Habilitation thesis EGLE MOCCIARO Brno 2022 FACULTY OF ARTS Paths of modality and beyond. The grammaticalisation of modals and directive m in Sicilian. Department of Romance languages and literatures Bibliographic record Author: Title of Thesis: Paths of modality and beyond. The grammaticalisation of modals and directive m in Sicilian. Year: Number of Pages: Keywords: Abstract Declaration Paths of modality and beyond. The grammaticalisation of modals and directive markers in Sicilian, TABLE OF CONTENTS Table of contents Volume I Preface. 8 List of figures and tables. 10 Abbreviations and transcription conventions. 11 1 Framing the study of modality in Sicilian: an introduction. 13 1.1 Why modality? 13 1.2 On Sicilian. 15 1.2.1 The lack of a standard and diatopic micro-variation. 15 1.2.2 Documentary history and existing corpora for studying Sicilian. 18 1.2.2.1 Medieval texts (14th and 15th centuries). 18 1.2.2.2 Beyond the Middle Ages (16th to 19th century). 23 1.2.2.3 Contemporary Sicilian. 26 1.3 Research on modality and modal means in Sicilian. 28 1.3.1 The dialectology approach. 28 1.3.2 Dialectology and beyond: three examples. 31 1.3.3 Modality and beyond: some peripheries. 40 1.4 Theoretical assumptions: three key notions. 43 1.4.1 General framework. 43 1.4.2 Subjectification. 49 1.4.2.1 The notion of subjectivity. 49 1.4.2.2 From subjectivity to subjectification. 57 1.4.3 Grammaticalisation. 67 1.4.3.1 Grammaticalisation and constructions. 72 1.4.3.2 Source concepts and patterns of change. 76 1.4.4 Semantic topography. 86 1.5 Outline of the volume: paths of modality and postmodality. 90 2 First path. From possession and volition to modal meanings (“Possession and volition in the development of modal meanings: A case study from Sicilian” 2009 [2019]). 95 2.1. Introduction. 95 TABLE OF CONTENTS 2.2 Theoretical assumptions. 97 2.2.1 Subjectification. 98 2.3 Aviri a + INF. 101 2.4. Vuliri + INF. 107 2.5. Modal meanings and temporal-aspectual features. 114 3 Second path. From volition to modal-passive meanings (“Vuliri + PP in the data of the Atlante Linguistico della Sicilia”, 2016). 118 3.1 Introduction. 118 3.2 The state of the art. 120 3.3 Data analysis. 123 3.3.1 Restrictions on use. 123 3.3.2 Formal and semantic properties. 124 3.3.3 From Sicilian to Italian. 127 3.4 The grammaticalisation of vuliri + PP: some considerations. 129 4 Third path. From volition to directive meanings (“Vuliri in ancient Sicilian between volition and modality”, 2018). 133 4.1 Subject of investigation and research methodology. 133 4.2 Volition and modality. 135 4.3 Volitional meanings. 139 4.3.1 Premodal lexical constructions. 139 4.3.2 Participant-internal volition: the modal vuliri + infinitive. 145 4.3.3 Subjects of volition. 151 4.4 Beyond volition. 153 4.4.1 Participant-oriented meanings. 153 4.4.2 Speaker-oriented meanings. 158 4.5 Beyond modality. 160 4.5.1 Optative expressions. 160 4.5.2 Irrelevance expressions. 164 4.6 Values of imminence. 168 4.7 To conclude. 170 TABLE OF CONTENTS 5 Forth path. From (itive) motion to illocution (“The grammaticalisation of motion verbs in Sicilian: the case of iri ‘to go’ in directive function”, 2019). 175 5.1 The grammaticalisation of motion verbs: an introduction. 175 5.2 Motion verbs and grammaticalisation. 180 5.2.1 The grammaticalisation of directive markers. 184 5.3 Data Analysis. 187 5.3.1 Texts and speakers. 187 5.3.2 Semantic and lexical characteristics. 188 5.3.3 Formal characteristics (decategorisation and phonetic erosion). 198 5.3.3.1 Inflection, defectiveness, phonetic erosion. 200 5.3.3.2 The element a: possible reconstructions. 207 5.4 A diachronic overview. 210 5.5 Conclusions. 217 6 Fifth path. A semantic map for modal verbs (“Participant oriented modality in contemporary Sicilian: an analysis of data from the Atlante Linguistico della Sicilia”, 2021). 219 6.1 Subject of investigation. 219 6.2 A semantic-functional map of participant-oriented modality domain. 220 6.3 Aviri a + infinitive. 223 6.4 Putiri/sapiri + infinitive. 227 6.5 Vuliri + infinitive. 230 6.6 Data from the translation questions. 235 6.7 Conclusions. 239 7 Towards a unified account: a modality’s semantic map for Sicilian. 241 References. 249 Volume II Commentary to Habilitation thesis. 278 Appendix to the Habilitation thesis (Original articles). 280 Preface. This work brings together a number of articles I have previously published, in most cases in Italian, as individual or collaborative research. Conceived for the habilitation thesis, the collection is, however, far from being purely instrumental. Rather, it represents a rational path along one of the main lines of investigation that have characterised in the past and still characterise my research activity, namely the expression of modality in Sicilian. Two aspects of this overall work should be highlighted. On the one hand, the diachrony of the reflection, which goes through different phases and gradually becomes more defined as far as the approach is concerned, while leaving the theoretical-methodological assumptions intact. On the other hand, the cooperative dimension of research, which also embraces the present and the near future (a conference on the topic was hosted in 2021 by our Department of Romance languages and literatures, in collaboration with the University of Palermo, the results of which are currently being published). The articles selected here reflect well, I believe, my contribution to this shared endeavour and, more generally, my research attitude, namely the interest in the analytical tools and methodology, especially those related to the theories of grammaticalisation, and their application to interpret linguistic data: in particular, the paths leading from cross-linguistically common notional sources, such as “volition” or “directional movement”, to grammatical meanings. It seemed to me that presenting for evaluation a research topic that has developed consistently over the years, also reflecting my vocation for dialogue and cooperative enquiry, could be a comprehensive way of outlining my scholarly profile. In order to facilitate a unified reading and to highlight internal connections and lines of development, individual articles have been framed in a theoretical-methodological structure, consisting of the introductory and the final chapters. This framework provides a detailed account of the research background and different theoretical approaches to language change that my thinking has gone through over the years. The five works presented fully reflect the originals in their content and structure. However, for the sake of consistency, they are presented in English, with minor adaptations from the Italian originals of most of them, which mainly facilitate the linguistic transition. In addition, to facilitate understanding by non-Sicilian readers, interlinear glosses are consistently provided for all examples in ancient and contemporary Sicilian. Finally, the reference list has been updated where necessary and, in order to avoid repetition and facilitate consultation, it is emancipated from the individual contributions and listed in one place, at the end of the work. Other minor changes compared to the originals are the addition of internal cross-references and the narrowing down of some notes, in particular, those relating to the corpora used, where repeated from one article to the next. Otherwise, everything new in the volume is contained in the theoretical introduction and, above all, in the final synthesis. The original articles are collected in an Appendix that constitutes the second volume of the habilitation thesis, together with its Commentary. Reference to them and, in the case of collaborative works, to the share of my contribution is made explicit at the beginning of each chapter. Brno, 15 August 2022 LIST OF FIGURES AND TABLES List of figures and tables Figure 1 Prototypical categories. 46 Figure 2 Objective and subjective construals (Langacker 1990:7). 59 Figure 3 Grounding (Langacker 1990: 10). 59 Figure 4 Grounding predications (Langacker 1990: 26). 60 Figure 5 Basic relation of possession (Langacker 1990: 29, adapted). 78 Figure 6 Path of development from obligation (Bybee et al. 1994: 240). 82 Figure 7 Path of development from desire and motion (Bybee et al. 1994: 240). 82 Figure 8 Path of development from ability, I (Bybee et al. 1994: 194). 84 Figure 9 Path of development from ability, II (Bybee et al. 1994: 199). 84 Figure 10 Basic motion event. 85 Figure 11 Modality’s semantic map (van der Auwera and Plungian 1998: 111). 89 Figure 12 Objectively construed relation (Langacker 1991: 216, adapted). 99 Figure 13 Subjectively construed relation (Langacker 1991: 216, adapted). 99 Figure 14 Degrees of salienece of the Ground (G). 100 Figure 15 Schematic import of aviri a. 102 Figure 16 Semantic map of aviri a. 107 Figure 17 Schematic import of vuliri. 108 Figure 18 Semantic map of vuliri. 113 Figure 19 Epistemic model (Langacker 1991: 243–244, adapted). 115 Table 1 Semantic-functional network of vuliri. 174 Figure 20 Path of development of [va IMP]. 217 Figure 21 Participant-oriented modality’s semantic map. 221 Figure 22 Semantic map of aviri a + infinitive. 223 Figure 23 Semantic map of putiri/sapiri + infinitive. 228 Figure 24 Semantic map of vuliri + infinitive. 235 Figure 25 Participant-oriented modality in Sicilian. 239 Figure 26 Modality’s semantic map of Sicilian. 248 ABBREVIATIONS AND TRANSCRIPTION CONVENTIONS Abbreviations and transcription conventions Morphological glosses (based on Leipzig rules) ACC accusative CLIT clitic COMP complementiser COND conditional CONJ conjunction DAT dative EVAL evaluative EXS existential F feminine FUT future GER gerund IMP imperative IMPS impersonal INF infinitive INT intensifier IPFV imperfective IPRF imperfect LE linking element LOC locative M masculine NEG negation PFV perfective PL plural PRS present PST past PTCP participle RED reduced form REL relative REFL reflexive SBJV subjunctive SG singular SLE syntactic linking element SUP superlative degree = morpheme segmentation ABBREVIATIONS AND TRANSCRIPTION CONVENTIONS City name acronyms. AG Agrigento CL Caltanissetta CT Catania EN Enna ME Messina PA Palermo RG Ragusa SR Siracusa TP Trapani Transcription rules for Sicilian.* Graphic-phonetic correspondances ṭ voiceless retroflexe occlusive [ʈ] ḍ voiced retroflex occlusive [ɖ] ṛ voiced retroflex flap [ɽ] š fricative post-alveolar [ʃ] ç voiceless postalveolar affricate [tʃ] or fricative post-alveolar [ʃ] j palatal approximant [j] â, ê, ô, û merging of consecutive vowels belonging to different morphemes Conversational signals : prolonged sound Other conventions SMALL CAPS Latin words or metaphors → Directionality of linguistic change * The texts in Sicilian conform to the choices of the source text and thus to different transcribers, both for contemporary Sicilian and for Sicilian of other chronological phases. In principle, the transcription is intuitive. Only a few recurrent and less readable forms are listed in the table. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 13 1 Framing the study of modality in Sicilian: an introduction. 1.1 Why modality? Modality is not a new topic in Sicilian studies. Sharp observations and remarks can be already found in the dialectological tradition, albeit often in a fragmentary way (cf. inter al. Leone 1995; Rohlfs 1968, 1969; Ruffino 1997a, 1997b, 2001; Varvaro 1988), and, in more recent years, a number of descriptions have been proposed using more generalist and theoretical models of analysis. These works have, from time to time, focused on and described central aspects of modality in Sicilian (cf. Bentley 1997b, 1998b, 2000c) or, otherwise, opened up new perspectives in terms of forms and functions within or around modality (cf. inter al. Brucale et al. 2022; Cruschina 2010a, 2010b, 2015; Cruschina and Bianchi 2021; Lo Baido Forthcoming; Scivoletto Forthcoming). What is still missing, however, is a systematic mapping of the functions into which the modal domain is articulated in Sicilian and the forms through which these functions are encoded. There is also a lack of programmatic attention to what can be called the “surroundings” of modality: that is, the peripheries of the modal domain, which are in contiguity with other domains, similar but not necessarily coinciding with it (tense, first of all, which among the contiguous domains is perhaps the most explored, but also directivity, evidentiality, mirativity etc.). In these areas, interesting interface phenomena occur that affect not only morphosyntax but also other levels of analysis, such as pragmatics. The sections forming this introductory chapter serve to ground the collected works within a coherent methodological and theoretical framework. I will begin by providing some coordinates – linguistic and socio-historical – within which the study of Sicilian is framed (section 1.2.1), the corpora that serve as a starting point for this study (1.2.2), the panorama of existing research (1.3), including both traditional and dialectological research (1.3.1) and, on the other hand, studies using more theoretical descriptive and analytical models to reinterpret traditional issues (1.3.2) or new modal subdomains and areas of cross-domain overlap (1.3.3). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 14 Next, I move on to describe the theoretical background against which the various works presented are set (1.4). This is in fact an articulated model, which brings together and integrates notions from other intimately connected models: namely the functionalist, usage-based and discourse-grounded approach (e.g., Bybee 2006, 2007, 2013; Bybee et al. 1994) and cognitive grammar (Langacker 1987, 1991). As we shall see, these approaches share the assumption that grammatical structures originate and develop in the concrete speakers’ use. In other words, it is the specific utterances produced in discourse that form the basis, due to frequency and prototypicality reasons, for those “constantly evolving set of cognitive routines”, which are the grammatical constructions (Langacker 1987: 57). A “form”, then, but not in a purely Saussurian sense: first of all, because the notion of “system” loses all synchronic rigidity and is rearticulated in a dynamic hypothesis of the facts of language, which are conceived of as inherently unstable and “emergent”, to use a famous expression by P. J. Hopper (1987). Two major consequences derive from this intrinsecally unstable view of language, namely the dissolution of the dichotomous synchrony vs diachrony relationship and the need to look at linguistic categories in prototypical terms, that is, as aggregates of exemplars occupying more or less peripheral positions from a certain categorical center. Language, in short, is not a state but a dynamic process. Or, to go back to Saussure, the langue is rooted and continuously realised in the parole. It is precisely the focus on usage that clearly explains the association with the insights derived from the linguistics of parole par excellence, namely sociolinguistics, which has lent its tools to several of the works presented here. However, the interpretive path proposed in the next sections only leaves the sociolinguistic model in the background and instead foregrounds some other theoretical issues. In particular, I will focus on three key notions on which the cognitivist and functionalist models just introduced largely converge and which make it possible to trace a common thread linking the different collected works. These notions are subjectification (1.4.1), grammaticalisation (1.4.2), and, especially concerned with representation models, semantic topography (1.4.3). Finally, a brief description of the articles collected in the next sections will be offered, which makes explicit how they relate to the theoretical notions presented (1.5). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 15 1.2 On Sicilian. 1.2.1 The lack of a standard and diatopic micro-variation. Sicilian is a Southern Italo-Romance language spoken in Sicily and its satellite islands (almost five million people in 2022, according to ISTAT, the Italian Institute of Statis- tics).1 It has no official status and is mostly employed in informal situations, while the national language, that is, Italian (or its regional variety) is employed also in more formal contexts and as a written language. Very schematically, it can be said that most Sicilians are bilingual (possibly with Sicilian as a non-native language), a certain amount is monolingual Italian (especially the latest generations), while Sicilian monolingualism is at best residual (cf. D’Agostino 2007 [2012]: 58–62; D’Agostino and Paternostro 2013: 447; Ruffino 2006).2 This language distribution is a consequence of the gradual spread of the national language starting with the birth of the Italian State in 1861, and then more and more rapidly in post-World War II republican Italy, up to the present day (cf. De Mauro 1963 for an in-depth analysis of the sociolinguistic situation in united Italy).3 In fact, the Hence, it belongs to the extreme subgroups, together with Salento and Calabrian varieties (Pellegrini 1977). Sicilian and Italian are not the only languages occupying the linguistic space of Sicily. A complete description should also include the different alloglot varieties resulting from immigrations dating back to the 11th – 13th centuries, coming from north-western Italy and connected to the Norman conquest, and immigrations dating back to the 15th –17th centuries, coming from Albania and connected to the Turkish invasion of the Balkan peninsula. Gallo-Italic varieties are still vital today in several centres in the provinces of Messina, Catania, Enna, Siracusa; Arbëreshe varieties are spoken in some centres in the Palermo area (for an overall description, see Trovato 2013). Alongside these historical minorities, the alloglot repertoires today also include the new immigrant languages, i.e., the languages of the new communities resulting from the immigration flows in recent decades from various non-European areas. These languages are only now beginning to receive systematic attention (see D’Agostino 2021; D’Agostino and Paternostro 2013: 480–487). 3 De Mauro (1963: 37) clearly explains how the terms “Italian” and “Italian-speaking” define situations that are not overlapping in Italy at the time of political unification. Access to Italian – a language used by a small circle of people and essentially a written code – was possible only for those who had access to higher education: that is, a very small percentage of the population according to the first census in 1861: “At the time of unification, therefore, the Italian population was for almost 80 percent deprived of the possibility of coming 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 16 relationship between Sicilian and Italian can best be described in terms of diglossia, that is, in terms of coexistence of codes with different status and prestige. It is only in this specific sociolinguistic perspective, which concerns the functional distribution of the two codes, that this relationship can be defined in terms of language (Italian) vs dialect (Sicilian).4 This dichotomy is an entirely socially grounded construct, as well explained by D’Agostino (2012: 74–78): a hierarchical relationship based on criteria external to the languages themselves, and inherently subject to fluctuations over time. From an internal – that is, merely structural – point of view, Sicilian is a historical-natural language exactly like its other sister languages, which stemmed out from spoken Latin on the territory of presentday Italy.5 As is well known, one of these Italo-Romance varieties, the Florentine, rose to into contact with the written use of Italian, that is, due to the already mentioned absence of oral use, of Italian without any other specification.” 4 As the use of Italian spreads, we move from a situation of diglossia to one called by Berruto (1987) dilalìa. In dilalìa there is functional overlap between language A and language B in informal uses and in primary socialisation; on the other hand, Italian (language A) remains the only possible one in functionally high domains. Dilalìa also explains the frequent phenomena of code-switching (i.e., switching, often unconsciously, from one code to another within the same communicative situation) and code-mixing (the presence of several codes within the same sentence), studied by Alfonzetti (1992 [2012]). 