PERFORM anact (do, or here, say nothing), but what one cannot do is NOT PERFORM an act\u201d.In order to see this, let us start with the negation at the propositional level. In(187), the adjunct clause [CP si tengo un perro] (if I have a dog) is negated:(187) Context: I\u2019m applying for a cat-friendly apartment. My mother asks mewhether landlords will accept my application if I include that I have a pet,and I say:Siiftengohaveunagato,catnoNEGsiiftengohaveunaperro.dog\u2018If I have a cat, not if I have a dog.\u2019DM si-clauses, however cannot be negated.(188) Context: An acquaintance is asking A about how to take care of their newdog since they have one, but A has a cat, not a dog. A replies:Sisitengohaveunagato,catnoNEG(*si)sitengohaveunaperro.dog\u2018[Si] I have a cat, not a dog.\u20195.2.1.4 Si-clauses cannot be conjoined nor disjoinedAlthough it would not be surprising for an illocutionary operator to be able toappear in conjoined clauses, si-clauses cannot appear in such cases, as (189):(189) [Context: A and B are talking. B mentions that A cannot know what it\u2019slike to have a dog because she only has a cat (when in reality she hasboth). A replies:]Sisitengohaveunagatocatyand(*si)(si)tengohaveunaperro!dog\u2018[Si] I have a cat, not a dog.\u2019194Although some illocutionary operators can be conjoined, typically there is nodisjunction at the speech act level.5 This is due to the fact that disjunction is an op-eration on propositions, and not illocutions, since the disjunctive operator wouldbe understood as cancelling the illoctutionary force of the sentence (Corr, 2016,p.124). In (190), combining two si-clauses with a disjunctive operator results inan ungrammatical sentence.(190) [Context: A and B are talking. B mentions how much she will miss A,even though they will meet the next day and, if that is not possible, onSaturday. A replies:]SisinosRECvemosseeman\u02dcanatomorrowoor(*si)([si])nosRECvemosseeelthefinde!weekend\u2018[Si] we will see each other tomorrow or the weekend!\u2019We have just discussed the syntactic distribution of si-CPs, but we have notyet focused on the syntactic position of DM si itself. This is the topic of the nextsection.5.2.2 Linearization: DM si is not leftmostIn order to locate DM si within the syntactic structure, I will first describe whatare the reasons to suspect that it occupies a very high position within the leftperiphery. Namely, I will discuss its position with respect to other high elementssuch as topics ( 5.2.2.1.1) and speaker-oriented adverbs (5.2.2.1.2), and show thatDM si precedes them all. However, in section 5.2.2.2 we will see that there arecertain elements that precede DM si: vocatives and discourse activating particles.5In the pragmatic framework introduced in Cohen and Krifka (2014) disjunction, althoughmechanically possible, does not yield a defined commitment space unless the speech acts are twodenegations. Krifka (2001) does mention the possibility of disjoined questions, but no other speechact seems to be disjoinable (besides denegations, as previously mentioned.1955.2.2.1 What DM si precedes5.2.2.1.1 TopicsThere are certain elements of the left periphery that can be preceded by DM si,one of them being topics such as overt subject pronouns (191) and left dislocatedelements (192). Typically, topics are considered to be high up in the C domainstructure\u2013in a cartographic approach (such as Rizzi\u2019s), they occupy a position inthe highest TopicP, above Focus but below ForceP. Overt subject pronouns havebeen analyzed as contrastive topic markers in Spanish (Mayol, 2010). In (191),the overt subject marker yo \u2018I\u2019 cannot precede DM si in a grammatical sentence:(191) OVERT SUBJECT PRONOUNa. SiSiyoIyaalreadyloitsab\u0131\u00b4a.knew\u2018[Si] I already knew it.\u2019b. *Yo si ya lo sab\u0131\u00b4a.DM si also precedes topicalized elements such as los libros (\u2018the books\u2019) inclitic left dislocation constructions like the one in (192).(192) TOPICALIZATION: CLITIC LEFT DISLOCATIONa. Sisilosthelibrosbooksyaalreadylosthemhehavele\u0131\u00b4do.read\u2018[Si] as for the books, I already read them.\u2019b. *los libros si ya los he le\u0131\u00b4do.Given these distributional facts, we can conclude that DM si is positionedabove topics in the syntactic structure.196ForcePForce\u2019ForcesiTopicPhola Topic\u2019TopoyoFinPhola Fin\u2019Fino IPya lo sab\u0131\u00b4aFigure 5.11: Position of DM si in ForceP with respect to topics.5.2.2.1.2 Discourse-oriented adverbsOther left-peripheral elements that cannot precede DM si are discourse-orientedadverbs (using the term in Speas and Tenny (2003), otherwise known as speaker-oriented adverbs). These adverbs are typically subdivided into three types: fromthe highest to the lowest in terms of hierarchy (shown in (193)), they are speechact adverbials, evaluative adverbials, and evidential adverbials (Jackendoff, 1972;Cinque, 1999; Ernst, 2009; Maienborn and Scha\u00a8fer, 2011):(193) Speech Act > Evaluative > Evidential\u2018honestly\u2019 > \u2018unfortunately\u2019 > \u2018evidently\u2019In Rizzi (1997), these adverbs occupy a position labelled Mod(ifier)P, betweenForceP and FiniteP, and below FocusP:(194) ForceP > FocusP >ModP > FinP > IP197In (195), evidentemente \u2018evidently\u2019 is an evidential adverb and comments \u201conthe quality of the evidence supporting the truth of the proposition, or on the man-ner in which the individual has come to learn of that fact or truth\u201d (Speas andTenny, 2003, p.329). Here again, DM si needs to precede evidentemente:(195) EVIDENTIAL ADVERBS6a. SiSievidentementeevidentlyesta\u00b4iscasado.married\u2018[Si] evidently he is married.\u2019b. *Evidentemente si esta\u00b4 casado.ForcePForce\u2019ForcesiModP(Evid)hola Mod(Evid)\u2019Mod(Evid)evidentementeFinPhola Fin\u2019Fin IPesta\u00b4 casadoFigure 5.12: Position of DM si in Force with respect to evidential discourseadverbs.In (196), desafortunadamente \u2018unfortunately\u2019 is an evaluative adverb and ex-6Note that (b) would be fine if there was an intonational break between evidentemente and si.198presses a judgment (or evaluation) of proposition. DM si precedes also this typeof speaker-oriented adverb.(196) EVALUATIVE ADVERBSa. SiSidesafortunadamenteunfortunatelyesta\u00b4iscasado.married\u2018[Si] unfortunately he is married.\u2019b. *Desafortunadamente si esta\u00b4 casado.ForcePForce\u2019ForcesiModP(Eval)hola Mod(Eval)\u2019Mod(Eval)oevidentementeFinPhola Fin\u2019Fino IPya lo sab\u0131\u00b4aFigure 5.13: Position of DM si in ForceILL with respect to evaluativediscourse adverbs.The last type of speaker-oriented adverb are speech act level adverbs such as\u2018honestly\u2019 or \u2018sincerely\u2019. They express the speaker\u2019s attitude towards the contentor the form of the proposition. In (197), DM si is shown to to precede since-ramente \u2018honestly\u2019; the reverse ordering, although dispreferred, is not entirely199ungrammatical. In fact, it is grammatical if the adverb \u2018sinceramente\u2019 and the restof the sentence are separated by an intonational break, as in (197d) :(197) SPEECH ACT ADVERBS7a. SiSisinceramentesincerelyaPREPquienwholethemalcanzareachconwith$301,000?$301,000\u2018[Si] honestly who can live on $301,000?\u2019 (Twitter)b. UshhushhsisisinceramentesincerelyyoIsab\u0131\u00b4aknewquethatDuqueDuqueerawasunainu\u00b4til,useless(. . . )\u2018 Ushh [si] honestly I already knew that Duque was useless.\u2019 (Twitter)c. SisisinceramentesincerelynoNEGtengohavenineitheridea.idea\u2018[Si] honestly I have no clue.\u2019d. ?Sinceramente si no tengo ni idea.e. Sinceramente, si no tengo ni idea.The data in (197e) can be explained through the interactional layer proposedby Wiltschko and Heim (2016): the adverb sinceramente \u2018honestly\u2019 would be po-sitioned in GroundSpeaker, as in the tree in (5.14): the adverb can be separated fromthe sentence by an intonational break, and its meaning fits the function associatedwith this layer:87It is important to note that in the case of dependent si, the si-Adv order is not allowed:(197) a. SinceramentesincerelysiifyoItuvierahadunabebe\u00b4babynoNEGlehimpondr\u0131\u00b4aputnada.anything.\u2018Honestly, if I had baby I wouldn\u2019t put anything on them.\u2019(Twitter)b. *si sinceramente yo tuviera un bebe no le pondr\u0131\u00b4a nada.8See Emonds (2004) on the structural significance of an intonational break200GroundSpeakerPGroundSpeaker\u2019GroundSpeakersinceramente,ForcePForce\u2019ForcesiIPno tengo ni ideaFigure 5.14: Position of DM si and the speech act level adverbsinceramente \u2018honestly\u2019.We can assume the same structure for (197a-c) as for the other two types ofdiscourse adverbs, as represented in the tree in (5.15). However, this analysisis suspiciously simple, since one would expect the speech act adverb to scopeover the entire speech act, including DM si. The data and judgements for thisco-occurrence are not particularly clear, and future work will help fine-tune theanalysis presented here. For now, the general conclusion is that DM si precedesmost discourse-oriented adverbs.201ForcePForce\u2019ForcesiModP(SA)hola Mod(SA)\u2019Mod(SA)oevidentementeFinPhola Fin\u2019Fino IPya lo sab\u0131\u00b4aFigure 5.15: Position of DM si in Force with respect to speech actdiscourse adverbs with an intonation break (marked by a comma).5.2.2.2 What DM si followsFunctionally, the role of DM si seems to match that of elements that have beenanalyzed as lexicalizing layers above CP that encode utterance-level information(Corr, 2016; Thoma, 2016; Corr, 2018). However, I have argued that based onits obligatory intonational integration within the sentence, and its direct selectionof sentence type, DM si is still within the CP domain. We therefore expect in-teractional discourse particles to precede DM si, and this prediction is borne out.There are three types of elements that are predicted to precede DM si: vocatives,Speaker-oriented discourse particles, and Addressee-oriented discourse particles.5.2.2.2.1 Addressee-oriented discourse activating particlesAddressee-oriented discourse activating particles such as oye (\u2018hey\u2019) typicallyprecede vocatives (Corr, 2016), and can be found in utterance initial position tograb the attention of the Addressee (Haegeman and Hill, 2013; Haegeman, 2014).202These particles are analyzed as being the highest heads of a syntactic layer thatrepresents speech act participants, namely in GroundAddressee:9GroundAddresseePGroundAddressee\u2019GroundAddresseeoyeGroundSpeakerPGroundSpeaker\u2019GroundSpeaker ForcePForce\u2019ForcesiCPpropositionFigure 5.16: Proposed syntactic positions for DM si and theSpeaker-oriented discourse activating particle oye.Given that we have proposed to position DM si below the interactional layer,we expect it to follow Addressee-oriented discourse activating particles. Thisprediction is borne out, as exemplified in (199):(199) a. OyelistensisiyoInoNEGhehavedichosaidnada.anything\u2018Hey, [si] I didn\u2019t say anything.\u2019b. #Si oye yo no he dicho nada.9This DM could also be analyzed as instantiating the Response later (Wiltschko, p.c.).2035.2.2.2.2 Speaker-oriented discourse activating particlesWe also expect DM si to appear below Speaker-oriented discourse activatingparticles, such as ay or anda (lit. \u2018walk\u2019 in the imperative form, but typically usedas an exclamation of surprise). These types of particles sit at the lowest projectionbeyond CP, and convey the Speaker\u2019s attitude towards the utterance (Haegemanand Hill (2013) and Wiltschko and Heim (2016) on other languages, and Corr(2016) on Spanish and other Ibero-Romance languages). This prediction is alsoborne out, as we see in the examples in (200). The structure is given in the tree inFigure 5.17.(200) a. Ayaysisiesisverdad!truth\u2018Oh, [si] it\u2019s true.\u2019b. *Si ay es verdad!GroundAddresseePGroundAddressee\u2019GroundAddressee GroundSpeakerPGroundSpeaker\u2019GroundSpeakerayForcePForce\u2019ForcesiCPes verdadFigure 5.17: Proposed syntactic positions for DM si and theSpeaker-oriented discourse activating particle ay.2045.2.2.2.3 VocativesVocatives in Italian and Romanian have been said to occupy one of the highestpositions within the left periphery (Moro, 2003; Hill, 2007, 2013). However, Corr(2016) notes that vocatives in Ibero-Romance languages cannot appear in absoluteinitial position when multiple DMs are used, as (201) shows:(201) a. Oye,listenguapo,good-lookingmira,DMyoInoNEGsoybe.1SGdelof.thePP,PPeh?DM\u2018Hey, mate, look, I\u2019m not a Tory, okay?\u2019 (Corr, 2016, p.77; ex.106-7)b. #Guapo oye mira, yo no soy del PP, eh?Given this restriction, Corr (2016) proposes that vocatives are merged in thespecifier position of what I have analyzed here as Force (her EvalP):GroundPGround\u2019Ground ForcePVocP Force\u2019Force IPyo no he dicho nadaFigure 5.18: Position of vocatives within the left periphery.We expect therefore to find DM si below vocatives, a prediction that is borneout as is shown in (202):205(202) a. Adriana,Adrianasisiesisjueves.thursday\u2018Adriana, [si] it\u2019s Thursday.\u2019b. *Si Adriana es jueves.The (simplified) structure for (202) can be seen in the tree in Figure 5.19:GroundPGround\u2019Ground ForcePVocPAdriana,Force\u2019ForcesiIPes juevesFigure 5.19: Position of DM si and vocatives.To sum up, I propose that DM si is an illocutionary operator that occupiesthe highest position in the CP domain, which I have labelled as ForceP, belowthe interactional layer above CP proposed in the recent literature (GroundP). Thepredicted relative order with respect to high CP elements (such as topicalizedelements) and low interactional-level elements (such as vocatives) is borne out.5.3 DiscussionIn the first half of the chapter, I proposed an analysis of si-CPs and DM si: si-CPsare root clauses that link the proposition they denote to the previous intra- andinter-sentential discourse; DM si is the head of the highest C projection, Force,and encodes the super-assertive illocutionary force of the sentence. I based this206proposal on diagnostics for illocutionary operators, as well as the relative positionof the particle with respect to other elements in the left periphery.This analysis is able to capture the multifunctionality of the particle si: aswe saw in Chapter 2, si can be used to introduce interrogative (203), and (root)exclamative clauses (204):10(203) NoNEGse\u00b4knowsiwhetheresisverdad.truthINTERROGATIVE\u2018I don\u2019t know whether it\u2019s true.\u2019(204) Sisisera\u00b4be.FUTfa\u00b4cil!easyEXCLAMATIVE\u2018Isn\u2019t it easy!\u2019Hernanz (2012) treats the particle si in root exclamatives as the head of FocusP,and interrogative si as the head of the Interrogative projection proposed by Rizzi(2001).11 She also positions conditional si as the head of ForceP, in parallel to thedeclarative finite complementizer que (we will come back to the complementizerparadigm in Chapter 6).10We will return to conditionals in Chapter 6.11The position of the interrogative complementizer in this specific projection was originallyproposed by Rizzi (2001). For a discussion on the evidence of exclamative si in FocusP, seeHernanz (2012).207ForcePForce\u2019Forcediscourse siInterrogativePhola Interrogative\u2019Interrogativeointerrogative siFocusPFocus\u2019Focoexclamative siFinPhola Fin\u2019Fino IPholaholaholaFigure 5.20: Position of si in different projections of the CP.A question that remains to be addressed is what the core lexical entry is thatis compatible with all these positions.12 Rodr\u0131\u00b4guez Ramalle (2011) proposes thatwhat DM si and interrogative si have in common is that they contrast alternatives:interrogative si contrasts polar alternatives, whereas DM si contrasts evidential al-ternatives. If we apply this same idea to our analysis, DM si seems to contrast twofelicity conditions: the one presented by the previous contribution to discourse,and the one it introduces. This would also explains why DM si cannot be used inout-of-the-blue contexts\u2013namely, because it needs something to react to. In (205)we see that Speaker A cannot use a si-clause in an out-of-the-blue context, but she12Another option is to assume that all these instances of the particle si do not share a corecommon lexical entry.208could use a bare clause:(205) Context: A enters a meeting room and sits besides a co-worker, B. A saysto B: (#Si)sihacedoesunad\u0131\u00b4adayprecioso!precious\u2018[#si] it\u2019s such a nice day!\u2019DM si needs something to contrast with and, therefore, it cannot be used out-of-the-blue.5.4 Rightmost and CP-external: DM noWe turn now to our second protagonist: DM no. Recall from section 5.1 that Iadopt an analysis of DM no that extends the syntactic structure of the clause to in-clude a projection above CP that encodes pragmatic roles syntactically (Haegemanand Hill, 2013; Wiltschko and Heim, 2016; Yang and Wiltschko, 2016; Jamieson,2019). Specifically, I adopt the Universal Spine Hypothesis structure proposedby Wiltschko (2014) and extended in Lam (2014); Thoma (2016) and Wiltschkoand Heim (2016) to account for interactional phenomena. I adopt the structureillustrated in the tree in Figure 5.21, where DM no is the head of GroundAddresseePgiven that it is used to confirm the Addressee\u2019s belief or intention. Given thepremises of the framework I am using, the function of a given lexical item can beused to know its position in the syntactic structure: since the interpretation of agiven lexical item depends on the place of association of that item in the syntacticstructure, we can reverse this link and use the interpretation of a lexical item incontext c to know its syntactic position.13In addition to these theory-based predictions, I will give independent syn-tactic evidence for DM no\u2019s position, namely selectional properties and its po-13This predicts that the same lexical item might have different interpretations depending onwhere it is associated in the spine. See Thoma (2016) for an analysis of Miesbach Bavariandiscourse particles that exploits this characteristic of the framework, as well as De\u00b4chaine et al.(2014) on noun class prefixes in Shona, as well as De\u00b4chaine and Wiltschko (2017) on reflexives.209sition relative to the rest of the sentence. The final analysis can be seen in Fig-ure 5.21. The final order is the result of moving the CP to the specifier position ofGroundAddresseeP (Munaro and Poletto, 2002), which will be further discussed insection 5.4.2:ResponsePResponse\u2019Response GroundAddresseePGroundAddressee\u2019GroundAddresseenoGroundSpeakerPGroundSpeaker\u2019GroundSpeaker CPpropositionFigure 5.21: Proposed syntactic positions for DM no.As we discussed in section 5.1, it has been noted in the literature that theuse of certain tags is sensitive to the relationship between speech act participants.Such effects have been attested for other DMs like Canadian eh and (American)English huh and right (Wiltschko and Heim, 2016). What Wiltschko and Heim\u2019sexamples in (206) and (207) show is that whereas huh and right can only be usedto confirm the truth of a proposition p, Canadian eh can also be used to ask forthe confirmation of the Speaker\u2019s assumption that the Addressee knows that theproposition is true.(206) John knows that Mary would like to have a new dog. He has not seen herin a long time. And he keeps wondering whether she got a new dog. One210day he runs into her while she\u2019s walking a new puppy. John utters:You have a new dog, eh\/huh\/right?= Confirm that p is true(Wiltschko and Heim, 2016, (example (6))(207) Mary is walking her new dog when she runs into John. She is expectingthat he would congratulate her on the new dog, but he\u2019s not mentioningit. She isn\u2019t sure anymore whether he actually realizes that she has a newdog. So she utters:I have a new dog, eh\/*huh\/*right?= Confirm that you know that p is true(Wiltschko and Heim, 2016, (example(7))We see the effects of speech act participant roles on the licensing of DM noespecially with predicates of personal taste and internal state predicates, as wesaw in Chapter 4. DM no is felicitous if added to a taste predicate whose judge isthe Addressee (second person), as in (208). Importantly, without the addition ofDM no, the sentence would be infelicitous or pragmatically invalid:14(208) Teyouhahasgustado,liked#(no?)no\u2018You liked it, [no?]\u2019In (208), it is pragmatically infelicitous to use a no-tagged clause with a tastepredicate with the Addressee as the judge: the Addressee is the sole judge ofwhether they like something, and asserting that the Addressee likes something isusually considered face-threatening.14Both this and the following point follow from the interrogative nature of the use of DM noand other confirmational markers, marked by the rising intonation. Regardless of the reason forthis characteristic, it still stands that the use of DM no is restricted in these cases.211The opposite happens when we exchange a second person for a first personjudge. If the Speaker is expressing her own judgements, it is not accepted toask the Addressee for confirmation of their commitment, since it is not needed toground the proposition:(209) Memehahasgustado,liked(#no?)no\u2018I liked it, [no?]\u2019The same effect is seen when the proposition denotes other perspectival predi-cates: internal state predicates. An example of an internal state predicate is \u201cto betired\u201d (in Spanish, estar cansado), exemplified in (210):(210) Esta\u00b4sarecansado,tired#(no?)no\u2018You are tired, [no?]\u2019(211) Estoyamcansada,tired(#no?)no\u2018I am tired, [no?]\u2019In (210), it would be pragmatically odd not to use the tag when confirming thatsomeone is tired: even if the Speaker in (210) knows the Addressee well and seesthat they have huge dark circles underneath their eyes and are yawning constantly,it would be a social faux-pas to dryly assert that the Addressee is tired. The use ofthe tag, however, allows this conversational move because of what it does: it asksthe Addressee whether making the statement \u2018You are tired\u2019 is valid.This is one of the distributional properties that the literature has used to pro-pose a syntactic position for this type of tag and particle (Munaro and Poletto,2002; Haegeman and Hill, 2013; Thoma, 2016; Wiltschko and Heim, 2016). Other212diagnostics, which will be discussed below in reference to DM no, are sentencetype restriction (discussed in 5.4.1.2), and scope of the particle (discussed in 5.4.1.1).5.4.1 Syntactic distribution: DM no is CP-externalFirst, let us focus on the syntactic distribution of DM no, applying the diagnosticswe used for DM si: what it restricts (section 5.4.1.1) and what it (does not) select(section 5.4.1.2), as well as its link to negation (section 5.4.1.3) and disjunction(section 5.4.1.4).5.4.1.1 DM no restricts root CPsAnalyzing DM no as the head of a projection above CP allows us to capture an-other important characteristic of this DM: it takes scope over the whole sentencewith which it occurs; that is, it asks for confirmation of the root CP, and cannot askfor confirmation of an embedded CP within the matrix clause. This is shown in(212) by the felicity of different continuations to the no-marked utterance. An an-swer that addresses the matrix clause (Adriana knowing p) is felicitious, whereasan answer that addresses whether the embedded p is true or not is not:(212) AdrianaAdrianasabeknowsquethatJuanJuanviene,comeno?no\u2018Adriana knows that Juan is coming, [no?]\u2019a. A: S\u0131\u00b4,yesloitsabe.knows\u2018Yes, she knows it.