5 The problem of the Latinisation of Southern Italy has long been debated. According to Rohlfs (1924, 1965, 1972d, 1974), the Sicilian Romance varieties are not the result of the ancient Romanisation (241 BCE), but varieties imported from neighbouring areas since the Norman conquest of the south (1061 CE). Before then, Greek prevailed (not only in Sicily, especially in the eastern part, but also in Salento and Calabria), due to uninterrupted continuity from the ancient language in use in Magna Graecia. On the other hand, in nonGreek (western) areas, the early process of Latinisation would have been interrupted by the beginning of Arab domination (827 CE). With the Norman rule, a new Romance wave would begin, leading to the decay of Arabic. Reacting to Rohlfs’ proposal, several Italian scholars (e.g., Alessio 1948, 1949; Parlangeli 1964- 1965) argued that Romanisation had completely obliterated ancient southern Greekness and that, therefore, southern Greekness was due to Byzantine colonisation from the time of the Justinian reconquest (535 CE). A more nuanced proposal is due to Varvaro (1979, 1981, 1995), who reads the ancient Sicilian linguistic situation not as a series of linguistic strata in succession (Rohlfs 1965), but as a composite and dynamic situation, which will stabilise in the Norman era. Varvaro observes that during this period, the island underwent a “great demographic, ethnic, social and cultural crisis”, which induced its inhabitants to abandon the countryside and regroup in a few large centres (Varvaro 1981: 116). In these centres, local Romance component acted as an element of cohesion for the island’s Arabic-speaking and Greek-speaking components and trig- 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 17 the role of language due to cultural and socio-historical reasons, finally acquiring “an army and a navy” – to use one of the most popular images in sociolinguistics to define the notion of language in contrast to dialect. Since the perspective adopted here is more internal than sociolinguistic (although it draws heavily on insights from sociolinguistics), I will prefer the labels “language” or “language variety”. In the absence of a reference standard, we cannot speak of a single Sicilian language (beyond the obvious consideration that no language is a monolithic block uniformly used by speakers). Traditionally, Sicilian has been divided into two main varieties (Piccitto 1951): western Sicilian and central-eastern Sicilian, characterised by the presence (western varieties) or absence (central-eastern varieties) of metaphony, i.e., an assimilatory process that raises mid-vowels when followed by high vowels (e.g., BŎNU > bbuònu, bbunu ‘good:M.SG’; BŎNA > bona ‘good:F.SG’). The western group includes the varieties used in the provinces of Palermo, Trapani, and central-western Agrigento; the central-eastern group is divided into central Sicilian, with the varieties used in the Madonie area, the varieties spoken in the provinces of Caltanissetta and Enna, and eastern Agrigento; and eastern Sicilian, which includes the varieties spoken in the provinces of Messina, Catania, Siracusa, and Ragusa. (On more recent and articulated classification aspects, see Ruffino 1991, 1997a, 1997b, 2001, who also includes morphosyntactic criteria; new proposals for classification are in Cruschina 2020.) As Sottile (2013: 230) well summarises, two firm points must guide the classification of Sicilian: 1) it is more appropriate to speak of a plurality of varieties – he speaks of dialects – rather than a single one; 2) many of these show a series of common features. Thus, the plurality of “Sicilians” has its basic unity by virtue of a group of significant and to some extent exclusive common traits. These common traits occur, then, in more or less extensive and compact areas delimiting, albeit not always clearly, specific “linguistic areas”. gered a process of assimilation of the different groups into a new cohesive social and linguistic body, sanctioning the birth of modern Sicilian (Varvaro 1981: 54; cf. also Alfieri 2013). This also explains, according to Varvaro, the lower internal fragmentation of Sicilian compared to other southern varieties. See also, from the point of view of language contact, Fanciullo (1984, 1996: 59). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 18 This situation, at once coherent and internally composite, has developed through a linguistic-historical path that follows that of the different dynasties that have dominated the island over the centuries, “in an ideal trajectory through Norman, Angevin and Aragonese Sicily, through Castilian and Habsburg Sicily and then into ‘Italian’ Sicily, distinguishing in this case between pre- and post-unification Sicily and ‘Italian’ Sicily, from its annexation to the Savoyard Kingdom to its current status as a region with autonomous status within the Republic” (Alfieri 2013: 570)6 . The linguistic history that develops along this ideal path is constantly marked by multilingualism and thus by contact phenomena. 1.2.2 Documentary history and existing corpora for studying Sicilian. 1.2.2.1 Medieval texts (14th and 15th centuries). Sicilian certainly has a long documentary history, although not very ancient compared to other Italo-Romance varieties. As for non-literary documentation, Romance vulgarisms and place and person names emerging from archival texts in Greek, Arabic and Latin are the only forms providing us with information on local Romance in the period before the 13th century (Varvaro 1995: 229). Even later, until the 14th century, most of the information about the local Romace and its use comes from external history (see Varvaro 1981; cf. also fn. 4). Before this there is of course the great experience of the Scuola poetica siciliana ‘Sicilian poetical school’, initiated by Frederick II of Swabia, emperor of the Holy Roman Empire from 1220 until his death in 1250. In support of his project of statist and secular centralisation in Sicily and the south of the kingdom, Frederick II promoted the formation of an organic intellectual élite, mostly embodied by court officials, from which emerged the first poetic movement in an Italo-Romance variety. However, while certainly implying the existence of a language of communication underlying the literary one, the Sicilian of 6 Here and elsewhere, unless otherwise indicated, all English translations of citations from works in Italian are due to the author. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 19 the Scuola is what Dante Alighieri (De vulgari eloquentia, I, XII, 2–4) defined as “illustrious”, a court language, purged of the most marked localisms, inspired by Latin and Provençal models (not to mention that many leading members of the Scuola were not of Sicilian origin). The fact that it was a language di scola (‘of school’) and not di popolo (‘of people’), as Alessandro Manzoni wrote in the Appendix to the Report on the Unity of the Language (1868), is confirmed by the total lack, during the entire Swabian era, of attestations of Sicilian in documentary and practical writings that instead precede or accompany the poetic production of other Italian regions (Alfieri 2013: 575). It should also be considered that these texts are known – to us as well as to Dante – through Tuscan versions, while the original Sicilian facies is transmitted to us by few witnesses (cf. Spampinato 2013 for an overview). Consequently, the nature of the language of use remains to be all but defined and to date only intuitable in some of its aspects. In fact, genuinely Sicilian texts can only be traced back to the 14th century, thus in the first decades of the Aragonese presence (1282–1516). At this stage, we finally find ample documentation of the use of Sicilian, both in literary production and in public communication, until the mid-15th century, when it begins to be replaced bu Tuscan (see 1.2.2.2). As far as poetry is concerned, the documentation is actually not very dense (e.g., Quaedam profetia or Lamento di parte siciliana, about 1354, cf. Spampinato 2013: 785–786). Prose production, on the other hand, is rich. An excellent overview of medieval prose texts is due to Pagano (2013: 792, from whom I derived the following information; cf. also Alfieri 2013), who explains that these are mostly vulgarisations (in other words, transpositions into the volgare, i.e., the local language) from Latin, Tuscan or Catalan, while the originals are much sparser. The most ancient vulgarisations, produced in the first decades of the 14th century in the milieu of the court of Frederick III, are respectively devotional, pedagogical and cultural in subject, namely: the Dialagu de Sanctu Gregoriu (Santangelo 1933), in which the Franciscan friar Giovanni Campulu from Messina translated The Dialogues of Saint Gregory the Great (540–604); the vulgarisation of the Factorum et dictorum memorabilium libri of Valerius Maximus, due to the Messina master of rhetoric Accursu di Cremona (Ugolini 1967); and the Istoria di Eneas (Folena 1956), in which another master 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 20 of rhetoric from Messina, Angilu di Capua, translated a Tuscan vulgarisation of a Latin epitome of the Aeneid. From the mid-14th century onwards, there is a general impoverishment in the quality of prose production, which seems to be linked to lower socio-cultural and sociolinguistic contexts, compared to the previous period. Probably dating from the 1370s is the Sposizione del Vangelo della Passione secondo Matteo (Palumbo 1954–1957), a commentary on some chapters of the Gospel of Matthew, an original work perhaps due to the franciscan Niccolò Cusacchi and addressed to the lower clergy, who were unable to understand Latin texts. Many other texts in the period and the next century are also religious in subject, namely: A) Treatises on religious and spiritual subjects, such as the late 14th -century Libru di li vitii et di li virtuti (Bruni 1973) and the Libru di lu transitu et vita di messer sanctu Iheronimu (Di Girolamo 1982), from the first half of the 15th century, both vulgarisations from Tuscan; the Epistula di miser sanctu Iheronimu ad Eustochiu, a late 15th century Sicilian version of a Tuscan vulgarisation from Latin (Salmeri 1999); the Meditacioni di la vita di Christu (Gasca Queirazza 2008) and the Munti della santissima oracioni (Casapullo 1995), both from Tuscan; other vulgarisations are the Soliloquia animae ad Deum (Pagano 2009) and, from Catalan, the Tractatu de savietati (Milana 2004); the Scala di virtù e vita de Paradiso is instead an original in Sicilian from the end of the 15th century. B) Confessions, such as the Ordini di la confessioni “Renovamini”, a vulgarisation from Tuscan of the Speculo by St Bernardine of Siena (Luongo 1989); four confessionals dating back to the 15th century and derived from Latin models, which propose question forms organised by type of sins (Branciforti 1953); and the Alfabetin, a vulgarisation of a Hebrew text that maintains the Hebrew characters despite the language being Sicilian (Sermoneta 1994). C) Religious rules and statutes, which regulate life within monasteries and confraternities, e.g., the Sicilian translation of the Regula di Santu Benedittu abbati, dating 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 21 from the end of the 15th century (Branciforti 1953); or the Regole di Santa Chiara (Ciccarelli 1983). D) Lives of saints, mainly vulgarisations, but also original writings, such as the Vita di San Corrado (Rotolo 1995), dating from the end of the 14th century or the beginning of the following century, and the late 15th -century Leggenda della Beata Eustochia; among vulgarisations, the Vita di Sant’Onofrio deserves attention, which, like the St Conrad, was elaborated in Franciscan circles and of which we have a 14th century version (Palma 1909) and a 15th -century one (Lorenzini 1995); and the vulgarisations of the Iacopo da Varazze’s Legenda aurea, in particular the lives of St. Christina, Sts. Cosmas and Damian (based on a Catalan vulgarisation), and St. Eutachio (Pagano 1999, 2004, 2011). In addition to devotional prose, we find scientific-like and historical-chronological texts. The former are mainly treatises on farriery (mascalcia), that is, works dedicated to the care of horses (although they sometimes include suggestions for humans). They follow in the wake of Frederick II’s experience, as they mainly vulgarise the De medicina equorum by Giordano Ruffo, one of the most prestigious members of the emperor’s entourage. The first translation into Sicilian dates from 1368 (De Gregorio 1905). Compared to the model, vulgarusations expresse a more socio-culturally lower culture, including, for example, magical practices. This holds true also for the Thesaurus pauperum, which vulgarises at the beginning of the 15th century a collection of recipes attributed to Pietro Ispano (Rapisarda 2001). Historiographical texts that deserve attention are the Rebellamentu di Sichilia (Barbato 2011), a chronicle of the Sicilian Vespers based on a Tuscan original and datable to the second half of the 14th century, and, of a more moralistic type, the Conquesta di Sichilia, which translates Goffredo Malaterra’s De Rebus gestis (Rossi-Taibbi 1954). As mentioned, at the same time as the first vulgarisations in the age of Frederick III, Sicilian also appeared in documentary writings, alongside Latin. The first attestation is a fiscal provision of 1320, the Capitula super cassia (Rinaldi 2005: 8–13), in which Frederick III issued a new tax (cassia) to meet war expenses. From the middle of the century onwards, the Sicilian gained more and more space in written communication, as the letters 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 22 written by the island’s nobles show, based on a precise political project of autonomy from the central power. From the end of the century, however, there is a gradual thinning out of public production in Sicilian, which is instead maintained in the epistolaries (Rinaldi 2005: xi–xii; cf. also Musso 2013: 818–819). In this context, the letters of Queen Eleanor of Anjou, wife of Peter IV of Aragon, sent from Catalonia between 1351 and 1375, are particularly relevant because they attest to a not only island use of Sicilian (Musso 2013: 820). Among other documentary texts, we find court documents, letters of exchange, inventories, wills, books of accounts etc. These texts show various sociolinguistic characteristics and also phenomena of contact with other vernaculars in the repertoire of the writers. However, on the whole, they show a certain morphosyntactic stability, “with firmly Sicilian outcomes common to the literary scripta and phenomena generally converging with ancient Italian” (Musso 2013: 822). A large part of the texts just presented have been published since 1951 in critical editions in the “Collezione di testi siciliani dei secoli XIV e XV” (Collection of Sicilian texts from the 14th and 15th centuries) of the Centro di studi filologici e linguistici siciliani of Palermo. The “Collezione” has 27 volumes to date (which are listed on the website: https://www.csfls.it/res/edizioni/catalogo/collezione-testi-siciliani-dei-secoli-xiv-xv/). The works of the Centro then converged, in editions that were eventually rechecked and updated, in the Corpus ARTESIA (Pagano et al. 2020). This was developed at the University of Catania in the early 2000s with the aim of publishing texts from the earliest attestations of the 14th century up to the first half of the 16th century, a period in which, as already mentioned, Sicilian was progressively replaced by Tuscan as the language of administration and public communication (http://www.artesia.unict.it/). The Corpus is hosted by the OVI (Opera del Vocabolario Italiano) and managed by the GATTO software (Gestione degli Archivi Testuali del Tesoro delle Origini), a programme developed for the Tesoro della Lingua Italiana delle Origini (TLIO). In the latest update (2020), 732 texts are indexed, with a total of 1,287,535 occurrences and 75,441 forms. In addition to the texts in the “Collezione”, the Corpus contains various texts published in different editions and unpublished texts accessible only in electronic editions, which sometimes replace old editions deemed philologically unreliable (http://artesia.ovi.cnr.it/). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 23 1.2.2.2 Beyond the Middle Ages (16th to 19th century). In contrast to Early Sicilian, texts from the modern (16th to 18th century) and earlycontemporary (19th century) ages are not yet collected in a single corpus, although we do have a (indeed conspicuous) number of individual editions for these centuries. Given the volume of production available and since, on the other hand, the texts from this period are not all equally central to the research presented in this volume, a general excursus will suffice here, with some focus on the texts I have used for checking linguistic facts in the diachrony of Sicilian, up to the threshold of contemporaneity. The most striking aspect that characterises the linguistic history of the island from the 16th century onwards is the sociolinguistic reconfiguration of the repertoire that makes Sicilian shift from language to dialect. As Alfieri (2013) explains, we are dealing with a de facto trilingualism, in which Castilian, imposed by the new Habsburg rule (1516–1713), was the leading language for institutional uses, while Sicilian and Tuscan competed in function and prestige until the transformation of bilingualism into diglossia, as in every other area of the peninsula (see also Lo Piparo 1987: 735). In other words, “Latin vs Romance bilingualism was now articulated between Latin and Tuscan and no longer between Latin and Sicilian” (Alfieri 2013: 569).7 Tuscan had begun to circulate as early as the Angevin-Aragonese period (as attested by the 14th - and 15th -century vulgarisations described in 1.2.2.1), also fostered by the spread of a Florentine bureaucratic-mercantile stratum throughout the Angevin territory, and on the other hand due to the clergy’s activity of popularising pauperist ideology, favoured by the Angevin rulers (Alfieri 2013: 578). Now, the spread of Tuscan fits into that “question of language” that characterises Italian intellectual debate of this epoch and that sees, starting with Bembo’s Prose della volgar lingua (1525), the establishment of Tuscan (better, of Florentine) as the model of the common language, 7 But within a more multifaceted linguistic space, shaped largely by the new territorial and immigration policies promoted by the Castilian government, which resulted in the presence of groups from northern Europe (Spanish, German and Flemish), placed at a high socio-economic level and, at a much lower level, Calabrian labourers employed in the countryside, especially but not only in the north-east of Sicily, and North African immigrants employed in domestic or rural service (Alfieri 1996; Alfieri and Iannizzotto 2013a: 589). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 24 of Italian. Thus, observes Varvaro (1984: 280, recalled by Alfieri and Iannizzotto 2013: 588), the Tuscan contribution to Sicilian linguistic history is different from Catalan and Castilian in that it is not conveyed by political power, but produced by a pre-existing situation of not exclusively literary prestige. As a result, Sicilian receded from public discourse and education (the latter entrusted to the clergy or private preceptors, but still in Latin), but survived in literary production, especially in manuscript literature (while printed literature favoured Latin or Tuscan), mainly in lyric poetry and prose on religious subjects (Alfieri and Iannizzotto 2013a: 592). In the sphere of lyric poetry, the Sicilian option is adopted by Antonio Veneziano (1514– 1593), from Monreale (Palermo), who was strongly inspired by Petrarchism, the fundamental model of 16th - and 17th -century Sicilian poetry. Alongside this, another current developed, aimed at a broader and less cultured audience and which had one of its expressions in the anonymous poem, known in many variants, La barunissa di Carini. In the next century, a text entirely in Sicilian is La notti di Palermu (1638), a play by Tommaso Anversa (Mistretta, Messina), also author of Lu primu tomu di l’Eneidi di Virgiliu. Rhymes in Sicilian were written by Simone Rau, from Palermo (1609–1659), in the context of Baroque lyrics, and by Paolo Maura (Mineo, Catania, 1638–1711). (See Zago 2013a for an overview of literary production in Sicily of the 16th and 17th centuries.) Even in the Illuminist Sicily of the brief Savoy (1713–1720) and Austrian (1720– 1734) dominations that followed the end of Spanish rule, and again in the first phase of Bourbon Sicily (1734–1860), we find, alongside a largely Tuscan-Italian production, a poetic production in Sicilian. This is represented first and foremost by the Palermo-born Giovanni Meli (1740–1815), author of La fata galanti, L’origini di lu munnu, Favuli morali inter alia; and by Domenico Tempio (1750–1821), from Catania, author of La carestia as well as numerous poems of an erotic nature. (See Zago 2013b for an overview of literary production in Sicily in the 19th century.) Apart from their literary value, which is more or less high, these works are worthy of note here because they constitute an important source for the study of language, which, as mentioned above, had lost ground in public and non-literary production. However, there is a parallel production in the 15th -18th centuries, consisting of epistolary exchanges, both 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 25 official and private, of practical writings, such as account books, and of texts of a bureaucratic nature, which testify to the multilingual and diamesically and diastratically composite nature of communication in these centuries. These texts, published in sparce or unpublished editions, have been studied by Sardo (2008) from the perspective of language contact and non-native language acquisition. Sardo defined these production as “interwritings”, a term that evidently echoes that of “interlanguage”, a key concept in theories of second language acquisition: “Indeed, they show recurring phenomena that refer to acquisition strategies among typologically related languages, whereby the writer activates hypotheses of congruence, correspondence, difference between the source and target language(s), as well as universal simplifying strategies” (Sardo 2013: 615). From this perspective, too, it can be observed how Spanish gradually gives way to Tuscan as the “target language”, in a process that has its fulfilment in the 1652 Viceregal Prammatica, which states that all public acts must be published in the lingua volgare, which means Tuscan.8 The different target languages interfere to varying degrees, also in relation to the sociolinguistic background of the writer, in the Sicilian interwritings, which thus represent a tool of great interest for linguistic research. In the 19th century, the sociolinguistic polarisation between Italian language vs. Sicilian dialect had been fully accomplished and would be formalised with the foundation of the Unified State in 1861, when Italian became the national language. Of course, Italian remained for a very long time a second language, learnt from books and therefore used by a small percentage of the population and, in any case, not always fully even by educated speakers, as the accounts of travellers testify (cf. Alfieri and Iannizzotto 2013b: 642; see De Mauro 1963: 37 about the slow spread of Italian, here discussed in 1.2.1, fn. 2). The locally spoken language is Sicilian. Alongside the literary production in Sicilian (which now also includes prose and to which it is not even partially possible to refer here, cf. Alfieri et al. 2013 for an overview), there are other texts that are very useful for linguistic analysis and in fact widely used for this purpose. I refer in particular to the collection of 8 This text certainly had a great cultural impact, if we consider the forms of communicative circulation of prammatiche, which were bandiate, that is cried out by the town crier, and then posted at the town Loggia, where they could be read and translated (Sardo 2013: 616). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 26 four hundred folk narratives (fables, amusing tales, historical and fanciful traditions, proverbs and sayings, and fairy tales, including variants), field-collected throughout Sicily and then transcribed and brought together by Pietrè in Fiabe, novelle e racconti popolari siciliani (1975). Despite some internal variations, these texts show an overall uniform morphosyntactic facies and a certain consistency or, at least, contiguity with today’s language that makes them classifiable as early contemporary Sicilian (cf. Brucale and Mocciaro 2017: 299). On the other hand, they differ, in terms of textual types and genres, both from the texts already examined (literary or bureaucratic or practical) and from those attesting to contemporary spoken Sicilian (1.2.2.3). 1.2.2.3 Contemporary Sicilian. The linguistic data available for contemporary Sicilian were mostly collected as a part of the Atlante linguistico della Sicilia (Linguistic Atlas of Sicily, ALS). This is a project launched in the 1980s at the University of Palermo, inspired by the experiences of the linguistic atlases developed in the Romance framework, but which systematically incorporated the instances of sociolinguistics (D’Agostino and Ruffino 2005; Ruffino 1995; Sottile 2018, 2019). As D’Agostino and Ruffino (2013) explain, ALS has been conceived and structured as a tool for documenting the use of spoken Sicilian (and the regional Italian of Sicily) by integrating horizontal (diatopic) and vertical (sociolinguistic variability) per- spectives. The Atlante contains two main sections, one ethno-dialectal and one socio-variational. The former collects ethnotexts, i.e., texts that are generally rather extensive, whose production has been stimulated by the interviewers’ questions (hence is semi-spontaneous) and are mostly monological and narrative in character, as the speaker narrates or describes facts, experiences, traditions (Paternostro and Sottile 2010: 603). The sociovariational section includes induced narrative speech, collected through an ad hoc questionnaire that stimulates semi-spontaneous conversations, some of which use the “blocked code” (in the last question of the questionnaire, respondents are asked to produce five minutes of speech in Italian and five in Sicilian). The section also contains the answers to the metalinguistic 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 27 retro-translation questions (from Sicilian to Italian and from Italian to Sicilian). A strong micro-areal sensitivity characterises both sections, becoming structural in the sociovariational one, with a systematic attention to different spatial areas, sub-areas, points at which linguistic data are collected. To date, ALS speech covers over 50 points and microareas, for a total of 1,200 surveys. This micro-spatial approach is complemented by a focus on the internal articulation of individual areas, taking the family as the basic unit, which is understood as a set of generations linked by a relationship of cultural and linguistic transmission (D’Agostino and Ruffino 2013: 1441). This means that within each area or microarea, different family types are investigated. In each type (a total of five), the linguistic-social complexity of the point is reproduced by the intersection of three generations and the variables “education” and “first language”, for a total of sixteen informants per collection point, plus two adolescents (strangers to the five families), who differ in their level of education and their parents’ first language.9 Although they differ somewhat in terms of methodology and research aims, the two sections share some fundamental assumptions. Firstly, the texts are in fact conceived as broad contexts in which the individual linguistic datum is to be placed and studied (Paternostro and Sottile 2010). Moreover, the linguistic representation systems of the two subcorpuses coincide, as they employ the same phono-orthographic and conversational transcription systems. A large part of the ethnotexts has been published in the various volumes of the series “Materiali e ricerche dell’Atlante Linguistico della Sicilia” (Materials and tools of the Linguistic Atlas of Sicily) as part of individual research on specific (linguistic and/or ethnological) facts. The data in the socio-variational section, on the other hand, are largely unpublished. However, most ALS texts have been poured into the Archivio sonoro delle parlate siciliane (Sound Archive of Sicilian Speeches; cf. Matranga 2011, 2016), which currently comprises around 5,000 hours of recordings, on various types of media, made over a period of thirty years in most centres in Sicily. Work is underway on the labelling of socio- 9 Further details will be provided in the individual articles where these data form the basis for the analysis. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 28 variational data and an overall digitisation and computerisation of all ALS data, with a view to an online searchable database in the medium term. 1.3 Research on modality and modal means in Sicilian. 1.3.1 The dialectology approach. The dialectological tradition has been mostly devoted to the morphological and syntactical description of moods or verbal constructions that specifically characterise the facies of Sicilian varieties compared to Italian or other Italo-Romance varieties. It would suffice here to refer to the various passages that punctuate the pages of Rohlfs’ fundamental Grammatica storica dell’italiano e dei suoi dialetti (1968: 301–306, 341–342, 349–356; 1969: 55–56, 61–69, 141–143). In this work, attention is also paid to the semanticfunctional characteristics of the forms described, in particular in delineating the uses of the moods, or their absence. Just think, for example, of the notations on the optative expres- sion: In those parts of southern Italy where the imperfect subjunctive replaces the conditional [...], this subjunctive is also used to express a discrete request, or a bland assertion [...], Sicilian vulissi chi currissimu ‘s/he would like us to run’ […], mi facissi stu piaciri ‘do me this favour’. (Rohlfs 1969: 67) Or, further on, the remarks about the “subjunctive of desire introduced by conjunction”: For those parts of southern Italy where the present subjunctive10 is replaced by the indicative, we cite the introduction through the preposition mu (mi) in southern Calabria, mi in northeastern Sici- 10 The present subjunctive has now generally disappeared in Sicilian and the South in general. Cf. Rolhfs (1969: 301–304). See also the fundamental observations in the more recent study by Loporcaro et al. (2018), which is focused on Pantesco, i.e. the variety of Pantelleria, and is conducted in the light of the ‘autonomous’ morphological model à la Maiden (1992; see also Maiden et al. 2011). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 29 ly, [...] Sicilian mi si setta ‘sit down!’, mi trasi ‘get in!’, mi veni cca ‘come!’, mi nun nèsci ‘don’t come out!’. (Rohlfs 1969: 68) And various other notations, most of the time only incidental, are also encountered elsewhere, e.g., on the necessive value of the type HABEO AD CANTARE, “a sort of future tense in which an idea of necessity is still somewhat implied” (Rohlfs 1968: 335). The three examples cited here are not by coincidence, but reflect the great attention that the phenomena in question (or related phenomena) have traditionally received on a formal level, schematically: 1) the presence vs absence of the conditional mood, an ItaloRomance innovation from the type ‘infinitive + HABĒBAM’ (CANTARE HABĒBAM > Anc. Fr. chantereie, Fr. chanterais, Anc. Sic. cantaria) or ‘infinitive + HABUI’ (CANTARE HABUI or Vulgar Lat. *HĔ(BU)I > It. canterei), which is not attested in present-day Sicily, except in a restricted area in the north-east of the island); and, related to this, the hypotheticalconditional structures (periodo ipotetico, cf. Rohlfs 1969: 139–153); 2) the particle mi ( (*cantarajo >) > It. canterò, Fr. chanterai, Sp. cantaré; cf. Rohlfs 1969: 331– 336). These are three key themes – we can certainly claim – in traditional historical dialectology and precisely for this reason the subject of renewed interest (and possibly criticism) in more recent literature (cf. 1.3.2). These themes are also addressed in another milestone of Sicilian dialectology, namely the Profile that Leone (1995) dedicated, more recently but with an equally classical approach, to the syntax of Sicilian. Compared to Rohlfs’ account, here we can find more ex- 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 30 tensive comments of a functional type, as in the case, for instance, of the following passage on the contextual expression of subjectivity in the absence of the present subjunctive: But the present indicative also replaces the present subjunctive, which no longer exists [...]. Now, the indicative is the mood of objectively existing facts; the subjunctive, on the other hand, is that of subjectivity, that is, of desired or thought facts. The use of the indicative with the value of the subjunctive is therefore generally possible only if the context succeeds in expressing the subjective aspect of the facts, that is, in communicating that they do not have, or do not yet have, an objective reality: Scappa prima ca torna iddu (‘Run away before he comes back’), Non puozzu riri ca mi piaci (‘I cannot say that I like him’), Mi scantu ca trasi (‘I am afraid he will come in’), Spieru ca veni (‘I hope he will come’). Neither returning, nor liking, nor entering, nor coming have an objective reality, being constructions of thought: but this clearly emerges from the verb of the main sentence (‘to fear’, ‘to hope’) or from the sentence as a whole. (Leone 1995: 38–39) And further on: Going back to the use of the indicative instead of the subjunctive [...], it still remains to be considered what semantic difference is established, in dependent proposition, between these moods, if both are possible. Indeed, Mi scantu ca trasi (‘I am afraid he comes in’) can be said, with reference to the past tense, Mi scantava ca trasia (‘I was afraid that he came’) (still with the indicative), but also Mi scantava ca trasissi (‘I was afraid that he would come’) (since the imperfect also exists in the subjunctive). Well, compared to the indicative, the subjunctive makes the eventuality of the hypothesised or feared fact feel more remote. (Leone 1995: 39) Returning to the level of linguistics forms, these seminal works also offer acute observations, on the one hand on constructions occupying more peripheral areas of the modal domain, see for instance the attention devoted to the directive construction va ‘go’ + imperative, described several times, mainly in formal terms, by Leone (1973, 1980: 130–131; 1995: 44–45; cf. also Rohlfs 1969: 166–167; cf. also Sorrento 1950; Tropea 1988; Varvaro 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 31 1988; and, framed in a rather theoretical than descriptive perspective, Sornicola 1976). On the other hand, it is worth mentioning the interest in alternative means of modal encoding (to which we will return in 1.3.3). These are by and large lexical markers that developed various modal nuances, such as frozen relics of the present subjunctive, e.g., nzam(m)ai or maisìa (literally meaning ‘never it be’, used with the deprecative function ‘heaven forbid!’); mpàzzica (based on the subjunctive of ‘to do’, i.e., literally ‘s/he doesn’t do that…’ with the same deprecative function as maisìa); pozza (literally, ‘it may’, hence ‘eventually’) (Leone 1980: 58–60, 1995: 33–34, 46–47; Tropea 1988: LXXVII, fn. 292); epistemic markers resulting from the crystallisation of entire clauses, e.g., pènzica (penzi + ca ‘think that’, hence ‘maybe’), capacica (capaci + ca ‘possible that’, hence ‘maybe’); or epistemicconditional markers, e.g., cusà (literally, ‘who knows; eventually; if’), siddu/suddu (‘if’ < Lat. SI + ILLUM), a li voti (‘sometimes’), macari (‘at least’) (Leone 1995: 34, 46; Rohlfs 1969: 184, 284; Ruffino 2001: 62; Sorrento 1920: 195–198; Varvaro 1988: 723). 1.3.2 Dialectology and beyond: three examples. The ideas and insights offered by the influential dialectological tradition described in the previous section have inspired the research conducted in subsequent years by several scholars who have been variously interested in modality and modal encoding, often combining dialectological interest with other theoretical-methodological perspectives. Without any claim to account for the wealth of existing studies, I will describe here three areas of the modal domain and modal encoding that have received considerable attention in recent decades, also as a consequence of their central role in traditional studies, as sketched in section 1.3.1. The first important example is provided by studies on conditional forms and constructions, which, as already mentioned, have already been widely described in the traditional literature (Leone 1958, 1974, 1980: 57–58 and 61–62, 1995: 34 n. 58, 40 n. 75; Leone and Landa 1984: 84; Rohlfs 1922, 1968: 141–142 and 340–341, 1980; Rossitto 1976; Varvaro 1988: 723). Among these, mention must first be made of the fundamental research carried out by Bentley at the end of the 1990s (1997a, 2000a, 2000b; Vincent and Bentley 1995), 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 32 which highlighted, already in the scripta of the 14th and 15th centuries, the presence of a conditional in -ìa, an outcome of the Romance type ‘infinitive + HABĒBAM’ (“an imperfect of the future”, according to Tagliavini 1949 [1982]: 261)”. According to Rohlfs (1969: 55), the fact that this conditional is a Romance innovation, combined with the consideration that Sicily was, for the most part, neo-Romanised after the collapse of Arab rule (Rohlfs 1968: 340; see about this issue 1.2.1 fn. 5), must exclude the endogenesis of the form, which would rather derive from a non-indigenous linguistic stratum “introduced into Southern Italy only after the year 1000, probably of Provençal (or Po Valley) origin, radiated into extreme Italy through literary or Chancery influences; [which] corresponds in its linguistic character to the many Romance neologisms of this area” (Rohlfs 1972d: 257– 258). In support of this thesis, Rohlfs provides morphological, phonetic and syntactic evidence: 1) complete inflection is generally lacking for the type cantaría; 2) the phonetic aspect (pruvirría instead of pruvarría) would show an imitation of literary Tuscan (provería, with -e- changing to -i- in Sicilian; curiously, he himself observed a couple of pages earlier that TĒLA > tila in the south would allow us to trace back the ending -ía directly to the Latin HABĒBAM > Sic. avìa); 3) in the other Romance languages the conditional in -ìa developed parallel to the future tense, but the latter would be however unknown in Southern Italy (Rohlfs 1969: 339–342; see Assenza 2018 for a detailed discus- sion). The conditional type in -ìa is actually absent in most of present-day Sicilian varieties, which therefore exhibit in the hypothetical structures a symmetrical morphosyntax in the protasis and apodosis, mostly with double subjunctive (Si pputissi, ‘u facissi ‘If I could I would do that’) or double imperfect (Si pputia, ‘u facia ‘Id.’). Only in the area of Messina and in some Gallo-Italic islands, the conditional may appear at the same time in the protasis and apodosis (Cci parrirìa ìu, s’u truvirìa ‘I would talk to him if I could find him’). Based on this distribution, Bentley (2000a: 6) observes that, [o]ne could argue that the prevalence of symmetrical types in the Sicilian dialect is a conservative trait, at least as far as the double subjunctive is concerned. In fact, in classical Latin the subjunctive appeared in both the protasis and the apodosis. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 33 However, she continues, The analysis of the evolution of conditional constructs in Sicilian does not seem to confirm the hypothesis formulated above, since in Old Sicilian we generally find the imperfect or pluperfect subjunctive in the protasis and the conditional in the apodosis. [...] The evidence from 14th and 15th century texts, therefore, suggests that the morphologisation of infinitive + HABĒBAM took place in Sicilian and that today’s structures do not derive directly from Latin. (Bentley 2000a: 7) Thus, in contrast to Rohlfs (1968: 339–342; see also 1969: 142; 1972d: 257–258) and on the basis of formal and textual observations (for which cf. already Schiaffini 1929), Bentley (2000a: 7, 14) argues for the authenticity of the paradigm in -ìa and hypothesises that its use in the hypothetical construction, observable as early as the 14th and 15th centuries, was specialised in the courtly and written domains, while the symmetrical construction with double subjunctive became characteristic of the informal varieties and survived, in fact, the former. This analysis, in addition to suggesting an uninterrupted development from Latin, “albeit uneven on the diaphasic and diamesic planes”, allows us to “place the case of Sicilian within the much broader framework of more common pan-Romance (cf. the emergence and expansion of the conditional) and interlinguistic (cf. the role of the imperfect indicative) tendencies” (Bentley 2000a: 14; cf. Comrie 1986). Bentley’s research inspired numerous other studies in the following years, which, by means of synchronic data collected in the field or by checking new diachronic data, confirmed the results or added important details. Among these works, mention should be made of those conducted in the socio-variational perspective of the ALS (Atlante Linguistico della), such as Castiglione (2004), who analyses the answers to the translation questions of the questionnaire (cf. 1.2.2.3), and Amenta (2007), who combines the sociolinguistic perspective with a functional one. Moreover, in this second work, the analysis of the areal diffusion of conditional constructions is substantiated by a survey of texts from other ages. This makes it possible to observe, in narrative texts of the 14th and 15th centuries (from the “Collezione”), the presence of the conditional in the apodosis of the hypothetical constructions of possibility and unreality and, on the other hand, the total absence of symmetrical 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 34 systems, both with double subjunctive and double conditional. On the other hand, symmetrical systems alternate in 19th century texts (Pitrè 1875). According to Amenta (2007: 511), this marks a significant distancing of contemporary Sicilian from the system of ancient Sicilian along an evolutionary line that had begun in 19th century texts. The informality of the register – she continues – may certainly have contributed to favour the use of symmetrical systems: Therefore, the expression of possible conditions in ancient Sicilian differs both from Latin and contemporary Sicilian, as far as the double imperfect subjunctive is concerned, and from the system of sub-standard Italian, in which hypothetical complex sentences with the double conditional are present. The only attestations with the conditional present in ancient Sicilian are apodosis in conjunction with protasis with the imperfect subjunctive, as is also the case in standard Italian. [...] As far as the unreal conditions are concerned, in the protases of the 14th -15th century attestations the subjunctive trapassato recurs homogeneously, while the apodoses are differentiated by the presence now of the conditional past tense, as occurs in standard Italian, now of the indicative imperfect or the passato remoto [i.e., the synthetic perfective past, EM]. (Amenta 2007: 507–509) Against a similar dialectological and socio-variational scenario, we find the accurate descriptions of double conditional constructions in north-eastern Sicily provided by Assenza (2012, 2018, 2021). She exploits the synchronic data obtained to once again question the hypothesis of the exogenous and literary origin of the conditional in -ìa, contesting point by point the so-called Rohlfs’ evidences, starting from the first morphological hypothesis, which was contradicted by new surveys that clearly showed complete paradigms for condi- tional.11 11 Cf. on the same ground the phonetic remarks by Barbato (2007: 180 and n. 256, who refers to Debenedetti 1932: 29) on the language of the Rebellamentu; but cf. also Loporcaro (1999: 76) on the Tuscan influence in Southern texts. Loporcaro, on the other hand, shows unequivocally the residual presence of the future ‘infinitive + HABEO’ in Southern Italo-Romance varieties, whose alleged absence would match that of the conditional according to Rohlfs (1968: 339–340). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 35 Similarly, De Angelis (2008: 97), who however, still referring to Rohlfs (1972), proposes that the double hypothetical conditional (si putí(v)a u fací(v)a) is “a syntactic calque of how the hypothetical period of the unreality of the present is constructed in ancient Greek, precisely through the imperfect indicative”: Ultimately, in both Calabria and Sicily the construction si purría u faría is documented in (or near to) areas of strong Greekisation and coexists in close contact with the imperfect double indicative construction. [...] And thus the genesis of the type si purría u faría is to be sought, at least in these areas, in the formal and structural affinities that the conditional in -ía shares with the imperfect forms; the dropping of -/v/- in the imperfect forms especially of second and third conjugation in the extreme southern varieties (Rohlfs 1968: 291), led to a remarkable formal similarity between the conditional forms in -ía and those of the imperfect, also in -ía; and just as the imperfect indicative was employed in both parts of the hypothetical period, in the same way the conditional was in certain cases extended to both protasis and apodosis. (De Angelis 2008: 98–99). Several of the above-mentioned authors have also dealt in recent years with the marker mi (mu/ma, in Calabrese) + indicative, which is the second case given here as an example of studies that combine dialectological interest and current models of analysis. This form is traditionally traced back to the Latin adverb MŎDŎ12 “in independent and dependent propositions, [...] characteristic of Calabrian-Sicilian dialects (mi trasi: ‘come in’; ci dissi mi trasi: ‘I told him to come in’), to express wishes and exhortations” (Sorrento 1920: 184; cf. also Sorrento 1912; 1950: 364–398) and, as Rohlfs (1972c: 334–335) points out, also in other important functions, subordinative rather than just imperative. Rohlfs relates the development of this conjunction with the loss of the infinitive in dependent clauses “in the 12 On the subordinator MŎDŎ, cf. Rohlfs (1972b: 327; cf. also Id. 1933: 50 n. 1), about the corresponding Calabrian form: “This mu represents the Latin MODO ‘promptly’ being identical to the It. mo ‘promptly, just now’, so that the sentence vogghiu mu dormu primitively meant ‘I want and already sleep’. We are dealing with a substitution of the infinitive, with a type of subordination that originally had the value of a coordination”. Cf. however Meyer-Lübke (1881, I, § 551). Alessio (1964) proposed an origin from MŎDŎ UT concessive. Radically alternative hypotheses are advanced by Damonte (2009), disputed on solid morphonological grounds by De Angelis (2016); see also Prantera and Mendicino (2013). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 36 Greek dialects of southern Italy (Calabria and Salento) and in the Italian dialects of those southern Italian provinces that at the time of Byzantine domination (and partly still in later centuries) must already have had, ex temporibus antiquis, a Greek or at least bilingual population, i.e., a population that in addition to Italian (or Italian dialect) also spoke Greek or vice versa” (Rohlfs 1972b: 318). It is not by chance that in Sicily, mi is used in the north-eastern triangle of the island (Sorrento 1912; Tropea 1965; cf. also Sornicola 1992 for some 15th century forms) and has as its “nucleus of maximum diffusion the territory between Messina, Taormina and Naso; this is precisely the territory where today’s local dialects preserve the greatest percentage of lexical Greekisms, a clear indication that also in Sicily the use of mi (as imitation or cast from the vulgar Greek νά ἵνα) developed in the environment of a bilingual population: again, Romance substance in a Greek spirit!” (Rohlfs 1972c: 338). The subject was revisited, at the beginning of the new century, by Damonte (2005), who tested a sample of speakers from north-eastern Sicily on the acceptability of sentences pre-constructed with mi. Damonte’s study is reviewed by Assenza (2005, 2008), who partially questions its procedural validity. Assenza (2008: 105) proposes to cross-check the metalinguistic data with those of semi-spontaneous speech, in order to assess what “beyond the grammatical acceptability of phenomena known from dialectological literature, comes to be configured today as obsolescent, or emerging, or winning in the linguistic consciousness and in the actual behaviour of speakers”. Read in the light of the parameters of diatopic, diastratic and generational variation, the data collected configure a situation of transition between old and new forms, characterised by inconsistencies and contradictions that open up an interesting gap between declared and actual usage: Although the grammaticality judgments formulated by older informants agree in evaluating as ‘correct’ and ‘acceptable’ – and in many cases even ‘preferable’ – all the classical occurrences of the construct with mi, the actual usages differ significantly from the evaluations expressed, configuring that phase of transition between archaic and innovative usages that we find advancing among the new generations of speakers. […] [T]he tendency is to contract its uses, reducing its syntactic contexts and preferring more modern and Italianised forms to it. (Assenza 2008: 118–119) 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 37 From a different methodological perspective, De Angelis (2013, 2016) retraces the debate on the etymology of the form mi, to arrive at the hypothesis, formulated on the basis of the typological literature and studies on grammaticalisation, that mi is a complementiser developed from the function of “marker of focus, affixed to verbal forms” covered by MŎDŎ already in Latin: a path, therefore, crosslinguistically frequent, from discourse to syntax (De Angelis 2016: 84–88). In later works, drawing on results achieved in formal syntax (cf. Ledgeway 1998, 2007, 2013 on Calabrian), De Angelis (2017a, 2017b) analyses the behaviour of mi in synopsis with the outcomes of Lat. QUĬD, also present in Sicilian (chi), and, on the other hand, the use of mi in independent sentences of volitional meaning, followed by a form of indicative imperfect to encode an invective or a curse. Finally, the development of mi in independent sentences is analysed in terms of insubordination by Ganfi (2018, 2021), who analyses first-hand data collected in the area of Messina. Insubordination is a theoretical construct that describes a process of reinterpreting subordinate phrasal structures as independent clauses (cf. Evans 2007; cf. also Cristofaro 2016; Lombardi Vallauri 2007). Through such a process, mi would acquire a modal status, both deontic (in orders, exhortations and auspices) and epistemic (in clauses expressing the speaker’s uncertainty about the actual realisation of an event, e.g., Non mi cadiu n’terra ‘Maybe he fell on the ground’). Moving to the third example in this section (which is also the one most directly related to the articles collected in this volume), we once again find Delia Bentley’s research. In addition to the studies on conditional, we also owe her a broad reflection on the relations between modality and tense, conducted through various works in the late 1990s (Bentley 1997b, 1998a, 1998b, 1998c, 2000c). In particular, she examined the functions of the construction ‘aviri a ‘have to’ + infinitive’, which coexists in ancient Sicilian texts with the synthetic outcomes of ‘infinitive + HABERE’ and replace them in later times. In addition to futurity, ‘aviri a + infinitive’ also expresses deontic and epistemic values. For this reason, the construction has never been considered in the literature as a genuine example of future (cf. Ebtener 1966: 38, who considers it a form of “emphatic future”; and Leone 1995: 36, fns 64, 65, who describes it as expressing “confident expectation”). As the natural conse- 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 38 quence of the absence of a future form, the expression of temporality in Sicilian would only be organised between the two poles of past vs non-past (Ebtener 1966: 38). To test the strength of these assumptions, Bentley conducted a careful analysis of the contexts of use of the construction, compared with the occurrences of the synthetic future in texts from different times, namely: 14th -century texts (both from the “Collezione” and other sources not influenced by the Tuscan model13 ), texts from the 19th -century (Pitrè 1875) and firsthand data from present-day Sicilian (Palermo). Results showed, on the one hand, the autochthonous character of the synthetic future (also supported by historical phonetic considerations; see also Loporcaro 1999: 109), which is the norm for the expression of the future in ancient texts (about 1100 cases against 300 of the periphrasis), but shows signs of clear decline in 19th century texts, where the periphrasis is much more widely used (about 1400 cases against 37 of the synthetic future; cf. Bentley 1998b: 122–124, 127). On the other hand, the qualitative analysis of the data supports the fully future value of ‘aviri a + infinitive’, which is not contradicted by the presence of modal nuances. More specifically, as long as the synthetic future prevails, the periphrasis has a purely deontic value (hora avimu a vidiri di l’huri sequenti ‘now we have to deal with the following hours’, Bentley 1998b: 127) and coexists in free distribution with the synonymic form diviri (cf. It. dovere ‘must, have to’ and equivalents in other Italo-Romance varieties), which will disappear in the later texts (in fact, its autochthonous character has also been questioned). Deontic value implies reference to the non-past tense not of the necessity itself but of the actions to be performed; therefore, we are not yet dealing with a predictive (epistemic) value, which would be entirely consistent with (and sometimes indistinguishable from) the futural sense. A clearly (subjective) epistemic-predictive sense is found in 19th century texts: Va taliali buonu ca cci nannu a essiri ca t’hannu a piaciri ‘Go look at them well that there must be some that you will like’ (Pitrè 1875, IV: 237), which contains a supposition about a present situation and a prediction (Bentley 1998: 130); A li tri anni […] chiantu sti favi, e vidi quantu mi nn’hannu a fari! ‘In three years, I will plant these beans and you will see how many more they will produce!’ (Pitrè 1875, II: 267), which is a prediction involving an 13 Indeed, documentary texts in most cases published later in the “Collezione”, see in particular Rinaldi (2005). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 39 inanimate subject (hence it excludes any agentive control over a future state of affairs; Bentley 2000c: 6). Bentley’s analysis is fully supported by crosslinguistic evidence of the notional link between temporality (future) and predictive modality: “What is conventionally regarded as future tense (in languages that are said to have a future tense) is rarely, if ever, used solely for making statements or predictions, or posing and asking factual questions about the future. It is also used in a wider or narrower range of non-factive utterances, involving supposition, inference, wish, intention and desire” (Lyons 1977: 816). This crosslinguistic scenario – Bentley explains – allow us to “dispel the myth of the so-called idea of ‘necessity’ implied by the periphrasis, or rather the view that the construction in question, since it always involves the modalisation of the sentence or utterance, should not be considered a future temporal structure” (Bentley 1997b: 49; see on this topic Ebneter 1966: 33–38; Piccitto 1955: 14; Rohlfs 1968: 335–336; Leone and Landa 1984: 67–71; Leone 1995: 36, fn. 65). According to Bentley (1998b: 133), in the history of Sicilian, the alternation between synthetic and analytic futures not only indicates a direction of semantic shift that has crosslinguistic parallels (from deontic to epistemic, see also Bentley 2000c), but also clearly exemplifies a phenomenon of layering (Hopper 1991: 22–24). By means of this notion, it is possible to describe the textual and sociolinguistic distribution of the two futural forms, namely the synthetic future belonged to a diastratically more formal style and, on the diamesic level, only to the written language, whereas the analytic construction, rare in the written, was presumably the prerogative of informal styles and, in fact, survived the decline of the ancient literary language. Bentley’s seminal study provided the theoretical-methodological basis for further research on the Sicilian construction ‘aviri a + infinitive’, based on data field-collected in the context of the Atlante Linguistico della Sicilia. This research has been especially carried out by Amenta (2006, 2010a, 2010b, 2020: 95–108; Amenta and Paesano 2010; but cf. already Arcuri and D’Agostino 1982) and, together with Bentley’s studies, constitutes the main antecendent of the work in this volume. Amenta’s research is part of a more general interest in the encoding of verbal categories in Sicilian, and their possible overlapping. While her early work is mainly focused on verbal moods and the modal uses of tenses (cf. Amenta 2004 on subjunctive and conditional moods in Sicilian and in regional Italian; see 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 40 also the above-mentioned 2007 work on conditional), a more precise interest in ‘aviri a + infinito’ emerges in later works (Amenta 2010a, 2010b; Amenta and Paesano 2010). Here, based on data deriving from the ALS sociolinguistic questionnaire and semispontaneous blocked code speech, the construction is analysed, together with other Sicilian periphrases (mainly aspectual), in the light of the criteria of periphrasticity proposed by Bertinetto (1990: 342–343, 1991), thus verifying the degree of internal cohesion, the non-separability of the constituents, the semantic and syntactic integration. On this background, Amenta describes the functional network of the construction, which ranges among deontic, epistemic and temporal (future) meanings (also depending on temporal selection), as already shown by Bentley, but with a more systematic focus on the sociolinguistic (diastratic and diaphasic, in particular) distribution of the construction. ‘Aviri a + infinitive’ proves to be highly grammaticalised throughout the sample, although – Amenta shows – in present-day Sicilian it undergoes renewed competition from synthetic forms due to the increasingly close relationship between Sicilian and (regional) Italian. 1.3.3 Modality and beyond: some peripheries. To conclude this survey of the existing literature on modality in Sicilian, we must at least mention some peripheries of modal expression, which open up interesting crossdomain perspectives, that is, perspectives that include different notional (modality, evidentiality) or sub-notional (epistemicity, conditionality) domains, as well as different levels of linguistic analysis (especially morphosyntax and pragmatics). Once again, I will present just a few illustrative facts of non-canonical modal marking, lexico-discursive in nature and, therefore, at the interface between morphology and pragmatics, which are at the core of a lively debate today. The first case is represented by a phenomenon already observed in traditional dialectology, namely the existence of lexical markers expressing various modal nuances (cf. 1.3.1). In various works in the last decade, Cruschina (2010a, 2010b, 2011, 2015; Cruschina and Remberger 2008) examined a number of lexical elements expressing epistemic and evidential meanings that have in common, on a formal level, the presence of a final mor- 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 41 pheme -ca, deriving from the univerbation of a verbal or adjectival form with the complementator ca ‘that’, namely dìcica (dici + ca ‘to say:PRS.3SG that’; cf. Sp. dizque, Sard. nachi, Gal. disque, Rum. cică), pàrica (pari + ca ‘to seem:PRS.3SG that’), pènzuca/pènzica (penzu/penzi + ca ‘to think:PRS.1/3SG that’), capàci ca (capaci + ca ‘possible:ADJ that’). Cruschina’s interest is first and foremost in the adverbial nature of these elements, since adverbs are traditionally said to be lacking in the southern varieties and in Sicilian, where adjectives play the same role (Rohlfs 1969: 243). In contrast, Cruschina argue for the adverbial nature of the forms in -ca, which result from a process of grammaticalisation that has affected the basic clauses (as already noted, albeit in different terms, in the traditional literature, e.g., Rohlfs 1969: 284). In this process, the development of evidential or epistemic nuances played a key role. As is well known, these are two contiguous categories, but often kept distinct in the literature: although both refer to the speaker’s knowledge, epistemicity signals the degree of confidence and certainty with respect to one’s assertions, and is thus inherently subjective and modal; whereas evidentiality specifies objectively the source of the information, i.e., the way in which the speaker came to know the reported information (cf. Aikhenvald 2004). However, other scholars admit possible overlaps, especially with regard to inferential evidentiality, which is very close to the notion of epistemicity (cf. van der Auwera and Plungian 1998; and for Romance languages, Pietrandrea 2005, 2007). Contrary to the common belief that in Romance languages evidentiality lacks dedicated markers and is, if anything, expressed through evidential strategies, Cruschina argues that dìcica and pàrica are fully grammaticalised evidential forms (as various semantic and syntactic constraints would show, and often also phonetic erosion, e.g., disca in some varieties), which typically report a common opinion or appearance, but also a rumour or gossip. On the other hand, pènzica and capacica would be – Cruschina claims – (equally grammaticalised) epistemic markers, ranging differently along a certainty > probability (pènzica) > possibility (capacica) scale. In a similar vein, we find the studies of Lo Baido (Forthcoming; Brucale et al. 2022), which focus on markers based on the verb of knowledge sapìri ‘to know’, namely sapiddu (literally, ‘knows he’, hence ‘who knows, maybe’, cf. Piccitto and Tropea 1977: 368), chi sacciu (literally, ‘what I know’) and cusà (literally, ‘who knows’; cf. Sp. quizá(s), anc. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 42 Port. quiça, Eng. who knows, Dutch wie weet; se De Smet and Van de Velde 2013; Ramat and Ricca 1998: 23). These markers are morphologically adverbs, the outcome – like those investigated by Cruschina – of a grammaticalisation process starting from broader syntactic segments. This process is connected, again, to the contextual development of modalising values, which relocate the markers in question from a merely morphosyntactic dimension to a pragmatic-discursive one. Lo Baido analyses the markers in question based on spontaneous speech collected in western Sicily, in particular in the area between Palermo and Trapani. This survey allows her to sketch the semantic network of the three markers in this area. Although similar, the values conveyed by the different markers do not completely overlap: sapiddu is often selected as a vehicle for attenuating the epistemic commitment of the speaker (and expressing indefiniteness in certain contexts) and as a medium devoted to the expression of emphasis. In its genuinely modal values, “[c]hi sacciu specialises in the activation of expectations of non-exhaustiveness, generally in argumentative contexts that place reasoning in a potential dimension in order to abstract categories from active and interchangeable exemplars with each other and with contextually relevant others” (Lo Baido Forthcoming). In the expression of epistemic values sapiddu and chi sacciu go hand in hand with another strategy, which is rather advanced in the process of change, namely cusà. In contrast to sapiddu and chi sacciu, cusà can also occur in conditional contexts and as a dubitative-corrective adverb, conveying a more specific value of eventuality (Brucale et al. 2022). At the very fringes of modality and modal encoding, we find discourse markers such as the forms investigated by Scivoletto (Forthcoming), who focused on bì ‘oh!’ (< vidi, imperative form of the verb ‘to see’), used in south-eastern Sicilian to express mirativity (and some other related functions). This is a notional category that has received some attention in linguistic typology since DeLancey (1997), which refers to surprise at an unexpected piece of information; thus, it is contiguous with evidentiality (Aikhenvald 2012). Almost never observed in the traditional literature (more interested, as we have seen, in the morphosyntactic side of modal encoding, cf. 1.3.1), bì certainly meets a new widespread interest in discourse markers (see for instance for the Romance field Molinelli and Ghezzi 2014) and the processes of grammaticalisation and pragmaticalisation that govern them 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 43 (Diewald 2011; Heine 2013). Scivoletto’s work is based on first-hand data from presentday Sicilian (both spontaneous speech and informal texts written by speakers on a selection of social media) and on diachronic data from different epochs (using the ARTESIA corpus for the 14th to 16th centuries and a selection of individual publications of poems, folk songs, etc. for the other centuries). The analysis of this corpus, especially the diachronic side, allowed Scivoletto to reconstruct the path through which the original imperative form developed a mirative meaning and, subsequently, a sense of correction/reformulation (imperative > call > mirative > correction), which is accompanied by a progressive formal reduction (vidi > vì > vih > [bɪ], with a final betacisation, which occurred in the 19th century along with the mirative shift), as is usual in processes of change leading from lexicon to grammar or pragmatics. Although they may appear marginal with respect to the themes that occupy the focus of this volume, studies such as those of Cruschina, Lo Baido and Scivoletto deserve attention both methodologically, because they employ analytical tools entirely consistent with those proposed here, and in the perspective of an overall mapping of modal expression in Sicilian. In fact, they exceed, in some cases heavily, the boundaries of morphosyntax, placing themselves more or less firmly in the domain of pragmatics and, with this, suggesting once again – and sometimes beyond intention – the lability of the boundaries between levels of analysis and the fundamentally non-modular nature of the language. 