\u2019b. B: #S\u0131\u00b4,yesJuanjuanviene.comes\u2018Yes, Juan is coming.\u2019213Similarly, the speaker in (213) is not asking for confirmation of the proposi-tion denoted by the noun complement clause it follows (le mintieron \u2018they lied tohim\u2019), but of the fact that the person in question has to admit that he had beenlied to. This is why the confirmation of p=\u2018they lied to him\u2019 is not a felicitouscontinuation, but the confirmation of p=\u2018he has to admit that they lied to him\u2019 is,as illustrated in (213a and b):(213) Tienehasquethatreconoceradmitelthehechofactdeofquethatlehimmintieron,liedno?[no]\u2018He has to admit the fact that they lied to him, [no?]\u2019a. A: #S\u0131\u00b4,yesesisverdad,truthlehimmintieron.lied\u2018Yes, it\u2019s true, they lied to him.\u2019b. B. S\u0131\u00b4,yesloittienehasquethatreconoceradmitparaforpasarpasspa\u00b4gina.page\u2018Yes, he has to admit it to be able to move on.\u2019We can conclude, then, that DM no scopes over the entire root CP and cannotscope over dependent clauses. This suggests that it has to be positioned high inthe structure in order to be able to c-command the whole matrix CP.5.4.1.2 DM no does not select for clause-typeAs we saw in Chapter 2, in terms of selectional properties, DM no does not seem toselect for any particular sentence type: it can anchor to declaratives, interrogatives,imperatives, and exclamatives (214):214(214) a. Hoytodayesismie\u00b4rcoles,wednesdayno?noDECLARATIVE\u2018Today is Wednesday, [no?]\u2019b. Do\u00b4ndewhereestara\u00b4will.beahora,knowno?noINTERROGATIVE\u2018Where could he be now, [no?]\u2019c. Estudiastudyunaratitowhile.DIMma\u00b4s,moreno?noIMPERATIVE\u2018Study for a little bit longer, [no?]\u2019d. Que\u00b4whatalegr\u0131\u00b4ajoyloitdeoftuyourhermano,brotherno?noEXCLAMATIVE\u2018How great the news about your brother, [no?]\u2019However, as was also discussed in Chapter 2, DM no selects a particular typeof speech act. Specifically, it does not anchor to three types of speech acts: ques-tions, commissives, or expressives (215). Under the assumption that illocutionaryforce is encoded in ForceP, this means that DM no has to be located in a positionwhere it can select ForceP\u2013such as GroundP.15(215) a. # Co\u00b4mo te llamas, no? QUESTIONWhat\u2019s your name, [no?]\u2019b. # Te lo juro, no? COMMISSIVE\u2019I swear, [no?]\u201915This is a simplification of how illocutionary force works. Even if we assume that illocutionaryforce is a result of a combination of various factors (sentence type, context of use, sometimes theperson of the subject (as is the case with promises)), the point is that DM no should be able toselect the clause at the point where it can perform the intended speech act.215c. # Muchas gracias, no? EXPRESSIVE\u2019Thank you very much, [no?]\u2019It is important to bear in mind that the judgements in (215) refer to the use ofthe anchor as a wh-question, a promise, and a thanking act, respectively: as wehave seen in Chapter 2 and Chapter 4, the sequences themselves are felicitous ifthey do not express these specific acts and are used in a rhetorical way.5.4.1.3 No-tagged utterances are undefined relative to negationOne of the properties of illocutionary operators at the CP-level was that they can-not be negated. This test, however, does not seem to be applicable to no-tagged ut-terances, or at least not as straightfowardly as applied for DM si in section 5.2.1.3.If we try to negate DM no, as in (216), the resulting utterance is nonsensical:(216) Estudiastudyunaratitowhile.DIMma\u00b4s,more*?noNEGno?no\u2018Study for a little bit longer, not [no?]\u20195.4.1.4 No-tagged utterances are undefined relative toconjunction\/disjunctionAnother one of the properties of illocutionary operators at the CP-level is that theycannot be disjoined. Just as with negation, this test does not seem to be applicablein such a straightforward way as with DM si:(217) ?Estudiastudyunaratitowhile.DIMma\u00b4s,moreno?nooorvetegoatodomir,sleepno?no\u2018Study for a little bit longer [no?] or go to bed, [no?]\u2019The issue here is that (217) above is felicitous if we use DM no with a non-confirmational function (as described in Chapter 2). But it would be infelicitous,216or at least pragmatically odd, to use confirmational no (with a rising intonation)in both positions.The only way to have disjunction in this case would be to have DM no afterboth assertions, as in (218):(218) Estudiastudyunaratitowhile.DIMma\u00b4smoreoorvetegoatodomir,sleepno?no\u2018Study for a little bit longer or go to bed, [no?]\u20195.4.2 Linearization: DM no is turn-finalDM no has an \u2018inert\u2019 syntactic life: it always appears after the anchor.16 DM nocannot appear sentence initially or sentence internally (220), this is true of DM no(with rising intonation), but phatic no can appear sentence internally.17(219) a. AdrianaAdrianatienehasunagato,catno?[no]\u2018Adriana has a cat, [no?]\u2019b. *No Adriana tiene un gato?c. *Adriana no tiene un gato?This is expected given the proposed analysis: in principle, we expect DMs thatare located in the interactional layer to appear in the peripheries, either the left or16Montan\u02dcez Mesas (2015) identifies a number of co-occurences of DM no and another DM, eh,as well as other discourse marking forms like mm or o sea \u2018I mean\u2019 as the anchor. Regarding theco-occurrence of DM no and DM eh, in her example DM eh follows DM no, and she claims thatthe combination of both reinforces its apellative meaning (similar to the Call-on-the-Addressee).However, both no and eh can be used as stand-alone utterances: no in its negative response markeruse, and eh to ask for a repetition of the previous utterance (Montan\u02dcez Mesas, 2015). It is thereforenot clear that these are sequences of DMs, as much as an anchor and a DM. The other examplesshe cites are also ambiguous in this sense.17See Chapter 2 for examples of non-confirmational uses of DM no.217the right periphery. In this case, we can derive the linear order of (220) followingthe proposal by Munaro and Poletto (2002) for Italian markers: the anchor (in thiscase, [Adriana has a cat]) moves to the specifier of the projection headed by DMno. 18Figure 5.22: Movement of the CP to the Specifier of GroundP, followingMunaro and Poletto (2002).This movement is caused by the need to meet a prosodic requirement, namelythe high rising intonation on DM no. Prosodically-required movement is attestedin other syntactic movements in Spanish, for example movement of the subject tosentence final position to receive nuclear stress (which falls on the last element ofthe clause) (Zubizarreta, 1998; Bonet, 2019).A potential issue with this proposal is that it does not predict the fact thatparentheticals can appear between the anchor CP and the DM:(220) AdrianaAdrianatienehasunagato,catquethatyoIsepa,knowno?[no]\u2018Adriana has a cat, as far as I know, [no?]\u2019This is not expected in a specifier-head configuration: a parenthetical couldnot intervene between a head and its specifier, but it could intervene between two18Again for ease of reading, I conflate both Grounding projections in Figure 5.22.218maximal projections (Munaro and Poletto, 2002). I propose a possible solutionfor this in the following section.5.5 DiscussionA possible solution for the challenge of linearization and DM no is to appealto a bi-clausal analysis of no-tagged clauses. A bi-clausal analysis of similarconstructions (such as English tag questions) has been proposed in the literature:English tag questions are composed of two clauses, an assertion and a question,where the VP of the second clause is elided because it is given (Sailor, 2009):(221) [CP . . . ], [CP isn\u2019t it VP]Let\u2019s compare what these two types of analyses would look like for DM no:one where we assume a bi-clausal structure with ellipsis of the given information,and a mono-clausal structure like the one we have proposed until now:(222) BI-CLAUSAL: [CP . . . ], [CP no CP]MONO-CLAUSAL: [CP ANCHOR [GroundP no [ t ] ] ]One of the reasons I didn\u2019t propose a bi-clausal analysis is that, unlike forEnglish tag questions, it is not as straightforward to assume that there is ellipsisof given information: in (224a) the VP in the second clause is easily recoverable,but not in the Spanish equivalent in (224b).1919Garc\u0131\u00b4a Vizca\u0131\u00b4no (2005) notes that in most cases, DM no can be swapped for a more \u201cfullyfleshed\u201d question. When the anchor is a non-evaluative assertion, the DM can be changed for thelonger sequence no es cierto? \u2018isn\u2019t it true?\u2019. When the anchor is an evaluative assertion, on theother hand, the DM can be swapped for either no es cierto? or no crees? \u2018don\u2019t you think?\u2019. Whenthe anchor is an imperative, the longer question has to be no crees? \u2018don\u2019t you think?\u2019 (this isillustrated in (223):(223) a. Man\u02dcanatomorrowesismartes,TuesdaynoNEGesiscierto?\/#nocertain\/NEGcrees?think.2SG\u2018Tomorrow is Tuesday, isn\u2019t it true?\u2019219(224) a. George washes his hands before dinner, does he not [wash his handsbefore dinner]? (Sailor, 2009)b. Jorge se lava las manos, no #[se lava las manos]?jorge REFL washes the hands no REFL washes the handsAnother reason is the restriction that the DM imposes on the speech act typeof the anchor: although it is not a clause typing restriction, it still can be used asevidence for the DM selecting its anchor.There are other two additional reasons to propose a mono-clausal analysis:one based on prosody, and another based on scope. I proposed that DM no (a)forms a prosodic unit with its anchor, although it is not integrated in the CP (sec-tion 2.2.1.2), and (b) sits high up in the left periphery, structurally above its anchor,which explains the scope of the DM (section 5.1.2).The relationship between DM no and prosody is tricky: as Cabedo Nebot(2013) notes, in the past there has been the assumption of a strict correlation be-tween DMs in general and prosodic breaks (in the sense that DMs are delimited byprosodic breaks). Cabedo Nebot (2013, p.204) mentions that orthographic con-ventions may also impose an expectation of a prosodic break between the anchorand a DM, an expectation that oftentimes is not met in casual conversation. Infact, in his corpus study, DM no forms an autonomous intonational unit only in16.60% of the cases: in other words, in the majority of the cases the DM is in-cluded as a peripheral element of the utterance, but does not form a prosodic unitof its own. The DM may not be fully integrated in the anchor itself, but it doesb. Que\u00b4whatbonito,prettynoNEGcrees?think.2SG\/\/noNEGesiscierto?certain\u2018How pretty, don\u2019t you think?\u2019c. Estudiastudyunapocobitma\u00b4s,morenoNEGcrees?\/#nothink.2SG\/NEGesiscierto?certain\u2018Study a bit longer, don\u2019t you think?\u2019220form a unit with it. Since I have assumed that prosodic integration is evidence ofsyntactic integration, I take this characteristic of DM no to favour a mono-clausalanalysis.This unit is not only prosodic: DM no takes scope over the whole anchor.Under the assumption that an expression \u03b1 can only scope over an expression \u03b2if \u03b1 c-commands \u03b2 , this would mean that the DM no c-commands the anchor CP(Hinzen, 2006; Thoma, 2016). A mono-clausal structure like (225) explains thisscope relation:(225) [GroundP no [CP ANCHOR]]A bi-clausal analysis, however, has the advantage of giving the ordering of theanchor and the DM \u2018for free\u2019, as well as explaining why we can find parentheticalsbetween the anchor and the DM: since the anchor and the DM form different max-imal projections, it is possible for a parenthetical structure to intervene betweenthem.If we assume that the ellipsis analysis can be applied to the structure like (226),we can still maintain an analysis where the DM no lexicalizes GroundAddressee, butin this case the GroundP of the second (interrogative) clause (see Jamieson (2019)for this analysis for Scots varieties):(226) AdrianaAdrianatienehasunagato,catno[no]es cierto?is\u2018Adriana has a cat, isn\u2019t it true?\u2019(227) BI-CLAUSAL: [GroundP . . . ], [GroundP no CP]There are advantages to both types of analysis, as we have just seen. A morein-depth study of this possibility, as well as a more nuanced understanding of therelationship between DM no, prosody, and parentheticals remains to be conducted.221Chapter 6Conclusions and outlookThe truth is an elephant described by three blind men. The first mantouches the tail and says it\u2019s a rope. The second man feels the roughleg and says it\u2019s a tree. The third man feels the trunk and says it\u2019s asnake.\u2014 Indian parable, quoted in S2E9 of The X Files.6.1 SummaryThe very broad aim of this dissertation was to study how two Peninsular SpanishDMs (si and no) are used, and what that can tell us about the different levels atwhich conversations operates. These two DMs aim to restore or maximize epis-temic alignment between Speaker and Addressee. On the one hand, DM si is an-alyzed as an illocutionary operator that marks that a proposition is already knownby all interlocutors, hence forcing an alignment within the Common Ground. Asfor DM no, it is used to align both beliefs and outcomes among interlocutors.These two DMs show that the Speaker has two versions of the CG in the Dia-logue Board (one representing what they know, and the other representing whatthey believe the Addressee to know), and that they are able to negotiate contextualupdates with the Addressee.I insisted on separating the contribution of the DM per se from the contribution222of the clause it appears with with respect to the rest of discourse. Si-clauses rejectthe conversational update proposed by the previous discourse move because afelicity condition is not met. A no-tagged clause is used when the Speaker wantsto confirm a previous epistemic or outcome bias.Another aim of this dissertation was to explore recent proposals to encodespeech act-level information in the syntactic structure. I proposed that DM si is anillocutionary operator situated within the CP domain, but as high as ForceP, basedon prosodic and syntactic evidence. I also proposed that DM no occupies an evenhigher position, outside of CP but still within the same clause. I follow recentanalyses in the syntactic literature in calling this higher structure the Groundinglayer: within it, DM no occupies the layer that represents the Addressee\u2019s infor-mation state (as perceived by the Speaker).One conclusion of this type of study is the complexity of multiple interactions.For example, in the case of DM si we had to account for the contribution of theprevious discourse, the si-marked clause as a whole within the discourse, and thecontribution of the DM si itself to the sentence it appears in. In the case of DMno, we had to distinguish its pragmatic effect from its semantic contribution onthe dialogue (checking whether the Addressee validates the anchor), as well askeeping track of the content of the anchor itself.1There are still many questions and further areas of research: in 6.2, I compareDM si and DM no to similar DMs, and try applying a similar analysis to the onedeveloped in this dissertation. Specifically, I compare two uses of illocutionaryoperators in Spanish: DM si to reportative que (6.2.1). In section 6.2.1.1 I contex-tualize the discussion about root complementizers within a bigger research ques-tion, namely that of insubordination\u2013the use of sentences marked as subordinatesas main clauses (Evans and Watanabe, 2016). I also compare the confirmationalDMs no and eh in 6.2.2: DM eh seems to thrive in exactly the contexts of use thatDM no tries to avoid, and this fact can be explained by the analysis presented in1To all this, we have to add the importance of intonation, which I have largely omitted in thisdissertation.223this dissertation. I also describe the possibility of cross-dialectal variation in theuse of these DMs in section 6.3. Finally, another interesting avenue of further re-search is the role that DMs may play in showing illocutionary differences betweendifferent speech act types: in section 6.4 I discuss recent studies that focus on thedifferent illocutionary effects of evaluative and non-evaluative assertions, a differ-ence that has played a role in the analysis of DM no in this dissertation\u2013namely,that the use of the DM disambiguates the possible interpretation of evaluative as-sertions as objective or subjective interpretations.6.2 Comparison with other markersI have alluded to other complementizers and tags that seem to function in similarways to DM si and DM no. In this section I propose a way to extend the analysispresented in this dissertation to these other markers.6.2.1 The complementizers si and queIn this dissertation I have described DM si as expressing superassertive force. Bythis I mean that, contrary to just proposing an update of the CG with p (what anassertion would do), a si-marked clause conveys that the proposition it denotes isalready part of the CG. We see this in (228): DM si cannot be used in a situationwhere it is virtually impossible for p (p=\u2018A is vegetarian.\u2019) to be in the CG:(228) Context: A and B are in a speed dating event. As an ice-breaker, B asks:B: Prefieres carne o pescado?\u2018Do you prefer meat or fish?\u2019A: (#Si)sisoyamvegetariana.vegetarian\u2018([Si]) I am vegetarian.\u2019224Interestingly, que \u2018that\u2019 can be used with the opposite effect: instead of su-perasserting a proposition, it merely presents it (as in Faller (2002); De\u00b4chaineet al. (2017))\u2013this is usually referred to as reportative que, since this particle, justlike si, seems to have many jobs.2 The use of que as a reportative is illustrated in(229), and its use as a declarative complementizer is illustrated in (230):(229) Context: An employee just had a meeting with her boss, who told her sheand her co-worker B shouldn\u2019t come to work tomorrow. She goes back toB\u2019s desk and says:Tu\u00b4,youquethatnoNEGvengamoscome.1SG.SUBJ!mann\u02dcana.tomorrow\u2018Hey you, [que] we shouldn\u2019t come tomorrow.\u2019(230) Memehahasdichosaidquethatesisvegetariana.vegetarian\u2018She has told me that she is vegetarian.\u2019The literature has noted that the reportative use of the complementizer que inroot clauses seems to introduce presentational force (Etxepare, 2010; Demonteand Ferna\u00b4ndez Soriano, 2014; Corr, 2016). As such, a proposition introduced byreportative que can be followed by a direct negation of the Speaker\u2019s belief in thatsame proposition, as exemplified in (231):(231) Context: A has been told by C that Hermann has had a son, but C is aknown liar and has announced the birth many times before. A tells B:2See, among others, Demonte and Ferna\u00b4ndez Soriano (2009, 2014); Corr (2016); Villa-Garc\u0131\u00b4a(2019).225Oye,hearquethatHermannHermannhahastenidohadunahijo,sonaunquealthoughhastauntilquethatnoNEGmemellamecallsnoNEGmemeloitcreo.believe\u2018Hey, [que] Hermann has had a son, although until he calls me I won\u2019tbelieve it.\u2019As we saw in Chapter 3, DM si contributes almost the opposite meaning, as wecan see in (232): the proposition DM si introduces cannot be not believed, sinceas Schwenter (2016b) points out, it is obviously true to the Speaker. Betweenthose two extremes is an unmodified, \u201cdefault\u201d assertion: a declarative that isnot modified by any root complementizer merely proposes an update of the CG(Farkas and Bruce, 2010).(232) Context: B tells A that now that Hermann is done with his PhD and hasno responsibilities he can travel the world and party all night. A tells B:Oye,listensisiHermannHermannhahastenidohadunahijo,son#aunquealthoughhastauntilquethatnoNEGmemellamecallsnoNEGmemeloitcreo.believe\u2018Hey, [si] Hermann has a son, although until he calls me I won\u2019t believeit.\u2019I propose that there is a three-way distinction in how a proposition can en-ter the conversation, and that this is reflected in how ForceP is marked: it caneither be empty, in which case the proposition will be asserted; it can be filledby si, in which case the proposition will be superasserted; or it can be filled byque, in which case the proposition will be presented. These three possibilities areillustrated in the tree in Figure 6.1:226ForcePForce\u2019Forcesi\/que\/ \/0IPpropositionFigure 6.1: Position of discourse si and reportative que.The picture is, of course, much more complicated than this: si-clauses inPeninsular Spanish also include exclamative clauses (discussed in Chapter 2 sec-tion 2.1.1.2); que can also appear in exclamative clauses. Both discourse andexclamative functions of the complementizers are illustrated in (233) and (234)respectively.3(233) a. DISCURSIVE FUNCTIONOye,hearquethateltheBarc\u00b8aBarc\u00b8ahahasganadowonlatheliga.league\u2018Hey, [que] Barc\u00b8a has won the Liga.\u2019 (Etxepare, 2010, p. 604)b. Sisiyaalreadyloitse\u00b4.know\u2018[Si] I already know.\u2019(234) a. EXCLAMATIVE FUNCTIONQuethatvienecomeseltheCoco!Boogeyman\u2018The Boogeyman is coming!\u20193 Corr (2016) identifies three different functions of que in root clauses; however, Demonte andFerna\u00b4ndez Soriano (2014) identify one of these as \u2019echoic\u2019, and not truly root clauses.227b. Sisisera\u00b4sbe.2SG.FUTtonto!dumb\u2018[Si] aren\u2019t you dumb!\u2019However, the picture can get even more complicated. Since que appears inmany positions within the CP structure, there seems to be a parallelism betweenthe uses of these two complementizers in root clauses versus non-root clauses, aspresented in Figure 6.2.Figure 6.2: Parallelism between the two Spanish complementizers.Digging deeper into the similarities and differences between si and que, andthe paradigm they could establish with a null complementizer (as in the case of\u201cbare\u201d assertion), remains for future research.6.2.1.1 Insubordination and DM siThe use of what are traditionally viewed as subordinating complementizers inroot clauses has been labelled insubordination (Evans et al., 2007). Examplesfrom various Indo-European languages are given in (235) (these are examples ofthe phenomenon of insubordination in general, that is, they include examples ofnon-finite clauses used as main clauses). This phenomenon describes a clause thatis either syntactically or morphologically marked for subordination but appears inroot environments. It is not rare cross-linguistically, and as we have seen, si is byno means the only complementizer in Spanish that can appear in root clauses.228(235) a. YoIirgo.INFatolathefiesta?partyJama\u00b4s!neverSPANISH\u2018Me go to the party? Never!\u2019b. Alzarsi,get.up.INF-REFLporci,pigsavetehavecapito?understoodITALIAN\u2018Get up, pigs, understand?\u2019c. BeimbyEintrittentrytieflowverneigen!bow.INFGERMAN\u2018Bow low on entering!\u2019d. John go to the movies? No way man! ENGLISHInsubordination is not restricted to Romance and Germanic languages. It hasbeen observed in Farsi (Ghomeshi, 2013), Tlingit (Cable, 2009, 2011), Japanese(Evans and Watanabe, 2016), Yupik (Mithun, 2008), Nisga\u2019a (Tarpent, 1991),Plains Cree (Cook, 2008), and Navajo (Mithun, 2008), to name a few (for moreexamples see Evans and Watanabe (2016)). In the following examples, matrixclauses bear morphosyntactic markers (boldfaced) that would otherwise have onlybeen observed in subordinate clauses:(236) FARSI (Ghomeshi, 2013, p.3)h\u00e6vaweatherb\u00e6dbadnistNEG.be.3SGxub-egood-be.3SGinjaherekePRTxeyliveryxub-egood-be.3SG\u2018The weather\u2019s not bad; it\u2019s nice; here, it\u2019s really nice.