1.4 Theoretical assumptions: three key notions. 1.4.1 General framework. As mentioned in 1.1, the five articles gathered and commented on here are framed in a theoretical and methodological perspective inspired to the usage-based models. These models form a paradigm in which insights from functionalist theories and some areas of cognitivism converge. In a nutshell, these theories hold that linguistic forms are created, acquired and used in the service of communicative and classificatory functions (cf. Givón 1979; Heine et al. 1991; Hopper and Thompson 1980; Hopper and Traugott 1993 [2003] 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 44 inter al.). The emergence of linguistic forms consists in the mise en grammaire14 (i.e., the translation into grammar) of needs that develop in the interaction of human beings with each other and with the environment and that are fixed in socially shared cognitive schemas. According to Tomasello (2003: 17), this involves not only historical languages and hence the emergence of individual language structures, but the phylogenesis of language itself, that is, the emergence of grammar as a human organising ability, which is “a cultural-historical affair [...] involving no additional genetic events concerning language per se” (Tomasello 2003: 9). Continuities and discontinuities in the world’s languages can be understood precisely in the light of such a process of constructing a grammar (which he calls “grammaticalisation”): The theory of grammaticalisation is able, at least in principle, to account for both the similarities between the world’s languages – based on cognitive, vocal-auditory information processing and pragmatic inference capacities that are common to all peoples with regard to social and communicative goals – and the fundamental differences between these languages, as different language communities use and grammaticalise different discourse sequences. The emergence of grammar is replicated at the ongenetic level, in the acquisition of language by individuals. Here again, functionalists (e.g., Bybee 2006, 2007) and cognitivists (in particular cognitive grammar and construction grammar linguists, cf. Goldberg 1995, 2006; Langacker 1987, to whom the label “usage-based” is actually due) share the hypothesis that grammatical structures are cognitive representations that generalise specific utterances realised in discourse, based on frequency criteria. 14 This expression is due to Anna Giacalone Ramat (1992), who uses it to distinguish the general process of construction of linguistic-grammatical systems as a whole (an acquisitional grammaticalisation) from the phenomena of grammaticalisation stricto sensu that concern individual phenomena of linguistic change. In this volume, the term “grammaticalisation” is used in the narrower and technical sense of emergence of a grammatical construction. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 45 While all linguists are likely to agree that grammar is the cognitive organization of language, a usage-based theorist would make the more specific proposal that grammar is the cognitive organization of one’s experience with language. As is shown here, certain facets of linguistic experience, such as the frequency of use of particular instances of constructions, have an impact on representation that we can see evidenced in various ways, for example, in speakers’ recognition of what is conventionalized and what is not, and even more strikingly in the nature of language change. The proposal presented here is that the general cognitive capabilities of the human brain, which allow it to categorize and sort for identity, similarity, and difference, go to work on the language events a person encounters, categorizing and entering in memory these experiences. The result is a cognitive representation that can be called a grammar. This grammar, while it may be abstract, since all cognitive categories are, is strongly tied to the experience that a speaker has had with language. (Bybee 2006: 711) Therefore, grammar and grammar usage are mutually dependent and largely overlap in linguistic analysis. As Bybee (2006: 730) summarises, “[u]sage feeds into the creation of grammar just as much as grammar determines the shape of usage”. The same processes govern language change, which is recognised as inherent to language, has a gradual character and “takes place as language is used” (Bybee 2013: 50). An important consequence of this understanding of language change – and of language more in general – is that “the categories and units of language are variable and […] form gradient rather than strictly bounded categories. Thus, linguistic structure is viewed as emergent – governed by certain regular processes, but always changing as it is re-created in the individual and in specific usage situations”. Thus – Bybee (2013: 50) continues, rather than a fixed, static set of representations, language is viewed as being affected by experience in an ongoing way even in adults. It also follows that we should not expect linguistic constructs such as segment, syllable, morpheme, word, or construction to have strict definitions, nor do we expect all the manifestations of these constructs in languages to exhibit exactly the same behavior. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 46 In her exemplar-centered approach, Bybee (2006: 717) proposes that new tokens of experience impact memory and the organisation of language categories to varying degrees, resulting in clusters. These exemplar clusters are categories that exhibit prototype effects, in other words, they are organised in terms of members that are more or less central to the category, rather than in terms of categorical features. The prototypical approach (dating, as is well known, from the cognitive psychology of Eleanor Rosch, e.g., 1971, 1973, 197515 ) is indeed central to the models described here, which have adopted it from the founding phase (e.g., Lakoff 1987; Taylor 1989). We can schematise the prototypical categorial organisation as in Figure 1. Figure 1. Prototypical categories. As Figure 1 shows, each category consists of specimens occupying different positions from the prototype (respectively, X and Y). Leaving aside the lengthy debate on what the prototype is, it is commonly assumed in the theoretical models we are dealing with that prototypicality derives from the frequency with which a certain specimen (or a certain 15 Partly indebted, however, to Wittgenstein‘s (1953) family resemblance model. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 47 cluster of properties) embodies a given category in a given language.16 Figure 1 also shows how, in this scalar view of categories, categorical boundaries are anything but clear-cut and leave room for various phenomena of transition and categorical ambiguity. This is the case with Xn, which also occupies a portion of Y, as a peripheral exemplar of this latter category; and of Yn, which crosses the boundaries of the category to which it belongs. The lability of the categorical boundaries illustrated by Figure 1 also affects the levels of linguistic analysis, in that it is common for specimens such as Xn to be at the crossroad between a lexical category and a grammatical category (e.g., Fr. avoir or It. avere, which are both verbs denoting possession and perfective auxiliaries involved in the compound forms of the past). In other words, in the model under discussion, the relationship between lexicon and grammar (as well as that between grammar and pragmatics, etc.) is extremely fluid, in the synchrony of linguistic forms and even more so in the diachrony of change (consider Latin HABĒRE from which both the Italian verb avere and the morpheme of future, -ò developed, e.g. canter-ò < CANTARE HABEO; see also Old Eng. willan ‘want’ > Eng. future gram will; Lat. volo ‘want’ > Rum. Future gram vrea; see 1.4.3 and 1.4.4). In short, linguistic categories, like any natural category, are open classes, whose inventory of specimens and prototypical core can restructure in diachrony, entailing significant fluctuations in the membership – and the ways of membership – of individual linguistic entities in a given category. 16 A prototype may concide with the most typical specimen, as for Langacker (1987: 371), according to whom, “[a] prototype is a typical instance of a category, and other elements are assimilated to the category on the basis of their perceived resemblance to the prototype”. Alternatively (and especially in the case of linguistic categories), a prototype can be conceived as a set of properties that specimens share to varying degrees, rather than as an objective specimen (e.g., Givón 1986). In both cases, the notion of prototype is highly culture-specific (and language-specific): for instance, while it is true that there are very clear examples of the category CUP (Labov 1973), it is also true that many of the central features of this category are fixed within a specific culture (the porcelain, the handle, the saucer, etc.). Even for linguistic categories too, the frequency and centrality of categorical properties varies from language to language, so that what is, for instance, the transitive prototype in one language does not necessarily coincide with the transitive prototype in another language, not unlike how the pigeon or the seagull embodies more central bird specimens in Europe than the bald eagle or the peregrine falcon, which are classified among the official African bird species. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 48 Another important consequence at the methodological level concerns the data and the way it is collected. Indeed, frequency and prototypical effects can only be observed in the analysis of natural data. As a consequence, in functionalist and cognitivist research (especially in grammaticalisation studies, cf. 1.4.3) the practice of collecting authentic speech (or writing) has become prevalent (and is now largely enhanced by the development of sophisticated electronic corpora): it is common now to address theoretical issues through the examination of bodies of naturally occurring language use. This practice has been in place for decades in the work of those who examine the use of grammar in discourse with an eye toward determining how discourse use shapes grammar […]. In fact, the study of grammaticization has played a central role in emphasizing the point that both grammatical meaning and grammatical form come into being through repeated instances of language use. This line of research along with the discourse research mentioned above indeed seeks to explain the nature of grammar through an examination of how grammar is created over time, thus setting a higher goal for linguistic explanation than that held in more synchronically oriented theory, which requires only that an explanatory theory provide the means for adequate synchronic description. (Bybee 2006: 712) On the other hand, the practice of using natural data has created a methodological convergence with sociolinguistic analysis, for which the use of this type of data is constitutive, as this discipline builds the analysis of synchronic variation, which is inherent to language use, precisely on the natural speech (Bybee 2006: 712). In the following sections, three key notions will be discussed that stem out from this general, functionalist and cognitivist scenario, or that decline it in greater detail and that serve to frame the various studies proposed in a coherent manner. These notions stand at different levels of analysis, namely that of codification, that of the origin of new forms, and, finally, that of linguistic description: 1) subjectification, a perspective way of representing linguistic events to which the description of modal meanings is intimately connected; 2) grammaticalisation, a model of linguistic change that serves to describe the development path of modal meanings from discourse; 3) semantic topology, a methodological 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 49 construct that provides a spatial representation of the semantic-functional networks of and among linguistic entities. 1.4.2 Subjectification. 1.4.2.1 The notion of subjectivity. The notion of subjectivity largely predates usage-based approaches. A foundational formulation is due to Èmile Benveniste (1958 [1966]),17 who in fact sketched the fundamental theoretical traits of the notion, from which all subsequent formulations are based. C’est dans et par le langage que l’homme se constitue comme sujet; parce que le langage seul fonde en réalité, dans sa réalité qui est celle de l’être, le concept d’“ego”. La “subjectivité” dont nous traitons ici est la capacité du locuteur à se poser comme “sujet”. Elle se définit, non par le sentiment que chacun éprouve d’être lui-même (ce sentiment, dans la mesure où l’on peut en faire état, n’est qu’un reflet), mais comme l’unité psychique qui transcende la totalité des expériences vécues qu’elle assemble, et qui assure la permanence de la conscience. Or nous tenons que cette “subjectivité”, qu’on la pose en phénoménologie ou en psychologie, comme on voudra, n’est que l’émergence dans l’être d’une propriété fondamentale du langage. Est “ego” qui dit “ego”. Nous trouvons là le fondement de la “subjectivité”, qui se détermine par le statut linguistique de la “personne”. (Benveniste 1958 [1966]: 259–260) As emerges from this definition, there is no univocal link between subjectivity and modality (as suggested, however, by the title of the section). For Benveniste, in fact, the notion of subjectivity is much broader and serves first and foremost to define relations of person, governed by an inherent logic of dialogue and reciprocity.18 Rather, modality is one of sev- 17 But cf. already the observations in Bréal (1897: 254-265), who devoted an entire chapter of his Essay de sémantique to the élément subjectif. See also Bühler (1934). 18 “La conscience de soi n’est possible que si elle s’éprouve par contraste. Je n’emploie je qu’en m’adressant à quel-qu’un, qui sera dans mon allocution un tu. C’est cette condition de dialogue qui est constitutive de la 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 50 eral linguistic categories emanating from subjectivity and requires specific grammatical forms to express itself. Indeed, Benveniste (1958 [1966]: 264) continues: En disant je souffre, je décris mon état présent. En disant je sens (que le temps va changer), je décris une impression qui m’affecte. Mais que se passera-t-il si, au lieu de je sens (que le temps va changer), je dis: je crois (que le temps va changer)? La symétrie formelle est complète entre je sens et je crois. L’est-elle pour le sens? Puis-je considérer ce je crois comme une description de moi-même au même titre que je sens? Est-ce que je me décris croyant quand je dis je crois (que...)? Sûrement non. L’opération de pensée n’est nullement objet de l’énoncé; je crois (que...) équivaut à une assertion mitigée. En disant je crois (que...), je convertis en une énonciation subjective le fait asserté impersonnellement, à savoir le temps va changer, qui est la véritable proposition. It is clear from Benveniste’s description that the modality referred to is the one that involves the speaker’s commitment to the degree of truth of the utterance – what the later linguistic literature would call the epistemic modality. The verb ‘to believe’, employed to exemplify the concept, introduces the speaker’s point of view and is, therefore, a signal of de-objectification of the truth of the utterance: it is the “I” that says “I”, to use Benveniste’s own words. The same applies to other verbs of opinion used in a parenthetic way (in Urmson’s 1952 terms), such as ‘to suppose’, ‘to presume’, ‘to assume’ or ‘to conclude’, in utterances such as You are, I suppose, Mr Smith or I presume that John has received a letter or, again, He has left the hospital, so I deduce that he is cured. On a purely denotative level, these verbs describe logical operations on a par with, say, ‘to reflect’ or ‘to reason’. However, in using ‘I assume’, I do not describe myself in the act of assuming, but draw an inference; in using ‘I deduce’ or ‘I conclude’, I do not describe myself in the act of concluding, but draw a logical conclusion from a premise: my subjectivity acts, breaking into the utterance and assuming a certain attitude towards what is said. This manifestation of subjectivity only becomes evident in the first person, who is no longer the subject of the personne, car elle implique en réciprocité que je deviens tu dans l’allocution de celui qui à son tour se désigne par je” (Benveniste 1958 [1966]: 260; see also Benveniste 1946 [1966]). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 51 utterance (subject de l’énoncé), but becomes the subject of the whole enunciation (subject de l’énunciation). Recalling (and in fact widening) Urmson’s (1952: 484) account that “the whole point of some parenthetical verbs is to modify or to weaken the claim to truth which would be implied by a simple assertion p”, Lyons (1977: 738) explains the role of verbs such those discussed by Benveniste in sentences such as I’ll be there at two o’clock, I promise you as a parenthetical use of performative verbs: the speaker adds to the first clause, with which he performs the illocutionary act of promising, a second clause which makes explicit the nature of his speech-act; and the parenthetical ‘I promise you’ confirms, rather than establishes, the speaker’s commitment (cf. I’ll be there at two o’clockthat’s a promise). In short – Lyons (1977: 739) summarises – these verbs are in fact indicateurs de subjectivité, as indicated by Benveniste, i.e., “devices with which the speaker, in making an utterance, simultaneously comments on that utterance and expresses his attitude towards what he is saying”. On the notion of subjectivity formulated by Benveniste, Lyons bases his description of the domain of modality, which deserves to be briefly summarised since it constitutes the main antecedent (and inexhaustible point of reference) of the functionalist and cognitive approaches to which the works presented here refer. It will be useful to explain immediately that, in Lyons, the notion of subjectivity exceeds the limits traced by Benveniste (which include, as we have seen, the epistemic area and the performative one, i.e., the domain of illocution) and serves to interpret both illocution and various types of modality, including epistemic and deontic, in an inclusive description of the modal domain. From a strictly linguistic perspective, Lyons (1977) marks a notional demarcation line between performative acts (or illocution) and modalities. The former refer to the types of actions that are performed in uttering words and that entail effects on the recipients (Austin 1962). In addition to promises, he particularly investigates the nature of mands (a term first 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 52 proposed by Skinner 1957), i.e., the subclass of directives that includes, hyperonymically, commands, demands, entreaties etc. that is to say, utterances which impose, or propose, some course of action or pattern of behaviour and indicate that it should be carried out. Mands differ from other subclasses of directives, such as warnings, recommendations and exhortations in that they are governed by the particular speakerbased felicity-condition that the person issuing the mand must want the proposed course of action to be carried out: if the speaker does not really want his mand to be obeyed or complied with, he is guilty of what Austin would call an abuse […]. Not only mands, but all personal directives, including warnings, recommendations and exhortations, are governed also by the more general addressee-based condition that the speaker must believe that the addressee is able to comply with the directive. One cannot appropriately command, request, entreat, advise, or exhort someone to perform an action, or demand that he perform an action, which one knows or believes he is incapable of performing. (Lyons 1977: 746) Lyons observes that in many languages (and typically in Indo-European languages) the difference between mands and statements is grammaticalised in the main verb of the sentences used to perform such acts, which is inflected in different grammatical moods. The Indo-European mood for mands utterances is the imperative, which, in the second person singular, carries no overt indication of person or tense (just as the singular vocative of nouns in Indo-European languages carries no overt indication of case), see for instance Lat. second-person singular imperative DIC ‘tell!’ vs present indicative DICIS ‘you tell’ (dic mihi quid fecerit ‘tell me what you did’ vs dicis mihi quid fecerit ‘you are telling me what you did’). Lyons notes – on the basis of an actually widespread opinion – that the lack of specific markers for the imperative might be due to the fact that, as a mood of desire and will, it is ongenetically more basic than the mood used to express statements, i.e., the indicative. In any case, mands are inherently future-oriented and vocative in nature: it is important to realize that commands and requests, of their very nature, are necessarily restricted with respect to the semantic distinctions that are grammaticalized, in many languages, in the cate- 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 53 gories of tense and person. We cannot rationally command or request someone to carry out some course of action in the past: the only tense distinctions that we might expect to find grammaticalized in the imperative, therefore, are distinctions of more immediate and more remote futurity. For similar reasons, the imperative is intimately connected with the second person (or vocative). (Lyons 1977: 746–747) Of course, in addition to the imperative, languages have a variety of other, more or less direct and more or less grammaticalised means of expressing commands, e.g., pseudostatements or pseudo-questions, used to modulate the illocutionary force to varying degrees (indirect speech acts, in Searle’s 1975 terms). Some of these means will be discussed for Sicilian in the articles presented in the next sections, where it will also be shown how indirect directivity can easily tap into the domain of modality. This is because, as Lyons (1977: 785–786) observes, the performance of indirect illocutionary acts calls into question conditions having to do with the knowledge, beliefs, will and abilities of the participants, which are the factors involved in the domain of the epistemic and deontic modality. The classification of modal subdomains proposed by Lyons builds on the classical distinction in the field of modal logic between (synthetic) propositions that are contingently true or false (or true or false in a given possible world) and (analytical) propositions that are necessarily true or false (or true or false in any logically possible world and, therefore, not empirically disproven). The central notions of this logic are, as is well known, necessity and possibility, related in terms of negation: “If p is necessarily true, then its negation, ~p, cannot possibly be true; and if p is possibly true, then its negation is not necessarily true” (Lyons 1977: 787). This type of logic allows us to interpret the meaning of must in: (a) Alfred must be unmarried. as a logical inference of Alfred is a bachelor since it is known that If Alfred is a bachelor, he must be unmarried. In other words, we normally infer logical necessity by activating our knowledges or beliefs, in this case about being a bachelor or married. This is alethic modality (< Gr. ἀληθής, alēthḗs ‘true’), which includes alethically necessary propositions, 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 54 that is, propositions that are necessarily true (or true in all possible worlds), and alethically possible propositions, that is, propositions that are not necessarily false (or true in at least one logically possible world). However, Lyons (1977: 791) continues, there are at least other two possible readings of (a): (b) ‘I (confidently) infer that Alfred is unmarried.’ (c) ‘Alfred is obliged to be unmarried.’ The propositions in (b) and (c) respectively express epistemic modality and deontic modality. According to Lyons, both imply subjectivity (contra other scholars, such as Kuryłowicz 1964, who considers subjectivity as the very feature which distinguishes epistemic from deontic modality). Epistemic modality (< Gr. ἐπιστήμη, epistḗmē ‘knowledge’) is intuitively close to alethic modality; however, (b) does not refer to what ‘is (objectively) known’ (as for the modal logic described above) but to what the speaker believes or thinks about the content of the proposition and which commits her to neither the truth nor the falsity of such a content (cf. I think that Edinburgh is the capital of Scotland) or commits her to its falsity (cf. I wish he had been to Paris; Lyons 1977: 795). However, according to Lyons, speaker’s commitment can both depend on a subjective evaluation (which is the more natural reading) and on more objective conditions. While the most likely (subjective) interpretation of Alfred may be unmarried is ‘I think Alfred is unmarried’ or ‘Perhaps Alfred is unmarried’, another scenario is still possible, which can be described as objective epistemic modality: But now consider the following situation. There is a community of ninety people; one of them is Alfred; and we know that thirty of these people are unmarried, without however knowing which of them are unmarried and which are not. In this situation, we can say that the possibility of Alfred’s being unmarried is presentable, should the speaker wish so to present it, as an objective fact. The speaker might reasonably say that he knows, and does not merely think or believe, that there is a 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 55 possibility (and in this case a quantifiable possibility) of Alfred’s being unmarried. (Lyons 1977: 798) The same holds true for must in Alfred must be unmarried: Suppose that, having established the marital status of every member of the community except Alfred, we find that we have identified exactly twenty-nine persons that are unmarried. In these circumstances, […] we should be justified in claiming to know that the proposition “Alfred is unmarried” must be true in some sense […] that ‘must’ bears in statements of alethic necessity. (Lyons 1977: 798) To summarise, while subjective epistemic utterances express opinions, hearsays or tentative inferences, rather than statements of fact (in fact, they are non-factive, like questions), objectively modalised utterances (be them alethic or epistemic) are statement, which can be accepted, denied or questioned (“I agree”, “It is not true”, “Is it so?”). They contain an it-is-so implication, which is missing in the subjectively modalised utterances, and which is “qualified with respect to a certain degree of probability, which, if quantifiable, ranges between I and 0. If the factuality of an epistemically modalized proposition (as it is presented by the speaker) is of degree I it is epistemically necessary; if its factuality is of degree 0 it is epistemically impossible” (Lyons 1977: 800). Languages may encode different degrees of factuality by means of grammatical or lexical items, for instance modal adverbs such as “certainly”, “probably”, “possibly”. Deontic modality (< Gr. δέον, déon ‘what is binding’) refers to obligation or permission, that is, it concerns the necessity or possibility of acts performed by morally responsible agents. Hence, it intersects (and to some extent overlaps, in Lyons’ description) the domain of illocution, as is evident from the following commentary, in which the terms “deontic” and “directive” freely alternate (emphasis is mine): When we impose upon someone the obligation to perform or to refrain from performing a particular act, we are clearly not describing either his present or future performance of that act. There is a 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 56 sense in which the sentence we utter can be said to express a proposition; but it is not a proposition which describes the act itself. What it describes is the state-of-affairs that will obtain if the act in question is performed; and we have already seen that directives can be analysed, along these lines, as utterances which impose upon someone the obligation to make a proposition true (or to refrain from making it true) by bringing about (or refraining from bringing about) in some future world the state-of-affairs that is described by the proposition. The notion of truth is not therefore irrelevant to the analysis of directives (the only class of deontically modalized utterances that we have so far been concerned with), but it applies less directly than it does in the analysis of subjectively or objectively modalized statements. (Lyons 1977: 823) Other two properties characterise deontic modality in contrast with epistemic utterances. The first property is the inherent future-orientation, in that the truth of a deontic proposition is determined in relation to a state of the world subsequent to (or at least contemporaneous with) the one in which the deontic event occurs (e.g., You muststate1 get outstate2 of here [now or soon]). The reference to a future world-state – as Lyons (1977: 825–826) explains – connects deontic modality to predictions, hence to intention, desire and will. This depends, according to Lyons, on the very origin of deontic meanings, which stem out from the instrumental function of language (compare this observation to Tomasello’s comments on phylogenesis of language, reported in 1.4.1): The origin of deontic modality, it has often been suggested, is to be sought in the desiderative and instrumental function of language: that is to say, in the use of language, on the one hand, to express or indicate wants and desires and, on the other, to get things done by imposing one’s will on other agents. It seems clear that these two functions are ontogenetically basic, in the sense that they are associated with language from the very earliest stage of its development in the child. It is equally clear that they are very closely connected. (Lyons 1977: 826) The second property is that deontic modality derives from some force or cause and different types of source or cause outline different more or less constraining forms of deontic modality (e.g., a human authority, a natural force, etc.): 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 57 If X recognizes that he is obliged to perform some act, then there is usually someone or something that he will acknowledge as responsible for his being under the obligation to act in this way. It may be some person or institution to whose authority he submits; it may be some more or less explicitly formulated body of moral or legal principles; it may be no more than some inner compulsion, that he would be hard put to identify and make precise. (Lyons 1977: 824) Although contiguous, deontic modality and illocution do not coincide. While the latter is normally performed as a directive act, deontic modality is not a direct imposition or permission (“I hereby impose the permission or grant the permission of performing a certain action”), but a statement, something that can be rephrased as “I hereby assert that you are obliged or allowed (by some unspecified authority) to perform a certain action”. As we will see (cf. 1.4.4), there are good reasons to stress the difference between the two related notions of deonticity and directivity, which is well outlined by Lyons even if not always operational in all the subsequent literature. This difference will be proved to be of crucial importance to distinguish certain phenomena of linguistic change and, at the same time, synchronic interaction between different semantic nuances of linguistic entities. 1.4.2.2 From subjectivity to subjectification. The notion of subjectivity is also assumed in Langacker’s cognitive grammar, in which it becomes a core principle. This centrality cannot be surprising, since cognitive approaches to semantics (not only cognitive grammar; cf. Lakoff 1987 inter al.) are, unlike and in open contrast to formal logic, explicitly anti-objectivist. As Langacker (1990: 5–6) explains: A foundational claim of cognitive semantics is that an expression’s meaning cannot be reduced to an objective characterization of the situation described: equally important for linguistic semantics is how the conceptualizer chooses to construe the situation and portray it for expressive purposes. An expression’s precise semantic value is determined by numerous facets of construal, including the level of specificity at which the situation is characterized, background assumptions and expec- 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 58 tations, the relative prominence accorded various entities, and the perspective taken on the scene. It is one aspect of perspective that concerns us here. The meaning of linguistic forms is, therefore, as is evident from the quotation, a matter of construal and perspectivisation, that is to say eminently subjective notions concerning the way in which speakers conceptualise and, then, linguistically represent events.19 As it is conceived as a space grammar (Langacker 1982), cognitive grammar schematically represents the notion of subjectivity as in Figure 2, which is derived from a couple of seminal articles by Langacker (1985, 1990, the latter significantly appeared in the first issue of the influential journal Cognitive linguistics). The Figure shows a viewer (or conceptualiser, V), who is offstage (i.e., outside the cognitive scene) in (a), and onstage (OS, i.e., inside the cognitive scene) in (b). In (b), V is arranged in the cognitive scene (or the perceptual field, PF) together with P (i.e., the perceived object). The relationship between P and V is indicated by the dotted lines. What is depicted in (b) is what Langacker calls an egocentric viewing arrangement. Transferred to a more directly linguistic level, this egocentric (in other words, subjective) view becomes a matter of grounding. Ground is the term commonly used in cognitive grammar to refer to the dimension of the enunciation, that is, “the speech event, its participants, and its immediate circumstances (such as the time and place of speaking)” (Langacker 1990: 9). As Figure 3 shows, the ground and its components (G) can be external to the linguistic representation of the events as in (a) (e.g. the child broke the glass, which is a rather objective way of presenting a certain event) or included in the scope of predicated event to a variable extent. In particular, G can be included in the maximal scope of predication (MS), while still remaining offstage, as an implicit and unprofiled reference point, as in (b) (e.g., yesterday in the child broke the glass yesterday) or 19 Of course, this is not a matter of individual subjectivity: rather, it is what speakers do within a specific linguistic culture, so to speak (i.e., in Saussurian terms, a fact that invests the langue). A transparent example is provided by the active (the child broke the window), passive (the window was broken [by the child]) or middle-intransitive (the window broke [because the child threw a stone]) constructions, which are different construals of the same cognitive scene, in which an agentive entity (which, depending on the case, is or is not in focus) exerts a force that provokes a change of state in another non-agentive entity. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 59 in the immediate scope (onstage region, IS), as in (c) (e.g., this and now in the child is breaking this glass right now). Figure 2. Objective and subjective construals (Langacker 1990: 7). Figure 3. Grounding (Langacker 1990: 10). What, in perfect continuity with Benveniste, first describes the deixis in general once again invests the dimension of modality in particular. Here Langacker’s discourse shifts from the plane of synchronic description to that of diachronic processes, in particular to the phenomena of grammaticalisation, “whereby ‘grammatical’ elements evolve from ‘lexical’ 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 60 sources” (Langacker 1990: 16).20 According to Langacker, some instances of grammaticalisation involve subjectification, a type of semantic shift that implies an increase of subjectivity or, in Langacker’s terms, “the realignment of some relationship from the objective axis to the subjective axis” (Langacker 1990: 17; see also the formalization reported in Section 2.2.1). Figure 4. Grounding predications (Langacker 1990: 26). In various works from the 1980s onwards and then more systematically in his Fundations of Cognitive Grammar (1991), Langacker uses the notion of subjectification to explain diverse phenomena of linguistic change including modality and, in particular, English 20 The inverted commas within the quotation obviously refer to the fact that, as already mentioned, in cognitive as well as in functionalist approaches, there is no sharp division between lexicon and grammar, which in fact form a continuum. This is a central tenet in grammaticalisation theories; see 1.4.1, 1.4.3. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 61 modals, which are said to function as grounding predications (rather than grammatical constituents in the strict sense). The development of modals (and auxiliaries more in general) can be represented as in Figure 4. This shows three different configurations. The first configuration (a) represents the conceptual import of the forerunners of will, may and can in Ancient English, respectively a verb meaning ‘want’, one attributing to its subject the physical ability to do something or, finally, the requisite of knowledge or some mental ability. Each of them – Langacker (1990: 25) claims – “invokes the conception of an associated process, which represents what it is that the subject wants or is somehow able to do”. This associated process is the one represented within the rectangole pushed forward by the subject (trajector, tr), in which the same subject (tr) performs a certain state of affairs (landmark, lm). This is still an objectively construed situation (in which the lm is encoded in English by means of an infinitival complement). On the other hand, (b) and (c) show the increasing participation of G in the process performed by tr. It is worth noting that these new grounded situations are placed in a timeline, where the present is the locus of G, whereas the process is something that will be realised in a subsequent timeframe. There are two main observations that follow from this description. The first concerns the temporal arrangement of the situation described or, rather, the relationship between tense and modality or, which is the same, between tense and reality, which is best addressed and formalised in the so called epistemic model by Langacker (1991: 240–249; discussed and operationalised here in 2.5). In a modalised situation, the present corresponds to the immediate reality, everything else is non-reality, albeit in different ways. The past exists only, possibly, in memory, as a reality known but no longer in act (known reality) or, otherwise, as a unknown reality (and it should be noted that the past of modals does not evoke past events but an increased degree of non-reality, cf. for instance I can sleep vs I could sleep [but I will not sleep] or [but I didn’t]). On the other hand, the future is not yet, it is precisely non-reality. Langacker’s formalisation is undoubtedly an elegant and effective representation, although the intuition is far from original. The thought inevitably runs to Saint Augustine: 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 62 I ask, Father, I affirm not: rule me, O my God, and direct me. Who is he that will tell me how there are not three times, as we learned when we were boys, and as we taught other boys, the past, present, and future; but the present only, because the other two are not at all? Or have they a being also; but such as proceeds out of some unknown secret, when out of the future, the present is made; and returns it into some secret again, when the past is made out of the present. For where have they, who have foretold things to come before, seen them, if as yet they be not. For that which is not, cannot be seen. And so for those that relate the things past, verily they could not relate true stories, if in their mind they did not discern them: which if they were none, could no way be discerned. There are therefore both things past and to come. (Confessions XI: 17; cf. Augustine 1912). For Saint Augustine, the past and the future do exist in the mind: there are a praesens de praeteritis ‘present time of past things’ in our memory, a praesens de praesentibus ‘present time of present things’ in our sight and a praesens de futuris ‘present time of future things’ in our expectation (Confessions XI: 20). As Lyons (1977: 815) already observed about Saint Augustine’s thought, “many philosophers would deny that we can make statements about the future at all, on the grounds that we cannot have knowledge, but only beliefs, about future world-states. What purports to be a statement describing a future event or state-of-affairs is therefore, of necessity, a subjectively modalized utterance: a prediction rather than a statement. Looked at from this point of view, the difference between It will be raining (tomorrow) and It may be raining (tomorrow) depends upon the speaker’s subjective evaluation of the probability of ‘it be raining’ being true of Wj [i.e., a certain possible world, EM].” The second observation is that, although the model is called epistemic, it actually also allows us to explain deontic values (for which Langacker uses the label ‘root’). Based on Sweetser’s (1982, 1990) classic account, the difference between the two types of modality depends on the locus of the potency directed towards the realisation of the state of affairs: in the case of the deontic or root modality, the power is a force located in the social sphere (in which an obligation is imposed or a permission is granted); in the case of epistemic meanings, the force moves into the domain of reasoning. According to Langacker, rather 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 63 than an abrupt metaphorical shift, as in Sweetser, we are dealing with various degrees of identifiability of the locus of potency: Without disagreeing [with Sweetser, EM], one can also see it as hinging on whether or not the locus of this potency can be identified with a particular individual, or at least with some delimited facet of present reality. For example, we speak of should having a root sense when the envisaged occurrence of the landmark process is construed as an obligation imposed by the speaker. A root sense may also involve thes reporting an obligation imposed by some other source of authority: it can be another person, an abstract entity such as a law, or even something as diffuse as societal norms or cultural expectations […] It thus seems natural to analyze an epistemic modal as representing a limiting case, that in which diffuseness of the locus of potency is pushed to its ultimate conclusion. In the senses we regard as epistemic, the locus of potency is the polar opposite of a focused, well-delimited source of authority – rather, it is identified holistically as the nature of evolving reality itself. Because the locus is undifferentiated and all-encompassing, factors that correlate inversely with objectivity, extension from a root to an epistemic meaning constitutes a type of subjectification. We could summarise Langacker’s reasoning in terms of a relation of continuity between deontic/root and epistemic meaning, which is both a relation of increasing abstractness and a pathway largely confirmed by the development of modals in historical languages, including Sicilian, as we will see (but see Bybee and Pagliuca 1985; Traugott 1989 for English modals).21 This path is marked by an increased subjectification, which is manifested in the loss of the temporal relation of succession between the modal and the event. As Langacker (1990: 29) explains, in the epistemic situation, we are not dealing with the probability that a certain situation will occur in the future, but rather to the possibility that it has already occurred. This usage reflects a shift from the concept of the evolution of reality itself to the evolution of our knowledge of reality (present). Since reality is largely external, whereas 21 It should be noted that not all scholars would accept a gradual interpretation of the relationship between epistemic and deontic modality. For instance, van der Auwera (p.c. 2012), who is generally in favour of gradience, objected that modality refers to the degree of certainty/commitment of the speaker or not. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 64 knowledge of reality is entirely the responsibility of those who conceptualise it, the evolutionary momentum is interpreted more subjectively when applied to the latter. The interest in diachronic side of subjectification is even more decisive in Traugott’s work. Moreover, compared to Langacker, Traugott’s approach is more firmly rooted in the pragmatic dimension and use, rather than in conceptualisation. In an influential work of 1989, dedicated to the rise of epistemic meanings in English, Traugott formulates some principles of linguistic change destined to have wide circulation and development in the subsequent years. As for Langacker, modalisation coincides with an increase in subjectivity, which includes the speaker in the semantics of the modal form. For Traugott, modal meanings (including grammatical and lexical meanings, i.e., modal verbs must, shall and will, speech act verbs, such as insist and suggest, and modal adverbs, such as probably, apparently, evidently) develop along a lexical to epistemic path, which is unidirectional in nature (as also for Bybee and Pagliuca 1985: 66). This path is as follows: lexical meanings > deontic (speech act like, i.e., related to will, obligation and permission, as in Palmer 1986 [2001]) > weak epistemic (corresponding to Lyons’ objective epistemic modality, a notion questioned by Traugott)22 > strong epistemic. A path – Traugott (1989: 43) warns – that allows previous meanings to coexist with new ones. Like Langacker, Traugott also expresses doubts as to whether this path can be interpreted entirely in metaphorical terms, as proposed by Sweetser (1982). The metaphor may at most have triggered the process, which is then, instead, scalar in nature: I suggest that a differentiation should be made between the development of the nonepistemic meanings and the development of epistemic meanings […]. A shift from owing concrete debts such as money to owing certain behaviors (cf. *sculan) can indeed be seen as metaphorical transfer based on similarities. A shift from a concrete meaning such as ‘set/stand on’ (cf. insist) to a speech-act verb can also be seen as a metaphorical transfer from the standing or setting of something on something to being firm about a proposition. But the shift to epistemics of conclusion, 22 What is obvious from the evidence (i.e., objectively epistemic according to Lyons) is already, for Traugott (1989: 36) a subjective evaluation of a state of affairs. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 65 belief, knowledge, hearsay, hypothetical conditionality, and so forth has little of the analogical mapping from one conceptual domain onto another that is characteristic of metaphor. Rather, the path can be seen as the conventionalizing of conversational implicatures. (Traugott 1989: 50, emphasis mine) Conventionalisation (that is, the inclusion in the semantics) of conversational implicatures produces pragmatic strengthening: “the development of epistemic and evidential meanings increases encoding of speaker informativeness about his or her attitude. There may be weakening of the semantics of deontics, but there is strengthening of focus on knowledge, belief, and the speaker’s attitude toward the proposition” (Traugott 1989: 49). In short, subjectification – and this is one of the most striking differences with Langacker’s approach to modals – is for Traugott an eminently pragmatic pathway, which does not contradict cognitivist interpretations, but shifts the analysis to a different plane, one that is entirely related to the use of language, rather than the conceptualisation of events: The main difference is the perspective: the metaphoric process of mapping from one semantic domain onto another is used in the speaker’s attempt to increase the information content of an abstract notion; the process of coding pragmatic implicatures is used in the speaker’s attempt to regulate communication with others. In other words, metaphoric process largely concerns representation of cognitive categories. Pragmatic strengthening and relevance as I use the terms largely concern strategic negotiation of speaker-hearer interaction and, in that connection, articulation of speaker attitude. (Traugott 1989: 51) Traugott (1995: 45–46) proposes that subjectification lies at the very core of any grammaticalisation process (although it also acts on purely lexical forms, so it is a broader notion than Langacker’s), where it operates along multiple dimensions, which relate to both pragmatics and morphosyntax, such as propositional function → discourse function, objective meaning → subjective meaning, non-epistemic modality → epistemic modality, nonsyntactic subject → syntactic subject → syntactic subject > speaking subject, free form → bonded form. In a word, a path from pragmatics to morphosyntax. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 66 The reason for the ubiquity of subjectiflcation presumably lies in the speaker’s attempts to communicate the relevance of what is said to the communicative event, which includes hearers as well as speakers, but which ultimately depends for its occurrence on the speaker. (Traugott 1995: 46) A further development of subjectification is intersubjectification, which involves “the development of meanings that encode speaker/writers’ attention to the cognitive stances and social identities of addressees, arises out of and depends crucially on subjectification” (Traugott 2003a: 124). At the origin of the notion we once again find Benveniste (1958 [1966]: 259–260; see fn. 19), who, as we have seen above, describes the ‘I/you’ dyad as constitutive of communication, since self-consciousness is only possible by contrast: I only use ‘I’ when addressing someone, who will be a ‘you’ in my speech. And again, in his essay on the nature of pronouns, Benveniste (1956 [1966]: 254) explains that the function of first- and second-person deictics solve the problem of intersubjective communication: “Leur role est de fournir l’instrument d’une conversion, qu’on peut appeller la conversion du langage en discours. C’est en s’identifiant comme personne unique prononçant je que chacun de locuteurs se pose tour à tour comme ‘sujet’”. However, while referring explicitly to Benveniste, Traugott actually proposes an entirely new concept of intersubjectivity: it is no longer, we might say, a non-referential subjectivity that inherently evokes the other (you), but a subjectivity that focuses on another subjectivity. This is what emerges from the definition she provides: intersubjectivity is the explicit expression of the SP/W’s [speaker or writer, EM] attention to the ‘self’ of addressee/reader in both an epistemic sense (paying attention to their presumed attitudes to the content of what is said), and in a more social sense (paying attention to their ‘face’ or ‘image needs’ associated with social stance and identity). (Traugott 2003a: 128) Just as subjectivity drives subjectification, intersubjectivity underlies intersubjectification, a process through which “meanings become centred on the addressee” (Traugott 2003a: 129). This is in fact onceived of by Traugott as an extension of subjectivity: 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 67 There cannot be intersubjectification without some degree of subjectification because it is SP/W who designs the utterance and who recruits the meaning for social deictic purposes. Like subjectification, it is part of a mechanism of recruiting meanings to express and regulate beliefs, attitudes, etc. Therefore intersubjectification can be considered to be an extension of subjectification rather than as a separate mechanism. (Traugott 2003a:134) The directionality of change is, according to Traugott, subjectification → intersubjectification and not vice versa (just as subjective meanings succeed, as we have seen, nonsubjective meanings and not vice versa). This hypothesis is semasiological, because it concerns first of all the constraints on the change that individual lexemes may undergo, but it also invests the onomasiological level, hence the shifts of meaning from one conconceptual domain to another (e.g. from the spatial domain to the markers of politeness in Japanese, not vice versa). Although Traugott uses it to explain phenomena such as honourifics, the notion it can easily pass into other domains, such as modality (cf. inter al. the recent Kookan and Nuyts 2021). Intersubjectification could also be helpful to interpret some of the constructions examined in the next sections, which, as we will see in due course, are used to by mitigate the illocutionary force of an order addressed to a ‘you’. In the discussion of subjectification and modalisation so far, repeated reference has been made to the notion of grammaticalisation, the mechanics of which have been partly explicated through the various cases presented. In what follows, more systematic attention will be paid to this fundamental model of change, which underlies all the cases discussed in the articles presented below. 1.4.3 Grammaticalisation. A key notion in the historical linguistics of recent decades, grammaticalisation constitutes the informing principle that binds all the contributions presented in this volume together and, therefore, deserves specific attention. However, many aspects of this articulated mechanism of linguistic change have already emerged, more or less explicitly, in the section 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 68 dedicated to subjectification – a notion, as we have seen, closely intertwined with the former. On the other hand, many other aspects will be dealt with in detail in the following sections. Therefore, the discussion here will highlight only a few basic issues, concerning the scope of the process, and then focus on a few semantic areas particularly involved in grammaticalisation processes, at the crosslinguistic level and in the work presented here (cf. 1.4.4). In general terms, grammaticalisation is a phenomenon of linguistic change whereby a given form shifts from the lexicon to the grammar (cf. Lat. PASSUM ‘step’ > Fr. negation pas; Lat. MENTE, which is the ablative of MENS ‘mind’ > Fr. and It. adverbial suffixes ment, -mente, e.g. cordialement, cordialmente ‘cordially, warmly’, but literally ‘with cordial mind’) or enriches its semantic network with new grammatical meanings alongside the basic lexical ones (e.g., It. avere is both a lexical verb meaning possession, as in ho un cane ‘I have a dog’, and a perfective auxiliary, as in ho mangiato una mela ‘I have eaten an apple’). This passage is anything but abrupt, but it admits several intermediate stages: For example, a lexical noun like back that expresses a body part come to stand for a spatial relationship in in/at the back of, and is susceptible to becoming an adverb, and perhaps eventually a preposition and even a case affix. (Hopper and Traugott 1993 [2003]: 6) What Hopper and Thompson describe in the above quotation is commonly referred to by the metaphorical label “cline”, that is, a unidirectional sequence of points of development, such as: content item → grammatical word → clitic → inflectional affix (Idd. 1993 [2003]: 6). This cline illustrates, for instance, the verb-to-affix grammaticalisation chain, a common change in Romance languages, such as those leading from a form of HABEO to Romance future or conditional forms (cf. 1.3.1, 1.3.2), for instance, Cl. Lat. [[CANTARE]V HABEOCOMPL] 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 69 > Late Lat. [CANTARE HABEO] > Fr. [chant-e-r-ai] where a verb+complement structure tightens the links between constituents and becomes a periphrasis that involves a (quasi)auxiliary (and which is subject to a number of syntactic restrictions, e.g., fixed order and non-separability of constituents), and then progressively loses not only syntactic autonomy but also phonetic weight (it becomes an affix). Grammaticalisation does not necessarily follow the path in its entirety and may be interrupted at more or less advanced stages. However, this all in all simple path (reported by Hopper and Traugott 1993 [2003]: 42–44) allows us to highlight two crucial facts implicated in grammaticalisation. Firstly, the stages involved, which are generally viewed in terms of a progressive loss of autonomy at different levels (Heine 1993: 53–54; Heine 2003: 759; Hopper and Traugott 1993 [2003]: 94–112): (1) Semantic. (2) Pragmatic. (3) Categorial (that is, morphosyntactic). (4) Accentual and phonetic. As for (1), the traditional account in terms of desemanticisation (in the strict sense of semantic bleaching, cf. Sweetser 1988) has been questioned by Traugott (1989, 1995), who, as we have already seen in 1.4.2.2, proposed that grammaticalisation rather produces a semantic increase, namely the acquisition of new pragmatic or abstract (grammatical) meanings. This is all the more true because – and this is the second fact to be highlighted – the forms on the left of the chain do not necessarily disappear as the change proceeds. In other words, different stages can co-exist with each other over time, so it is not surprising for a full form to synchronically overlap with a grammaticalised or grammaticalisig form (e.g., as mentioned, It. avere is both a verb of possession and an auxiliary; It. andare means ‘to go’ and, at the same time, it functions as a passive auxiliary). Losing semantic autonomy rather means – as we will see shortly – that certain aspects of the meaning gene- 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 70 ralise to new contexts (second stage, that is, the pragmatic one), and this can produce ambiguity of interpretation. The third stage, i.e., decategorisation, entails the loss of certain properties as a full verb (e.g., the possibility of being conjugated in all forms, of forming imperatives or passives, or of being negated separately), while (4) means loss of accentual autonomy (i.e., the form becomes a clitic) and, at the limit of the process, being eroded phonetically (e.g., It canter-ò < CANTARE+HABEO; Am. Eng. gonna < going to). The reconstruction of such a path, which has a solid diachronic basis in historical languages, entails some consequences of enormous relevance on a theoretical level. Not only is grammaticalisation a fluid process at the diachronic level, but this fluidity reverberates in synchrony, affecting the structure of linguistic categories, categorial boundaries (between lexicon and grammar) and the semantic networks of individual linguistic items, which may include in fact lexical and (more) grammatical meanings, the boundaries of which are not infrequently blurred (cf. 1.4.1). This is clearly summed up by Heine (1993: 53), who explains that “the presence of overlaps is responsible for the fact that grammatical categories are inherently ambiguous in certain uses”. It is precisely the assumption of a non-discrete heuristics that explains the alliance between functionalist and cognitivist approaches. However, these approaches differ (but do not contradict each other) in a number of respects, primarily in the identification of the motivations of the process, which are said to be more pragmatically based or more cognitively based. We have already discussed Traugott’s pragmatic reasons in the previous section and mention was also made of the metaphorical hypotheses proposed in the cognitivist field by Sweetser. An approach that undoubtedly combines these two poles of reasoning (which is then the one taken in the works proposed here, cf. in particular 4.2) is that put forward by Heine. There are two fundamental aspects that deserve attention here. One concerns in fact the origin of grammatical meanings. According to Heine (1993; Heine et al. 1991) at the basis of grammaticalisation processes we find basic words, denoting universals of experience that are culturally independent as they express basic human relations with the environment (generic movement or generic identity). These source concepts are schematised according to very simple patterns of the propositional type (event schemas), which – it is important to emphasise – are me- 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 71 taphorical abstractions from certain aspects of physical experience and, therefore, are at a cognitive, hence pregrammatical level (see also Langacker 1987: 168–169). Among such schemas, we can mention: Location (X is to Y); Motion (X goes from/to Y); Activity (X does Y); Desire (X desires Y); Change of state (X becomes Y); etc. (Heine 1993: 27–30). The actual linguistic encoding is of course language specific, although a certain crosslinguistical recurrance can be observed in conveying such notions by means of verbs such as ‘be’, ‘stay’, ‘go’, ‘do’, ‘continue’, ‘begin’, ‘end’, ‘want’, ‘become’, etc. These verbs express very general meanings, containing only a schematic description of the events and lacking all the details that are instead expressed by other similar verbs (for instance, ‘to go’ and ‘to come’ only describe the pattern “source/path/goal”, without the further specifications contained in verbs such as ‘to swim’, ‘to walk’, etc., which also provide information about the manner of motion). This generality (which also corresponds to greater semantic lightness) plays a crucial role in increasing the number and types of contexts in which linguistic entities can be employed and makes them particularly suitable for grammaticalisation: frequency is, in fact, a fundamental component of any grammaticalisation process (as well as being a consequence of it; cf. Bybee et al. 1994; Bybee 2003, 2005). Since they are cognitively grounded, it is not surprising that certain processes of grammaticalisation follow similar paths crosslinguistically, e.g., motion towards a goal → intention → future; will or desire → intention → future (cf. Bybee 2003; Heine and Kuteva 2002; Kuteva et al. 2019). Up to this point, Heine’s approach would seem not to deviate too much from earlier cognitivist proposals that identified metaphorical abstractions as the origin of grammaticalisation processes. However, in Heine’s model, metaphor functions at best as the trigger of a process that not only develops gradually, but is also entirely operative in language, in discourse: The process from cognition to grammar has both a discrete and a continuous perspective. The former is mainly psychological in nature and suggests an analysis in terms of metaphor, while the latter appears to be essentially pragmatic: it is highly context-dependent and exhibits a metonimical structure. (Heine et al. 1991: 70) 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 72 This explanation overcomes the contradiction (already pointed out by Langacker and Traugott, as we have seen in 1.4.2.2) between the discreteness of the metaphorical transfer from one conceptual domain to another and, on the other hand, the gradualness of the transition from minus to plus to grammatical that is characteristic of the grammaticalisation process. As an explanation of the gradualness of the process, Heine et al. (1991: 70) give the classic example of the verb to go in sentences such as: (1) He is going to town; (2) He is going to eat; (3) He is going to do his very best to make you happy; (4) The rain is going to come. In these examples there is no discontinuity between the various senses of going, but progressive shifts of focus (metonymies or pragmatic inferences; see also Hopper and Traugott 1993 [2003]: ch. 4; Traugott and König 1991) on different parts of the meaning: the physical-spatial movement expressed in (1) is marginal in (2), which conveys a sense of ‘intention’ and, secondarily, ‘prediction’; (3) expresses the same meanings as (2), but loses the spatial one altogether; in (4) the only possible meaning is ‘prediction’ (cf. also 5.1.1). Metonymical shifts originate within the contextual dimension, in a process called by Heine et al. (1991: 71–72; Heine 2003: 579) “context-induced reinterpretation”, which involves three ideal stages: (i) A certain linguistic expression F with a meaning A is recruited to be grammaticalised. (ii) F acquires a second usage pattern, B, with the effect of creating an ambiguity between A and B. (iii) Finally, either A is lost or, alternatively, there are contexts in which the only possible meaning is B; in the latter case, the effect is the polysemy of F. The intimate connection of this model with the functionalist approach à la Hopper and Traugott is evident. 