\u2019(237) NAVAJO (Mithun, 2008)T\u2019a\u00b4h e\u00b4\u0131\u00b4 a\u00b4a\u00b4di sida\u00b4 le\u00b4\u0131\u00b4. T\u2019o\u00b4o\u00b4 baaya\u00b4niizii-o\u2018Hont\u2019ah was still sitting back there. I got so embarrassed-SUB.\u2019229When it comes to the relationship between si-clauses and insubordination,there is a possible grammaticalization process: si-clauses may have originatedas adjunct conditional clauses. These conditional clauses could have, in fact, beenpremise conditionals (following Haegeman (2003)). This type of conditional hasbeen given many other names in the literature, but they essentially put a conditionat the level of the speech act as opposed to the propositional level (Haegeman callsthis other type of conditionals event conditionals). This contrast between differentlevels of modification was also noted by Sweetser (1990), who distinguishes be-tween content, epistemic, and speech act conditionals (the latter would correspondto the premise conditionals):(238) a. If Paris is the capital of France, two is an even number. CONTENTb. If she\u2019s divorced, she\u2019s been married. EPISTEMICc. If I may say so, that\u2019s a crazy idea. SPEECH ACTIt is possible that DM si introduced at some point a speech act conditional, andthat over time the main clause (or CP1 in Figure 6.3) was elided. In fact, a possibleparaphrase of si- clauses is along the lines of \u201cHow can you assert\/ask\/order that,if p is the condition that holds\u201d (cf. Schwenter (1998)).Figure 6.3: si-clause as peripheral adverbial.Such a construction would target the felicity condition of the previous dis-course move (something that I have proposed DM si does) while re-asserting thestatus of p as an element of the CG. Since the main clause is easily recoverable230from the context and refers back to it, it would make sense that this part wouldbe elided. This would also account for the fact that si-clauses can be analyzedas being discourse anaphoric (Cook, 2008): as we have seen in this dissertation,si-clauses refer back to something mentioned in the previous discourse. Cook(2008) described a similar type of clauses in Plains Cree, which she calls dis-course anaphoric, and proposes to schematize them as in (239).(239) XPantecedent . . . CPanaphorIn the case of si-clauses this XPantecedent would be a CP, as we have seen inFigure 6.3. More data, and especially more diachronic data, is needed to flesh outthis hypothesis.6.2.2 The DMs no and ehMuch of the research done on DMs (in Spanish and in other languages) has com-pared different DMs in order to define their contribution to the utterance. DMno has been mainly compared to another Spanish DM, eh, as well as other tagssuch as \u2018verdad?\u2019 or \u2018cierto?\u2019 (both translatable as \u2018is(n\u2019t) it true?\u2019) (Zorraquinoet al., 1999; Stenstro\u00a8m, 2005; Garc\u0131\u00b4a Vizca\u0131\u00b4no, 2005; Montan\u02dcez Mesas, 2007;Rodr\u0131\u00b4guez Mun\u02dcoz, 2009; Roggia, 2012; Montan\u02dcez Mesas, 2015).In general, eh seems to thrive in exactly the contexts of use where no is infelici-tious: commissives (240), directives (241), and expressives (242) (Garc\u0131\u00b4a Vizca\u0131\u00b4no,2005; Montan\u02dcez Mesas, 2007).(240) a. Teyouloitprometo,promiseeh?eh\u2018I promise, [eh]?\u2019b. #Teyouloitprometo,promiseno?no\u2018I promise, [no]?\u2019231(241) a. Ca\u00b4llateshut.upya,alreadyeh?eh\u2018Shut up already, [eh]?\u2019b. #Ca\u00b4llateshut.upya,alreadyno?no\u2018Shut up already, [no]?\u2019(242) a. Muchasmanygracias,thankseh?eh\u2018Thank you very much, [eh]?\u2019b. #Muchasmanygracias,thanksno?no\u2018Thank you very much, [no]?\u2019The literature on eh mentions that this particle modifies the utterance as op-posed to the proposition itself (Cuenca and Castella` (1995), cited in Montan\u02dcez Mesas(2015)).4 This is exactly what I claimed for DM no, but whereas no tables a pro-posal for update, eh tables the update itself. This is modelled in Figure 6.4 for thesentence in (243) (command + eh), where the outcome o=\u2018Study more\u2019 is addeddirectly to the shared TDL:(243) a. Estudiastudyma\u00b4s,moreeh?eh\u2018Study more, [eh]?\u20194Garc\u0131\u00b4a Vizca\u0131\u00b4no (2005) claims, however, that eh serves more often as a propositional strength-ener (refuerzo proposicional) than as an illocutionary strengthener (refuerzo ilocutivo). The datashe uses as illustration of the former function can also be analyzed as examples of illocutionarystrengthening (Garc\u0131\u00b4a Vizca\u0131\u00b4no, 2005, 94: 5).232ACURRENT PROJECTEDDCATDLACGTDL oBo=\u2018Study more\u2019BCURRENT PROJECTEDDCBTDLB oo=\u2018Study more\u2019CGTDL oBo=\u2018Study more\u2019KA2TABLEKA2Figure 6.4: Speaker utters a modified command with DM eh.Pragmatically, the effect of eh as opposed to no is to increase the degreeof strength of the illocutionary point (Garc\u0131\u00b4a Vizca\u0131\u00b4no, 2005; Montan\u02dcez Mesas,2015). I formalize this in the same way I formalized the mitigating effect of no:the use of eh increases the degree of strength of the command Estudia ma\u00b4s (\u2018Studymore\u2019) in (244), whereas the use of no decreases the strength:(244) a. Estudia ma\u00b4s, eh?ILL= COMMAND(b, p)STRENGTH: +1b. Estudia ma\u00b4s.ILL= COMMAND(b, p)STRENGTH: 0c. Estudia ma\u00b4s, no?ILL= COMMAND(b, p)STRENGTH: -1233In a sense, what DM eh is doing is \u2018doubling-down\u2019 on the anchor: not onlyis the Speaker in (245) making a promise, but she wants the Addressee to reallyknow that that update has taken place.(245) Teyouloitprometo,promiseeh?eh\u2018I promise you, [eh]?\u2019Also with assertions there is a sense of ratification of the update. In (246) theSpeaker is talking about going down to hell for love and not regretting it (becauseshe \u2018came back with two angels\u2019), but insists that in spite of not regretting it, shestill went down to hell:(246) Perobutbajargo-down.INFbaje\u00b4,go-down.1SG.PSTeh?ehBajar,go-down.INFbaje\u00b4.go-down.1SG.PST\u2018But I DID go down, [eh]? I did go down.\u2019 (from a monologue by actressRossy de Palma, used in the song \u2018PRESO (Cap. 6: Clausura)\u2019 by singerRosal\u0131\u00b4a)In the analysis presented here, where the DM is unmistakably linking thewhole speech act in the anchor with the update, this ratifying use is explained:the Speaker deploys DM eh to inhibit the non-acceptance of the anchor by theAddressee.6.3 Cross-dialectal variationIn this dissertation I have focused on Peninsular Spanish si and no: although bothmarkers appear in, as far as I know, all Spanish varieties, the conditions of useare probably different from variety to variety. This is not surprising, since theuse of DMs is sometimes a defining characteristic of certain dialectal varieties of234languages, e.g. the use of the DM eh in Canadian English.5In the Spanish literature as a whole, the DM that has been most prominentlydiscussed is eh (Montan\u02dcez Mesas, 2015). Besides the studies focusing on differ-ent Peninsular varieties, there are studies on the use of eh as a control de contacto\u2018contact check\u2019 in Chilean Spanish (San Mart\u0131\u00b4n Nu\u00b4n\u02dcez, 2011), as a multifunc-tional marker in Dominican Spanish (Roggia, 2012), and as a hesitation markerin Sanjuanero Spanish (Graham, 2013). Regarding no, it has been most oftenstudied as a mitigating device both in Peninsular Spanish and in Mexican Spanish(Fe\u00b4lix-Brasdefer, 2004a), as well as in Puerto Rican, Dominican, and Mexicaninstitutional settings (Flores-Ferra\u00b4n, 2017). A wide cross-varietal comparison ofDM no has not been conducted as far as I know, with the exception of Ucle\u00b4s Ra-mada (2018) which compares the use of mitigating control checks (\u2018marcadoresde control de contacto\u2019) in Peninsular Spanish and Mexican Spanish in interviewsettings. An interesting future research avenue would be to compare Caribbeanvarieties of Spanish with Peninsular varieties, as these two varieties use very dif-ferent prosodic contours to mark polar questions (Beckman et al., 2002).6.4 Illocutionary differences between evaluativeand non-evaluative assertionsThere is a debate in the literature regarding the differences between what I havecalled evaluative and non-evaluative predicates at the lexical level, but as Beltrama(2018) notes, much less has been said of differences at the illocutionary level. Inhis paper, he reports the results of two experiments where participants rated re-sponses to assertions (evaluative and non-evaluative) and polar questions. In thefirst experiment, they rated how confirmation, denial, and silence were judged. Inthe second experiment, the participants rated the naturalness of disagreement witha denial of an assertion (evaluative vs non-evaluative) and a polar question. On thebasis of these experiments, Beltrama concludes that the illocutionary behaviour of5Of course, eh appears in other varieties of English, most notably in New Zealand English(Meyerhoff, 1994; Stubbe and Holmes, 1995; Vine and Marsden, 2016; Schweinberger, 2018).235evaluative assertions is different from that of non-evaluative assertions, and thattheir effect on discourse falls somewhere between the effect of a non-evaluativeassertion and a polar question. Beltrama\u2019s experiments show that denials are ratedas less disruptive by speakers. Specifically regarding the effect of evaluative as-sertions, he claims that evaluative assertions do not result in an update of the CGwith p in the absence of an overt response.The work in this dissertation is further evidence that there is indeed an illo-cutionary difference between evaluative and non-evaluative assertions, shown bythe effect of DM no on evaluative assertions, and its co-occurrence with only onetype of evaluative assertion (namely, those that are ambiguous between a \u2018gen-eral\u2019 reading and a \u2018Speaker-as-judge\u2019 reading). In fact, the use of DM no can beseen as a way of avoiding this last effect (not updating the CG until a response isgiven). The use of DMs, in addition to possible replies, can therefore be used toinvestigate the different illocutionary effects of different types of utterance.236BibliographyBartels, C. (1997). Towards a compositional interpretation of English statementand question intonation. PhD thesis, University of Massachusetts at AmherstAmherst, MA. \u2192 page 140Beckman, M., D\u0131\u00b4az-Campos, M., McGory, J. T., and Morgan, T. A. (2002).Intonation across Spanish, in the tones and break indices framework. Probus,14(1):9\u201336. \u2192 page 235Bello, A. ([1984] (1847)). Grama\u00b4tica de la lengua castellana destinada al uso delos americanos. Madrid: EDAF. \u2192 page 50Beltrama, A. (2018). Subjective assertions are weak: Exploring the illocutionaryprofile of perspective-dependent predicates. In Proceedings of Sinn undBedeutung, volume 22, pages 160\u2013173. \u2192 pages 18, 235Beninca`, P. (2001). The position of topic and focus in the left periphery. InCinque, G. and Salvi, G., editors, Current Studies in Italian Syntax. EssaysOffered to Lorenzo Renzi, pages 39\u201364. \u2192 page 39Bertrand, A., Heim, J., Thoma, S., and Wiltschko, M. (2015). Let\u2019s all besystematic about confirmationals, eh?. Talk at the European Summer School ofLanguage, Logic, and Information. August 3-14. Bolzano, Italy. \u2192 page 69Beyssade, C. (2012). Confirmation requests and biased questions: assertions,questions, both or neither?. Talk at IX Workshop on Formal Linguistics, Rio deJaneiro, 29-30 August; and RALFe, Paris, 29-30 November. \u2192 page 66Beyssade, C. and Marandin, J.-M. (2006). The speech act assignment problemrevisited: Disentangling speaker\u2019s commitment from speaker\u2019s call onaddressee. Empirical Studies in Syntax and Semantics, 6:37\u201368. \u2192 pages14, 31, 83, 147, 151, 160, 178, 183237Biberauer, T., Haegeman, L., and Van Kemenade, A. (2014). Putting our headstogether: Towards a syntax of particles. Studia Linguistica, 68(1):1\u201315. \u2192page 6Blakemore, D. (2001). Discourse and Relevance Theory. In Deborah Schiffrin,D. T. and Hamilton, H. E., editors, The Handbook of Discourse Analysis,chapter 5, pages 100\u2013118. Blackwell, Malden, MA. \u2192 page 7Blakemore, D. (2004). Discourse markers. In Horn, L. and Ward, G., editors, TheHandbook of Pragmatics, pages 221\u2013240. Blackwell, Malden, MA. \u2192 page 7Bonet, E. (2019). Spanish phonology at the syntax interface. In Colina, S. andMart\u0131\u00b4nez-Gil, F., editors, The Routledge Handbook of Spanish Phonology,page 325\u2013340. \u2192 page 218Briz, A. (1998). El espan\u02dcol coloquial en la conversacio\u00b4n. Barcelona: Ariel. \u2192pages 7, 11, 77, 140Bu\u00a8ring, D. and Gunlogson, C. (2000). Aren\u2019t positive and negative polarquestions the same? ms. \u2192 pages 132, 163Cabedo Nebot, A. (2013). Sobre prosodia, marcadores del discurso y unidadesdel discurso en espan\u02dcol: evidencias de un corpus oral esponta\u00b4neo. Onoma\u00b4zein,(28):201\u2013213. \u2192 pages 12, 64, 189, 220Cable, S. (2009). Insubordination in the languages of the Pacific Northwest. Ms.,University of Massachusetts, Amherst. \u2192 page 229Cable, S. (2011). Insubordination in Tlingit: An areal effect. Northwest Journalof Linguistics, 5(1):1\u201338. \u2192 page 229Cardinaletti, A. (2011). German and Italian modal particles and clause structure.The Linguistic Review, 28(4):493\u2013531. \u2192 page 3Castroviejo Miro\u00b4, E. (2006). \u201cWh\u201d-exclamatives in Catalan. PhD thesis,Universitat de Barcelona. \u2192 page 147Cinque, G. (1999). Adverbs and functional heads: A cross-linguistic perspective.New York: Oxford University Press. \u2192 page 197238Clark, H. H. and Brennan, S. E. (1991). Grounding in communication.Perspectives on Socially Shared Cognition, 13(1991):127\u2013149. \u2192 pages7, 19, 183Clark, H. H. and Carlson, T. B. (1981). Context for comprehension. In Long, J.and Baddeley, A., editors, Attention and performance IX, page 313\u2013330. \u2192page 15Clark, H. H. and Marshall, C. R. (1981). Definite knowledge and mutualknowledge. In Joshe, A., Webber, B., and Sag, I., editors, Elements ofdiscourse understanding, pages 10\u201363. \u2192 pages 1, 15Clark, H. H. and Schaefer, E. F. (1989). Contributing to discourse. CognitiveScience, 13(2):259\u2013294. \u2192 pages xiv, 7, 19, 20, 183Cohen, A. and Krifka, M. (2014). Superlative quantifiers and meta-speech acts.Linguistics and Philosophy, 37(1):41\u201390. \u2192 page 195Coniglio, M. and Zegrean, I. (2012). Splitting up Force. Main ClausePhenomena: New horizons, 190:229\u2013256. \u2192 pages 37, 172, 177Cook, C. E. (2008). The syntax and semantics of clause-typing in Plains Cree.PhD thesis, University of British Columbia. \u2192 pages 11, 229, 231Corr, A. (2016). Ibero-Romance and the syntax of the utterance. PhD thesis,University of Cambridge. \u2192 pages31, 39, 40, 78, 156, 172, 177, 181, 182, 188, 190, 192, 193, 195, 202, 204, 205, 225, 227Corr, A. (2018). \u2018Exclamative\u2019and \u2018quotative\u2019 illocutionary complementisers inCatalan, European Portuguese and Spanish. Languages in Contrast,18(1):69\u201398. \u2192 pages 8, 39, 40, 202CREA (2020). Corpus de referencia del espan\u02dcol actual. Real AcademiaEspan\u02dcola. \u2192 page 124Crespo, I. and Ferna\u00b4ndez, R. (2011). Expressing taste in dialogue. In SEMDIAL2011: Proceedings of the 15th Workshop on the Seantics and Pragmatics ofDialogue, pages 84\u201393. Citeseer. \u2192 page 60Crible, L. (2017). Towards an operational category of discourse markers. InFedriani, C. and Sanso`, A., editors, Pragmatic markers, discourse markers and239modal particles: New perspectives, pages 99\u2013124. John Benjamins,Amsterdam. \u2192 page 6Cuenca, M.-J. and Castella`, J.-M. (1995). Una caracteritzacio\u00b4 cognitiva de lespreguntes confirmato`ries (question tags). Caplletra. Revista Internacional deFilologia, (18):65\u201384. \u2192 page 232Czerwionka, L. (2012). Mitigation: The combined effects of imposition andcertitude. Journal of Pragmatics, 44(10):1163\u20131182. \u2192 page 169De\u00b4chaine, R.-M., Cook, C., Muehlbauer, J., and Waldie, R. (2017). (De-)constructing evidentiality. Lingua, 186:21\u201354. \u2192 pages 14, 18, 141, 142, 225De\u00b4chaine, R.-M., Girard, R., Mudzingwa, C., and Wiltschko, M. (2014). Theinternal syntax of shona class prefixes. Language Sciences, 43:18\u201346. \u2192 page209De\u00b4chaine, R.-M. and Wiltschko, M. (2017). A formal typology of reflexives.Studia Linguistica, 71(1-2):60\u2013106. \u2192 page 209Degand, L. (2019). The paradox of discourse markers: Evidence fromproduction under cognitive load. In 6th International Conference onDiscourse Markers in Romance Languages. \u2192 page 4Demonte, V. and Ferna\u00b4ndez Soriano, O. (2009). Force and finiteness in theSpanish complementizer system. Probus, 21(1):23\u201349. \u2192 pages 9, 225Demonte, V. and Ferna\u00b4ndez Soriano, O. (2014). Evidentiality and illocutionaryforce: Spanish matrix que at the syntax-pragmatics interface. In Dufter, A. andA\u00b4lvaro S. Octavio de Toledo, editors, Left sentence peripheries in Spanish:Diachronic, variationist, and typological perspectives, chapter 8, pages217\u2013253. John Benjamins, Amsterdam. \u2192 pages 186, 225, 227Denis, D. (2013). The social meaning of eh in Canadian English. Proceedings ofthe 2013 Canadian Linguistics Association. \u2192 page 40Denis, D., Wiltschko, M., and d\u2019Arcy, A. (2016). Deconstructedmultifunctionality: Confirmational variation in Canadian English through time.Talk presented at DiPVaC3, University of Ottawa, May 4-6. \u2192 pages 65, 79DeRose, K. and Grandy, R. E. (1999). Conditional assertions and \u201cbiscuit\u201dconditionals. Nou\u02c6s, 33(3):405\u2013420. \u2192 page 95240Domaneschi, F., Romero, M., and Braun, B. (2017). Bias in polar questions:Evidence from English and German production experiments. Glossa: ajournal of general linguistics, 2(1). \u2192 page 132Egg, M. (2013). Discourse particles, common ground, and felicity conditions. InGutzmann, D. and Ga\u00a8rtner, H.-M., editors, Beyond Expressives: Explorationsin Use-Conditional Meaning, pages 125\u2013150. Brill, Leiden. \u2192 pages75, 94, 103Elvira Garc\u0131\u00b4a, W. (2016). La prosodia de las construcciones insubordinadasconectivo-argumentativas del espan\u02dcol. PhD thesis, Universitat de Barcelona.\u2192 pages xvii, 188Elvira-Garc\u0131\u00b4a, W., Roseano, P., and Ferna\u00b4ndez-Planas, A. M. (2017). Prosody asa cue for syntactic dependency: Evidence from dependent and independentclauses with subordination marks in Spanish. Journal of Pragmatics,109:29\u201346. \u2192 pages 44, 46, 49, 193Emonds, J. (2004). Unspecified categories as the key to root constructions. InPeripheries, pages 75\u2013120. Springer. \u2192 pages 3, 176, 200Ernst, T. (2009). Speaker-oriented adverbs. Natural Language & LinguisticTheory, 27(3):497\u2013544. \u2192 page 197Etxepare, R. (2010). From hearsay evidentiality to samesaying relations. Lingua,120(3):604\u2013627. \u2192 pages 225, 227Evans, N. et al. (2007). Insubordination and its uses. In Finiteness: Theoreticaland empirical foundations. Oxford University Press. \u2192 page 228Evans, N. and Watanabe, H. (2016). Insubordination. Amsterdam: JohnBenjamins. \u2192 pages 44, 223, 229Faller, M. (2002). Semantics and pragmatics of evidentials in Cuzco Quechua.PhD thesis, Stanford University. \u2192 pages 170, 225Farkas, D. F. and Bruce, K. B. (2010). On reacting to assertions and polarquestions. Journal of semantics, 27(1):81\u2013118. \u2192 pagesxiv, 1, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 29, 78, 80, 82, 136, 226241Farkas, D. F. and Roelofsen, F. (2017). Division of labor in the interpretation ofdeclaratives and interrogatives. Journal of Semantics, 34(2):237\u2013289. \u2192 page13Fe\u00b4lix-Brasdefer, J. C. (2004a). Interlanguage refusals: Linguistic politeness andlength of residence in the target community. Language learning,54(4):587\u2013653. \u2192 pages 11, 169, 235Fe\u00b4lix-Brasdefer, J. C. (2004b). La mitigacio\u00b4n en el discurso oral de mexicanos yaprendices de espan\u02dcol como lengua extranjera. In Pragma\u00b4tica sociocultural:estudios sobre el discurso de cortes\u0131\u00b4a en espan\u02dcol, pages 285\u2013302. Ariel. \u2192pages 11, 169Flores-Ferra\u00b4n, N. (2017). Mitigation and indirectness in Puerto Rican,Dominican, and Mexican Spanish in an institutional discourse setting. InGonza\u00b4lez-Rivera, M., editor, Current Research in Puerto Rican Linguistics,chapter 8, pages 157\u2013183. Routledge, London. \u2192 page 235Fraser, B. (1980). Conversational mitigation. Journal of pragmatics,4(4):341\u2013350. \u2192 page 169Garc\u0131\u00b4a Vizca\u0131\u00b4no, M. J. (2005). El uso de los ape\u00b4ndices modalizadores \u00bfno? y\u00bfeh? en espan\u02dcol peninsular. In Selected Proceedings of the Second Workshopon Spanish Sociolinguistics. Cascadilla Proceedings Project, Somerville, MA,pages 89\u2013101. \u2192 pagesxiii, 11, 12, 62, 63, 65, 67, 68, 72, 73, 78, 128, 150, 219, 231, 232, 233Geluykens, R. (1987). Intonation and speech act type: An experimental approachto rising intonation in queclaratives. Journal of Pragmatics, 11(4):483\u2013494.\u2192 page 140Ghomeshi, J. (2013). The syntax of pragmaticalization. In Proceedings of the2013 Annual Conference of the Canadian Linguistic Association. \u2192 page 229Ginzburg, J. (1996). Dynamics and the semantics of dialogue. In Seligman, J.and Westersta\u02dahl, D., editors, Logic, language and computation. CSLI,Stanford. \u2192 pages 13, 19Ginzburg, J. (2012). The interactive stance. Oxford University Press. \u2192 pages13, 19, 131242Giorgi, A. (2010). About the Speaker: Towards a syntax of indexicality. Oxford:Oxford University Press. \u2192 pages 39, 173Gold, E. (2005). Canadian Eh?: A survey of contemporary use. In Proceedingsof the 2004 Canadian Linguistics Association Annual Conference. \u2192 page 40Gold, E. and Tremblay, M. (2006). Eh? and Hein?: Discourse particles ornational icons? Canadian Journal of Linguistics\/Revue canadienne delinguistique, 51(2-3):247\u2013263. \u2192 page 40Go\u00b4mez Gonza\u00b4lez, M. d. l. A\u00b4. (2014). Canonical tag questions in English,Spanish and Portuguese: A discourse-functional study. Languages inContrast, 14(1):93\u2013126. \u2192 pages xiii, 12, 63, 67, 68, 71, 72, 73, 78, 132, 171Graham, L. A. (2013). Comparing hesitation markers in Sanjuanero Spanish.Dia\u00b4lago de la Lengua, 5:66\u201377. \u2192 page 235Gras, P. and Sansin\u02dcena, M. S. (2017). Exclamatives in the functional typology ofinsubordination: Evidence from complement insubordinate constructions inSpanish. Journal of Pragmatics, 115:21\u201336. \u2192 pages 4, 5Grosz, P. (2010). German doch: An element that triggers a contrastpresupposition. Proceedings of the Chicago Linguistic Society, 46:163\u2013177.\u2192 page 87Gunlogson, C. (2004). True to form: Rising and falling declaratives as questionsin English. New York: Routledge. \u2192 pages 16, 64, 140Gunlogson, C. (2008). A question of commitment. Belgian Journal ofLinguistics, 22(1):101\u2013136. \u2192 pages 13, 14, 15, 16, 25Gutie\u00b4rrez-Rexach, J. (1996). The semantics of exclamatives. Syntax at sunset.UCLA Working Papers in Linguistics, pages 146\u2013162. \u2192 page 147Gutie\u00b4rrez-Rexach, J. (1999). Spanish exclamatives and the interpretation of theleft periphery. In D\u2019hulst, Y., Rooryck, J., and Schroten, J., editors, RomanceLanguages and Linguistic Theory 1999: Selected papers from \u2018GoingRomance\u2019 1999, pages 167\u2013194. John Benjamins, Amsterdam. \u2192 pages46, 47Gutie\u00b4rrez-Rexach, J. (2008). Spanish root exclamatives at the syntax\/semanticsinterface. Catalan Journal of Linguistics, 7(1):117\u2013133. \u2192 page 147243Haegeman, L. (2003). Conditional clauses: External and internal syntax. Mind &Language, 18(4):317\u2013339. \u2192 pages 95, 230Haegeman, L. (2014). West Flemish verb-based discourse markers and thearticulation of the speech act layer. Studia Linguistica, 68(1):116\u2013139. \u2192pages 39, 181, 182, 202Haegeman, L. and Hill, V. (2013). The syntacticization of discourse. Syntax andits limits, 48:370\u2013390. \u2192 pages 3, 39, 172, 178, 181, 190, 202, 204, 209, 212Han, C.-h. (2002). Interpreting interrogatives as rhetorical questions. Lingua,112(3):201\u2013229. \u2192 pages 31, 78, 132, 156Heim, J., Keupdjio, H., Lam, Z. W.-M., Osa-Go\u00b4mez, A., Thoma, S., andWiltschko, M. (2016). Intonation and particles as speech act modifiers: Asyntactic analysis. Studies in Chinese Linguistics, 37(2):109\u2013129. \u2192 pages39, 64Heim, J., Keupdjio, H., Lam, Z. W.-M., Osa-Go\u00b4mez, A., and Wiltschko, M.(2014). How to do things with particles. In Proceedings of the 2014 annualconference of the Canadian Linguistic Association. \u2192 pages 23, 40, 64, 65Heim, J. M. (2019a). Commitment and Engagement: The Role of Intonation inDeriving Speech Acts. PhD thesis, University of British Columbia. \u2192 pages13, 23, 64, 80, 140, 163, 177, 184Heim, J. M. (2019b). Turn-peripheral management of common ground: A studyof Swabian gell. Journal of Pragmatics, 141:130\u2013146. \u2192 pages 13, 39Hernanz, M. L. (2012). Sobre la periferia izquierda y el movimiento: elcomplementante si en espan\u02dcol. In Brucart, J. M. and A\u00b4ngel J. Gallego, editors,El movimiento de los constituyentes, pages 151\u2013171. Visor Libros, Madrid. \u2192pages 47, 48, 207Hernanz, M. L. (2015). Sobre la periferia izquierda de la oracio\u00b4n en espan\u02dcol:entre la polaridad y la (ad)miratividad. In Talk at the Jornadas Complutensessobre la grama\u00b4tica del espan\u02dcol en honor del profesor Ignacio Bosque, Madrid,May 27. \u2192 pages 47, 50, 54Hill, V. (2007). Vocatives and the pragmatics\u2013syntax interface. Lingua,117(12):2077\u20132105. \u2192 pages 173, 205244Hill, V. (2013). Vocatives: How syntax meets with pragmatics. Brill. \u2192 page 205Hinzen, W. (2006). Mind design and minimal syntax. Oxford University Press.\u2192 page 221Holmes, J. (1984). Modifying illocutionary force. Journal of pragmatics,8(3):345\u2013365. \u2192 page 169Jackendoff, R. S. (1972). Semantic interpretation in generative grammar. \u2192page 197Jamieson, E. A. (2019). Questions, biases and \u2018negation\u2019: Evidence from Scotsvarieties. PhD thesis, The University of Edinburgh. \u2192 pages13, 23, 39, 173, 209, 221Kaplan, D. (1999). The meaning of ouch and oops. In Cornell conference oncontext dependency. \u2192 page 118Kimps, D. (2018). Tag questions in conversation: A typology of theirinteractional and stance meanings, volume 83. John Benjamins PublishingCompany. \u2192 page 66Kiss, A. (2018). Attributing illocutionary acts in spanish by the tag question no?Paper presented at the Annual Conference of the Canadian LinguisticAssociation. \u2192 page 140Ko\u00a8lbel, M. (2004). III\u2014Faultless disagreement. Proceedings of the Aristoteliansociety, 104(1):53\u201373. \u2192 page 60Kratzer, A. (1999). Beyond ouch and oops: How descriptive and expressivemeaning interact. In Cornell conference on theories of context dependency,volume 26. Cornell University Ithaca, NY. \u2192 page 118Krifka, M. (2001). Quantifying into question acts. Natural language semantics,9(1):1\u201340. \u2192 page 195Krifka, M. (2008). Basic notions of information structure. Acta LinguisticaHungarica, 55(3-4):243\u2013276. \u2192 page 7Krifka, M. (2014). Embedding illocutionary acts. In Roeper, T. and Speas, M.,editors, Recursion: Complexity in cognition, pages 59\u201387. Springer,Switzerland. \u2192 page 192245Krifka, M. (2015). Bias in commitment space semantics: Declarative questions,negated quetions, and question tags. Proceedings of Semantics and LinguisticTheory (SALT), 25:328\u2013345. \u2192 pages 13, 132Ladd, D. R. (1981). A first look at the semantics and pragmatics of negativequestions and tag questions. Papers from the Regional Meeting ChicagoLinguistics Society, (17):164\u2013171. \u2192 page 132Lam, Z., Thoma, S., and Wiltschko, M. (2013). Thinking about you. InWorkshop on Interfaces at the Left Periphery, University of Michigan. \u2192 page65Lam, Z. W.-M. (2014). A complex ForceP for speaker-and addressee-orienteddiscourse particles in Cantonese. Studies in Chinese Linguistics, 35(2):61\u201380.\u2192 pages 13, 39, 177, 184, 209Landone, E. (2009). Los marcadores del discurso y la cortes\u0131\u00b4a verbal enespan\u02dcol. Bern: Peter Lang. \u2192 page 128Lasersohn, P. (2005). Context dependence, disagreement, and predicates ofpersonal taste. Linguistics and philosophy, 28(6):643\u2013686. \u2192 pages 17, 60Lasersohn, P. (2009). Relative truth, speaker commitment, and control of implicitarguments. Synthese, 166(2):359\u2013374. \u2192 page 60Lee, B. P. (2001). Mutual knowledge, background knowledge and shared beliefs:Their roles in establishing common ground. Journal of pragmatics,33(1):21\u201344. \u2192 page 15Levinson, S. (1983). Pragmatics. Cambridge Textbooks in Linguistics.Cambridge University Press. \u2192 page xxiLewis, D. (1979). Scorekeeping in a language game. In Semantics from differentpoints of view, pages 172\u2013187. Springer. \u2192 page 13Lewis, D. (2006). Discourse markers in English: a discourse-pragmatic view.Approaches to discourse particles, 43:59. \u2192 page 6Maienborn, C. and Scha\u00a8fer, M. (2011). Adverbs and adverbials. In Semantics.An international handbook of natural language meaning; Volume 2, pages157\u2013183. Berlin, New York: Mouton de Gruyter. \u2192 page 197246Malamud, S. A. and Stephenson, T. (2015). Three ways to avoid commitments:Declarative force modifiers in the conversational scoreboard. Journal ofSemantics, 32(2):275\u2013311. \u2192 pages xiv, 13, 14, 16, 23, 26, 27, 29, 136Mart\u0131\u00b4n-Gonza\u00b4lez, J. (2003). The syntax of sentential negation in Spanish. PhDthesis, Harvard University. \u2192 page 62Mayol, L. (2010). Contrastive pronouns in null-subject Romance languages.Lingua, 120(10):2497\u20132514. \u2192 page 196McGrath, M. and Frank, D. (2018). Propositions. In Zalta, E. N., editor, TheStanford Encyclopedia of Philosophy. Metaphysics Research Lab, StanfordUniversity, spring 2018 edition. \u2192 page xxiMeyerhoff, M. (1994). Sounds pretty ethnic, eh?: A pragmatic particle in NewZealand English. Language in Society, 23(3):367\u2013388. \u2192 pages 40, 235Mithun, M. (2008). The extension of dependency beyond the sentence.Language, 84(1):69\u2013119. \u2192 page 229Miyagawa, S. (2013). Surprising agreements at T and C. Talk given at RomaniaNova VI, Natal, Brazil. \u2192 pages 173, 190Mo\u00b4ccero, M. L. (2010). Las preguntas confirmatorias como indicadoras deposicionamiento intersubjetivo. Estudios filolo\u00b4gicos, (45):67\u201378. \u2192 page 11Moltmann, F. (2010). Relative truth and the first person. Philosophical studies,150(2):187\u2013220. \u2192 page 60Montan\u02dcez Mesas, M. P. (2007). Marcadores del discurso y posicio\u00b4n final: laforma \u00bfeh? en la conversacio\u00b4n coloquial espan\u02dcola. Estudios de Lingu\u00a8\u0131\u00b4stica.Universidad de Alicante, 21. \u2192 page 231Montan\u02dcez Mesas, M. P. (2008). El ape\u00b4ndice \u00bfno? en la conversacio\u00b4n coloquialespan\u02dcola. Bolet\u0131\u00b4n de Filolog\u0131\u00b4a, 43(2):117\u2013174. \u2192 pages 62, 63Montan\u02dcez Mesas, M. P. (2015). Marcadores discursivos conversacionales yposicio\u00b4n final. Hacia una caracterizacio\u00b4n discursiva de sus funciones enunidades de habla. PhD thesis, Universitat de Valencia. \u2192 pagesiii, 12, 37, 63, 65, 77, 128, 140, 217, 231, 232, 233, 235247Montol\u0131\u00b4o Dura\u00b4n, E. (1999). \u00a1Si nunca he dicho que estuviera enamorada de e\u00b4l!Sobre construcciones independientes introducidas por si con valor replicativo.Oralia: Ana\u00b4lisis del discurso oral, (2):37\u201370. \u2192 pages iii, 10, 50, 74Moore, J. D. and Pollack, M. E. (1992). A problem for rst: The need formulti-level discourse analysis. Computational linguistics, 18(4):537\u2013544. \u2192page 3Moro, A. (2003). Notes on vocative case. a case study in clause structure.Amsterdam Studies in the Theory and History of Linguistic Science, pages247\u2013262. \u2192 pages 173, 205Munaro, N. and Poletto, C. (2002). Ways of clausal typing. Rivista diGrammatica Generativa, 27:87\u2013105. \u2192 pages xvii, 3, 210, 212, 218, 219Ortega, J. (1985). Ape\u00b4ndices modalizadores en espan\u02dcol: los comprobativos.Estudios roma\u00b4nicos dedicados al profesor Andre\u00b4s Soria Ortega, pages239\u2013255. \u2192 pages iii, 11, 63, 128Oyharc\u00b8abal, B. (1993). Verb agreement with nonarguments: On allocutiveagreement. Generative Studies in Basque Linguistics, pages 89\u2013114. \u2192 pages173, 190Paul, W. (2009). Consistent disharmony: sentence-final particles in Chinese.Cambridge Occasional Papers in Linguistics, 5:1\u201324. \u2192 page 3Paul, W. (2014). New perspectives on Chinese syntax, volume 271. Walter deGruyter GmbH & Co KG. \u2192 page 39Pickering, M. J. and Garrod, S. (2004). Toward a mechanistic psychology ofdialogue. Behavioral and brain sciences, 27(2):169\u2013190. \u2192 page 1Pierrehumbert, J. and Hirschberg, J. B. (1990). The meaning of intonationalcontours in the interpretation of discourse. In P. R. Cohen, J. M. and Pollack,M. E., editors, Intentions in Communications, pages 3271\u2013311. \u2192 page 140Pons Border\u0131\u00b4a, S. (2006). A functional approach to the study of discoursemarkers. In Fischer, K., editor, Approaches to discourse particles, pages77\u2013100. Elsevier, Amsterdam. \u2192 page 37248Portner, P. (2004). The semantics of imperatives within a theory of clause types.Proceedings of Semantics and Linguistic Theory (SALT), pages 235\u2013252. \u2192pages 17, 18, 19, 31, 160Potts, C. (2007). The expressive dimension. Theoretical linguistics,33(2):165\u2013198. \u2192 page 118Reese, B. and Asher, N. (2007). Prosody and the interpretation of tag questions.In Proceedings of Sinn und Bedeutung, volume 11, pages 448\u2013462. \u2192 pages13, 132, 133Reese, B. and Asher, N. (2010). Biased questions, intonation and discourse.Information Structure: Theoretical, Typological, and ExperimentalPerspectives, pages 139\u201373. \u2192 page 132Rett, J. (2008). A degree account of exclamatives. Proceedings of Semantics andLinguistic Theory (SALT), 18:601\u2013618. \u2192 page 147Rett, J. (2011). Exclamatives, degrees and speech acts. Linguistics andPhilosophy, 34(5):411\u2013442. \u2192 page 147Rizzi, L. (1997). The fine structure of the left periphery. In Elements ofgrammar, pages 281\u2013337. Springer. \u2192 pages 3, 36, 172, 176, 197Rizzi, L. (2001). On the position \u201cint (errogative)\u201d in the left periphery of theclause. Current studies in Italian syntax: Essays offered to Lorenzo Renzi,pages 287\u2013296. \u2192 pages 177, 207Rodr\u0131\u00b4guez Mun\u02dcoz, F. J. (2009). Estudio sobre las funciones pragmadiscursivasde \u00bfno? y \u00bfeh? en el espan\u02dcol hablado. RLA. Revista de lingu\u00a8\u0131\u00b4stica teo\u00b4rica yaplicada, 47(1):83\u2013101. \u2192 pages11, 35, 63, 65, 70, 71, 72, 73, 78, 128, 169, 231Rodr\u0131\u00b4guez Ramalle, T. M. (2011). Sobre si y la organizacio\u00b4n del margenpreverbal en espan\u02dcol. LEA: Lingu\u00a8\u0131\u00b4stica espan\u02dcola actual, 33(2):199\u2013222. \u2192pages 10, 25, 50, 51, 54, 75, 89, 95, 110, 208Roggia, A. B. (2012). Eh as a polyfunctional discourse marker in DominicanSpanish. Journal of Pragmatics, 44(13):1783\u20131798. \u2192 pages 37, 231, 235Rojas-Esponda, T. (2014). A qud account of german\u2019doch\u2019. In Proceedings ofSinn und Bedeutung, volume 18, pages 359\u2013376. \u2192 page 94249Romero, M. and Han, C.-h. (2004). On negative yes\/no questions. Linguisticsand Philosophy, 27(5):609\u2013658. \u2192 pages 92, 132, 163Ross, J. R. (1970). On declarative sentences. Readings in Englishtransformational grammar, 222:272. \u2192 pages 37, 38, 173, 178Rudin, D. (2018). Rising above commitment. PhD thesis, UC Santa Cruz. \u2192pages 13, 14, 16, 17, 21, 78, 82, 83, 152Rudin, D. and Beltrama, A. (2019). Default agreement with subjective assertions.In Proceedings of SALT 2019, volume 29, pages 82\u2013102. \u2192 page 18Sacks, H. (1987). On the preferences for agreement and contiguity in sequencesin conversation. In Button, G. and Lee, J. R., editors, Talk and SocialOrganisation, chapter 2, pages 54\u201369. Multilingual Matters. \u2192 page 1S\u00e6b\u00f8, K. J. (2009). Judgment ascriptions. Linguistics and philosophy,32(4):327\u2013352. \u2192 page 60S\u02c7afa\u00b4r\u02c7ova\u00b4, M. (2005). The semantics of rising intonation in interrogatives anddeclaratives. In Proceedings of sinn und bedeutung, volume 9, pages 355\u2013369.\u2192 page 140Sailor, C. (2009). Tagged for deletion: a typological approach to VP ellipsis intag questions. UCLA Master\u2019s thesis. \u2192 pages 219, 220San Mart\u0131\u00b4n Nu\u00b4n\u02dcez, A. (2011). Los marcadores interrogativos de control decontacto en el corpus PRESEEA de Santiago de Chile. Bolet\u0131\u00b4n de filolog\u0131\u00b4a,46(2):135\u2013166. \u2192 pages 128, 235Sanders, T. J. and Noordman, L. G. (2000). The role of coherence relations andtheir linguistic markers in text processing. Discourse processes, 29(1):37\u201360.\u2192 page 7Sbisa`, M. (2001). Illocutionary force and degrees of strength in language use.Journal of pragmatics, 33(12):1791\u20131814. \u2192 page 169Schiffer, S. (1972). Meaning. Oxford: Oxford University Press. \u2192 page 15Schlo\u00a8der, J. J. (2018). Assertion and rejection. PhD thesis, University ofAmsterdam. \u2192 pages 13, 60250Schweinberger, M. (2018). The discourse particle eh in New Zealand English.Australian Journal of Linguistics, 38(3):395\u2013420. \u2192 pages 40, 235Schwenter, S. (2014). Pragmatics of conditional marking: implicature, scalarity,and exclusivity. Routledge. \u2192 page 52Schwenter, S. A. (1998). The pragmatics of conditional marking: implicature,scalarity, and exclusivity. PhD thesis, Stanford University. \u2192 pages50, 95, 230Schwenter, S. A. (2000a). Spanish connectives and pragmatic implicatures. InHispanic linguistics at the turn of the millennium: Papers from the 3rdHispanic Linguistics Symposium, pages 292\u2013307. \u2192 pages 52, 53Schwenter, S. A. (2000b). Viewpoints and polysemy: Linking adversative andcausal meanings of discourse markers. In Couper-Kuhlen, E. and Kortmann,B., editors, Cause\u2013condition\u2013concession\u2013contrast: Cognitive and discourseperspectives, pages 257\u2013282. Mouton de Gruyter, Berlin. \u2192 pagesiii, 10, 50, 52, 53Schwenter, S. A. (2016a). Independent si-clauses in spanish. Insubordination,115:89. \u2192 pages 11, 24, 26, 44, 49, 50, 51, 55Schwenter, S. A. (2016b). Meaning and interaction in Spanish independentsi-clauses. Language Sciences, 58:22\u201334. \u2192 pagesiii, 10, 11, 50, 54, 55, 73, 75, 77, 87, 88, 89, 91, 124, 226Searle, J. R. (1969). Speech acts: An essay in the philosophy of language.Cambridge: Cambridge University Press. \u2192 pagesxiii, 32, 33, 75, 96, 102, 108, 113, 118Searle, J. R. (1976). A classification of illocutionary acts. Language in Society,5(1):1\u201323. \u2192 pages xiii, 31, 75Searle, J. R. (1979). A taxonomy of illocutionary acts. In Expression andmeaning: Studies in the Theory of Speech Acts, pages 1\u201329. CambridgeUniversity Press, Cambridge. \u2192 pages xiii, 30, 31Searle, J. R. and Vanderveken, D. (1985). Foundations of illocutionary logic.Cambridge: Cambridge University Press. \u2192 pages 94, 170251Speas, P. and Tenny, C. (2003). Configurational properties of point of view roles.In DiSciullo, A. M., editor, Asymmetry in Grammar (Vol 1), pages 315\u2013344.John Benjamins, Amsterdam. \u2192 pages 3, 39, 172, 178, 179, 181, 197, 198Stalnaker, R. (1978). Assertion. Syntax and Semantics, 9:315\u2013322. \u2192 page 15Stalnaker, R. (2002). Common ground. Linguistics and philosophy,25(5-6):701\u2013721. \u2192 pages 14, 24, 76Starks, D., Thompson, L., and Christie, J. (2008). Whose discourse particles?New Zealand eh in the Niuean migrant community. Journal of Pragmatics,40(7):1279\u20131295. \u2192 page 40Stenstro\u00a8m, A.-B. (2005). Teenagers\u2019 tags in London and Madrid.Contexts-Historical, Social, Linguistic. Studies in Celebration of Toril Swan,pages 279\u2013291. \u2192 page 231Stephenson, T. (2007). Judge dependence, epistemic modals, and predicates ofpersonal taste. Linguistics and Philosophy, 30(4):487\u2013525. \u2192 pages 17, 60Stojanovic, I. (2007). Talking about taste: disagreement, implicit arguments, andrelative truth. Linguistics and Philosophy, 30(6):691\u2013706. \u2192 page 60Stubbe, M. and Holmes, J. (1995). You know, eh and other \u2018exasperatingexpressions\u2019: An analysis of social and stylistic variation in the use ofpragmatic devices in a sample of new zealand english. Language &Communication, 15(1):63\u201388. \u2192 page 235Sudo, Y. (2013). Biased polar questions in English and Japanese. In Beyondexpressives: Explorations in use-conditional meaning, pages 275\u2013295. Brill.\u2192 pages 132, 133, 163, 166Sweetser, E. (1990). From etymology to pragmatics: Metaphorical and culturalaspects of semantic structure. Cambridge: Cambridge University Press. \u2192page 230Tarpent, M.-L. (1991). The morpheme--and the mysteries of nisgha syntax. 26thICSNL, pages 315\u2013347. \u2192 page 229Thaler, V. (2012). Mitigation as modification of illocutionary force. Journal ofPragmatics, 44(6-7):907\u2013919. \u2192 page 169252Thoma, S. C. (2016). Discourse particles and the syntax of discourse: evidencefrom Miesbach Bavarian. PhD thesis, University of British Columbia. \u2192pagesiv, xvii, 13, 15, 16, 19, 23, 33, 39, 40, 80, 172, 177, 183, 185, 202, 209, 212, 221Torres Bustamante, T. (2013). On the syntax and semantics of mirativity:Evidence from Spanish and Albanian. PhD thesis, Rutgers University. \u2192pages 54, 126Travis, C. E. (2006). The natural semantic metalanguage approach to discoursemarkers. In Fischer, K., editor, Approaches to discourse particles, pages219\u2013241. Elsevier, Amsterdam. \u2192 page 37Truckenbrodt, H. and Maienborn, C. (2012). Semantics of intonation. Handbookof semantics. \u2192 page 140Ucle\u00b4s Ramada, G. (2018). La atenuacio\u00b4n de los marcadores de control decontacto en PRESEEA: un estudio comparativo entre Espan\u02dca y Me\u00b4xico.RILCE: Revista de Filolog\u0131\u00b4a Hispa\u00b4nica, 34(3):1313\u20131335. \u2192 pages 128, 235Ucle\u00b4s Ramada, G. (2020). Epistemic (a)symmetries and mitigation in thedescription of conversational markers: the case of Spanish \u00bfno? CorpusPragmatics, 4:107\u2013131. \u2192 pages 128, 169Van Rooy, R. and S\u02c7afa\u00b4r\u02c7ova\u00b4, M. (2003). On polar questions. Proceedings ofSemantics and Linguistic Theory (SALT), 13:292\u2013309. \u2192 pages 132, 133Vanderveken, D. (1990). Meaning and Speech Acts: Volume 1, Principles ofLanguage Use. Cambridge: Cambridge University Press. \u2192 pages 32, 94, 170Villa-Garc\u0131\u00b4a, J. (2015). The Syntax of multiple-que Sentences in Spanish: Alongthe left periphery. Amsterdam: John Benjamins. \u2192 page 9Villa-Garc\u0131\u00b4a, J. (2019). Recomplementation in English and Spanish: Delineatingthe CP space. Glossa, 4(1). \u2192 pages 9, 225Vine, B. and Marsden, S. (2016). Eh at work: The indexicality of a new zealandenglish pragmatic marker. Intercultural Pragmatics, 13(3):383\u2013405. \u2192 page235Waldie, R. J. (2012). Evidentiality in Nuu-chah-nulth. PhD thesis, University ofBritish Columbia. \u2192 page 162253Ward, G. and Hirschberg, J. (1985). Implicating uncertainty: The pragmatics offall-rise intonation. Language, pages 747\u2013776. \u2192 page 140Wilson, D. and Sperber, D. (1990). Outline of relevance theory.HERMES-Journal of Language and Communication in Business, (5):35\u201356.\u2192 page 7Wilson, D. and Sperber, D. (1999). Relevance and relevance theory. In MITEncyclopedia of the Cognitive Sciences. Citeseer. \u2192 page 7Wilson, D. and Sperber, D. (2002). Relevance theory. Malden, MA: Blackwell.\u2192 page 7Wiltschko, M. (2014). The universal structure of categories: Towards a formaltypology. Cambridge: Cambridge University Press. \u2192 pagesxvii, 183, 184, 185, 209Wiltschko, M. (2016). Response particles beyond answering. Order andstructure in syntax I, page 241. \u2192 pages 19, 178Wiltschko, M. (2017). Ergative constellations in the structure of speech acts. InThe Oxford Handbook of Ergativity. \u2192 pages xvii, 39, 40, 172, 178, 183, 184Wiltschko, M., Denis, D., and D\u2019Arcy, A. (2018). Deconstructing variation inpragmatic function: A transdisciplinary case study. Language in Society,47(4):569\u2013599. \u2192 page 29Wiltschko, M. and Heim, J. (2016). The syntax of confirmationals. Outside theClause: Form and function of extra-clausal constituents, 178:305. \u2192 pagesiv, xiv, xvi, xvii, 3, 6, 13, 15, 16, 27, 28, 29, 37, 39, 40, 64, 140, 142, 172, 173, 174, 178, 179, 183, 184, 185, 189, 200, 204, 209, 210, 211, 212Wiltschko, M., Thoma, S., Osa-Go\u00b4mez, A., Keupdjio, H., Lam, Z., and Heim, J.(2015). Towards a typology of the grammar of discourse: a case-study ofconfirmationals. Paper presented at the 11th Annual Conference of theAssociation of Linguistic Typology. \u2192 page 30Yang, X. and Wiltschko, M. (2016). The confirmational marker ha in NorthernMandarin. Journal of Pragmatics, 104:67\u201382. \u2192 pages14, 29, 79, 133, 152, 158, 169, 209254Zanuttini, R. (2008). Encoding the addressee in the syntax: Evidence fromEnglish imperative subjects. Natural Language & Linguistic Theory,26(1):185\u2013218. \u2192 page 173Zimmermann, M. (2009). Discourse particles in the left periphery. In Dislocatedelements in discourse, pages 208\u2013239. Routledge. \u2192 page 3Zimmermann, M. (2011). Discourse particles. von heusinger, maienborn andportner (eds.) 2011, handbook of semantics (hsk 33.2). \u2192 page 13Zorraquino, M., Antonia, M., and Portole\u00b4s, J. (1999). Los marcadores deldiscurso. Grama\u00b4tica descriptiva de la lengua espan\u02dcola, 3:4051\u20134213. \u2192 page231Zubizarreta, M. L. (1998). Prosody, focus and constituent order. \u2192 page 218255","@language":"en"}],"Genre":[{"@value":"Thesis\/Dissertation","@language":"en"}],"GraduationDate":[{"@value":"2020-11","@language":"en"}],"IsShownAt":[{"@value":"10.14288\/1.0394124","@language":"en"}],"Language":[{"@value":"eng","@language":"en"}],"Program":[{"@value":"Linguistics","@language":"en"}],"Provider":[{"@value":"Vancouver : University of British Columbia Library","@language":"en"}],"Publisher":[{"@value":"University of British Columbia","@language":"en"}],"Rights":[{"@value":"Attribution-NonCommercial-NoDerivatives 4.0 International","@language":"*"}],"RightsURI":[{"@value":"http:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/","@language":"*"}],"ScholarlyLevel":[{"@value":"Graduate","@language":"en"}],"Title":[{"@value":"Epistemic (mis)alignment in discourse : what Spanish discourse markers reveal","@language":"en"}],"Type":[{"@value":"Text","@language":"en"}],"URI":[{"@value":"http:\/\/hdl.handle.net\/2429\/75792","@language":"en"}],"SortDate":[{"@value":"2020-12-31 AD","@language":"en"}],"@id":"doi:10.14288\/1.0394124"}