1.4.3.1 Grammaticalisation and constructions. The second part of Heine’s discussion, concerning the contextual dimension of change, leads us directly to an issue that is still rather debated today, which concerns the scope of the process. In none of the modern approaches, and especially none of those just mentioned, is grammaticalisation conceived as a change affecting a single lexical entity. As Elizabett Closs Traugott (2003b: 624–625; cf. also Hopper and Traugott 1993 [2003]: 18–22) 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 73 has observed, this characterisation is actually constitutive of the notion, first formalised in a famous work by Antoine Meillet (1912 [1921]),23 L’évolution des formes grammalicales, appeared in the context of Indo-European studies. In this work, the notion is certainly employed in reference to a process involving individual words, namaly “le passage d’un mot autonome au rôle d’élément grammatical” (Meillet 1912 [1921]: 131). To exemplify this passage, Meillet cites the first person form of ‘to be’ in French, i.e., suis, in several contexts: (i) je suis celui qui ‘I am the one who’, in which the form is an autonomous one (as it means ‘existence’); (ii) je suis chez moi ‘I am at home’, where it still maintains a certain autonomy (‘location’); (iii) je suis malade ‘I am sick’, je suis maudit ‘I am cursed’, in which the lexical value now fades into a grammatical one (a copula); (iv) je suis parti ‘I left’, je suis allé ‘I have gone’, je suis promené ‘I went for a walk’, in which suis is a grammatical element in its own right (an auxiliary). According to Meillet (ibid.), “[i]l est pourtant clair […] que suis est dans: je me suis promené le même mot que dans: je suis ici; mai il est devenu une partie constituante d’une form grammaticale”. It is already evident from this quotation that, for Meillet, the transition from one sense to another occurred in or, at any rate, affects a specific syntactic context. The discussion of word order, which was “expressive” in Latin (i.e., pragmatic and fundamentally free), while it developed a grammatical value in French (in which cases Latin disappeared as in most Romance languages), is even more explicit in this sense: La valeur expressive de l’ordre des mots, que l’on observé en latin, a été remplacé par une valeur grammaticale. Le phenomene est de même ordre que la “grammaticalisation” de tel ou tel mot; au 23 Of course, the idea of a path from lexical to grammatical precedes the creation of the word to define it. According to Heine (2003: 576), forerunners of the notion of grammaticalisation can be considered Étienne Bonnot de Condillac (1746) and, from a linguistic perspective, Franz Bopp (1816). On the other hand, the contributions of Meillet and other Indo-Europeists, such as Kuriłowicz (1965), in the first part of the 20th century would not, Heine argues, have added any major theoretical insights to what had already been proposed in the previous century. A different opinion is expressed by Traugott (2003b; Hopper and Traugott 1993 [2003]: 18), who recognises Meillet’s central role and sees Wilhelm von Humboldt (1925) as the main pre- cursor. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 74 lieu que ce sois un mot employé en groupe avec d’autres qui prenne la caractère de “morphéme” par un effect de l’habitude, c’est une maniére de grouper le mots. (Meillet 1912 [1921]: 148) In commenting on such an early analysis, Traugott (2003b: 625) observed that, after Meillet and until the turn of the Millennium, most work on grammaticalisation ignored (or explicitly rejected) broader syntactic phenomena such as word order and the notion of grammaticalisation was mostly understood in fact as a change of individual forms. See for instance Kuriłowicz’s (1965: 69) definition, whch had in fact broad circulation: “Grammaticalization consists in the increase of the range of a morpheme advancing from a lexical to a grammatical or from a less grammatical to a more grammatical status, e.g. from a derivative formant to an inflectional one”. It is only from the end of the last century that the idea that grammaticalisation also involves segments above the word begins to gain ground; see for instance Lehmann (1982 [2015]) who, in his influencial Thought on grammaticalisation, includes in the scope of the notion both morphemes and whole constructions. At any rate, although a mention to constructions appears occasionally in works on grammaticalisation, the constructional character of the phenomenon has not been thematised until re- cently. This thematisation mainly concerns Traugott’s reflection and embraces the debate with other colleagues working within the grammaticalisation framework. In an article significantly entitled “Constructions in grammaticalisation”, Traugott (2003b) proposed to interpret grammaticalisation as the development of constructions, understood in a pretheoretical sense, i.e., as involving syntagmatic strings larger than single words. A similar position is taken by Bybee (2003: 602), who in the same volume where Traugott’s article appeared, states that: “In fact, it may be more accurate to say that a construction with particular lexical items in it becomes grammaticized24 , instead of saying that a lexical item becomes grammaticized”. We have already discussed the central role Heine assigns to the syntagmatic context in which grammaticalisation originates; however, he (2003: 581) cau- 24 Grammaticization is a competing term of grammaticalisation. For discussion about different theoretical implication above the two labels, see inter al. Lehmann (1982 [2015]: 11–12). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 75 tions against the possibility of univocally identifying grammaticalisation processes with constructions. This might seem surprising when considering his entire logic but, he argues, although it is irrefutable that the development of grammatical elements is shaped by the constructions in which these elements occur, “there are no convincing examples so far to suggest that instances of grammaticalisation processes can be identified exclusively in terms of constructions”. Heine’s caution reflects the barrage of criticism that grammaticalisation theories had to deal with in those years. These criticisms mainly concerned, in a nutshell, the risk of tracing any form of linguistic change back to grammaticalisation, considering also, we might add, the effects of the constructionist turn that was spreading in those very years and which extended the notion of construction from syntax to any conventional associations of form and meaning (Fillmore et al. 1988; Goldberg 1995). In his article, Heine addresses these criticisms point by point, often cutting down their size. In a different vein, one that denounces her militancy in the ranks of constructionism, Traugott (2003b: 645) clearly explains where the boundaries of grammaticalisation should be marked: A focus on strings or constructions rather than lexical items alone might appear to extend the domain of grammaticalization too far. It is true that little will be excluded from the study if we think of “grammaticalization” as an approach, a way of constructing the data, with focus on interactions between structure and use and on gradualness […] But grammaticalization is not coterminous with change. Phonological changes with no morphological effects will be excluded, as will semantic change involved in lexicalization, for example in the shift from one major category to another […], or word formation and compounding […] insofar as “grammaticalization” refers to a type of change, rather than an approach, it is not coterminous with morphosyntactic change. Rather it is […] the process whereby lexical material in highly contrained pragmatic and morphosyntactic contexts is assigned grammatical function, and once grammatical, is assigned increasingly grammatical, operator-like function. In more recent work by Traugott, the notion of construction takes on the more technical sense of a form-function pair (in a word, of any linguistic sign) and the entire construct of 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 76 grammaticalisation is explicitly subsumed under that of constructionalisation, understood as the creation of a formnew-meaningnew combination of signs (Traugott and Trousdale 2013: 22).25 In this perspective, grammaticalisation is one aspect or a set of special cases of constructionalisation. However, although certainly relevant to understanding some recent lines of research, we will not dwell further on the notion of constructionalisation, as it is not employed effectively in the articles collected here. In what follows, therefore, the term “construction” will be employed in the more traditional sense of morphosyntactic string. Within these strings, phenomena of change emerge that can be interpreted in terms of grammaticalisation. 1.4.3.2 Source concepts and patterns of change. We now take a step back to the source concepts underlying grammaticalisation processes, as described by Heine et al. (1991) and mentioned in 1.4.3. As seen, because they are grounded in basic, recurring patterns of experience, these concepts and the words that encode them are crosslinguistically frequent. In their fundamental review of conceptual shifts from lexicon to grammar, Heine and Kuteva (2002; Kuteva et al. 2019) report a long list of one-to-one correspondences between source and target. Although it might seem an oversimplification of the paths of grammaticalisation, which Heine himself defined as gradual and far from abrupt or merely metaphorical,26 it is 25 It may be useful to note that a constructionist approach underpins Langacker’s cognitive grammar, as described in 1.4.2; indeed, the cognitive models are explicitly inspired by construction grammar as first proposed by Fillmore et al. (1988), and subsequently allied quite intimately with Goldberg’s (1995) and other constructionist perspectives, such as Croft’s (2001, 2013) Radical Construction Grammar. In this sense, Traugott’s constructionist turn marks a further reduction in the internal distance to usage-based models. 26 The authors are perfectly aware of this possible contradiction and explain: “We are fully aware that this procedure rests on a gross simplification of the facts. [...] In the present book the perspective preferred is one according to which the process exhibits an interlocking pattern where uses of a grammaticalizing item share features with earlier, less grammaticalized uses of the same item. These uses, which have been described with reference to an “A>A/B>B” scenario […], exhibit a chain-like structure which cannot easily be divided into more or less separable ‘points’.” (Heine et al. 2019: 8). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 77 nevertheless a very useful frame of reference for appreciating certain recurrent patterns of change.27 Based on this framework and supplementing it with indications from other cognitivist and typologically informed studies, the attention will be focused here on a few source concepts that underlie the studies reported in this volume and the way they encode modal functions in various world languages. These are POSSESSION, DESIRE or VOLITION and (andative) MOTION, as well as a couple of other concepts less central to the discussion but no less relevant, namely POTENCY and KNOWLEDGE. Kuteva et al. (2019: 341–342, and reported references) report several cases in which a verb meaning ‘have’ (H-POSSESSION) develops a deontic meaning of necessity (or a menaing of obligation). This occurs for instance in Latin: VENIRE HABES come:INF have:2.SG ‘You have to come.’ Among Indo-European languages, the same pattern is found in German, where haben ‘have, own’ + zu ‘to’ > auxiliary of obligation (e.g., Er hat zu gehorchen ‘He has to obey’); in English, where have to functions as an obligation marker (e.g., You have to wash your hair); in Czech (mít ‘have’ + infinitive > ‘should, be to’) and Lithuanian (ture ‘have’ + infinitive > ‘must’); in Spanish (tener ‘hold, have’ > obligation auxiliary tener que + infinitive ‘have to’, ‘must’); and in some creoles (cf. Negerhollands ha ‘have’ + fo, conjunction > ‘must’, obligation marker). Similar patterns are found in various non-IndoEuropean languages. Heine et al. also observe that not only H-POSSESSION is involved in grammaticalisation. For instance, B-POSSESSION, that is, possession based on belonging, is found in German, where gehören ‘belong to’ > auxiliary marking deontic modality when involving participial main verbs (e.g. Er gehört eingesperrt ‘He should be/ought to be locked up’). 27 In fact, typological perspective appears to be more consistent with the grammaticalisation approach than the constructionalist one, since constructions and their history are irreducibly language-specific (Kuteva et al. 2019: 16). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 78 The relationship of possession is generally recognised as very basic and has in fact been the subject of wide attention in the cognitivist literature. In his work on subjectification, Langacker (1990: 29–33) sketched a schematic configuration of this notion (as the basis of perfective values, but the description also applies to our subject matter). He proposed that, in the process of abstraction that underlies grammaticalisation paths, the relationship of possession is assumed by languages to be a generic relation of control of the possessor over the possessed: Such considerations have led me to hypothesize that the linguistic category of possession has an abstract basis (i.e. a schematic characterization applicable to all class members) with respect to which ownership, part/whole, and kinship relations constitute special, prototypical cases. What all possessives share, I believe, is that one entity (the “possessor”) is used as a reference point (R) for purposes of establishing mental contact with another, the target (T). (Langacker 1990: 30). This configuration is described as in Figure 5: Figure 5. Basic relation of possession (Langacker 1990: 29, adapted). Figure 5 shows a relationship between a possessor (tr) and a possessed entity (lm), represented as a transmission of energy (i.e., a direct physical control) from the former to the latter (the double dotted arrow). However, this control is rather attenuated under normal circumstances, that is, merely potential or habitual (since “a person who possesses an ob- 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 79 ject – Langacker 1990: 29 explains – does not necessarily hold it or make any other kind of physical contact: I do have a shovel, but I am not now using it or touching it; and if I have a bar of gold, I will probably store it in a vault and never touch it at all”). This potentiality is further attenuated in the process of abstraction underlying its generalisation. This merely abstract and metaphorical possession, expressed by the single dotted arrow, corresponds to the mental path through which the participants in the speech act establish a contact between R and the target (e.g., I have a privilege, a right, a prerogative). According to Langacker, this function as a point of reference is what all possession relations have in common, even though it may be overshadowed, from time to time, by more objective notions with greater conceptual content (e.g., ownership). And it is precisely this schematic relation that triggers various shifts towards grammar. In the case of the shift towards modality, we could adapt Langacker’s model by saying that the target is, in this case, an activity to which the reference point is not yet related and in fact must relate, from the speaker’s point of view.28 It is generally acknowledged in both the typological and cognitivist literature (see, for example, Langacker’s 1991 proposals, discussed in 1.4.2) that deontic meanings are more basic than epistemic ones (and precede them historically); the latter, in fact, no longer refer to a reality in the future (something that has to be or can be done by a participant in the event), but to a mental attitude of the speaker. This chronological dependence, which also involves the semantic network of possession-based modals, has been studied from a typological perspective by Bybee and colleagues (see in particular Bybee et al. 1994). More in the spirit of the initial usage-based studies on modals, they interpret the shift from deontic to epistemic in metahorical terms. See for instance Bybee and Pagliuca (1985: 73), who proposed that: “The obligation sense of have to predicates certain conditions on a willful agent: x is obliged to y. The epistemic sense is a metaphorical extension of obligation to 28 Less nuanced metaphorical interpretations can be found in the literature. See for example Bybee et al. (1994: 184): “These constructions indicate that an agent possesses [...] an activity”, and this activity – they continue, is a projected one”. The frequent extension of H-POSSESSION to intentional an/or futural meanings depends on this aspect (Bybee et al. 1994: 178, 240; Kuteva et al. 2019: 288–289, 340). An articulated metaphorical interpretation is due to Talmy (2000: 449–452) which is based on his force dynamics model. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 80 apply to the truth of a proposition: X (a proposition) is obliged to be true”. Regardless of the acceptability of a merely metaphorical interpretation (see section 1.4.1 for a discussion of this point), there is historical evidence that epistemic meanings follow deontic ones and not vice versa (Bybee et al. 1994: 195; Heine et al. 2019: 288–289).29 Another common source for modal meaning is desire (towards an object) or volition (towards an event). Desire is generally understood as more basic than volition towards the realisation of a state of affairs, although it is far from easy to find diachronic evidence of this path of development, which in most cases can only be reconstructed. This reconstruction can also benefit of the conceptual and semantic analogy with possession (desiring an object implies a transfer of energy, albeit only potential and projective) or, alternatively, with motion (the subject of desire can be conceived as moving psychologically towards the object of desire). In both cases, we can refer to a schematic configuration very close to that in Figure 5, that is, a minimal action chain initiated by one entity transmitting energy (physical and, by extension, metaphorical) to another entity, in a temporal sequence that fundamentally embodies an oriented spatial configuration (which is obviously only conceptual), with a starting point, a path, and an end point (cf. Langacker 1991: 285).30 Kuteva et al. (2019: 453–457) list three main groups of frequent modal or modal-like meanings based on desire or volition: deontic necessity, proximative (an aspectual-like meaning) and future (a temporal meaning closely related to modality). An obligation construction (‘ought to’) based on volition is found in English, presumably developed in the early twentieth century, which appears to be strongly associated with second person subject referents (e.g., You want to mind your step). Other transparent cases are in Mehri (spoken in Yemen and Oman), where ḥōm ‘want’ developed in a cohortative marker (e.g., nǝḥōm nǝghōm ‘let’s go’), and in the Amazonian language Tucano, where ia ‘want (auxi- 29 Interestingly, both studies report typologically different examples that do not include the Romance languages, let alone Sicilian. Cf. Seychelles French-based Creole: nu it bezuȇ:HAVE.TO dezan ȃ-vil “we had to go to town”, i bezuȇ:HAVE.TO pe ale “he is probably leaving” (Heine et al. 2019: 289). 30 In the physical-spatial correlates of these configurations, we can observe a change of state and a change of position respectively, the latter being source concept for the former on the basis of the metaphor CHANGE OF STATE IS CHANGE OF POSITION, itself derived from STATES ARE LOCATIONS (Lakoff 1987). 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 81 liary)’ conveys impersonal obligation when occurring with impersonal subjects. Other languages recruiting ‘want’/‘wish’ verbs to encode general necessity or obligation are Abkhaz, Awa Pit, Modern Chinese, Kannada, Marathi, Maricopa, and Punjabi (Bybee et al. 1994: 186: Kuteva et al. 2019: 455; Narrog 2012: 161). As (aspectual) proximatives (‘being about to’), want-verbs are found in Old English (Hit wolde dagian ‘The day was about to break’), Persian (Mixast:WANT.PST.3SG bemirad ‘He was about to die’), Hungarian, Old Chinese, Ewe etc., as reported in Kuteva et al. (2019: 455–457). They observe (p. 457) that this path of development derives from a context-induced interpretation when the main verb denotes an event that the subject is not expected to actually want to occur, e.g. ‘to die’, ‘to be lost’; in these cases, a proximative meaning emerges as more probable than the volitional one; a further step occurs when ‘to want’ is generalised to inanimate subjects, so that the volitional meaning is suspended and only the proximative one survives. Future meanings are found in a large number of non-related languages, such as. Rumenian (voi cînta ‘I will sing’ < Lat. VOLO CANTARE ‘I want to sing’), Modern Greek (thelô ina ‘I wish that’ > tha, future tense), Old English (willan > will; see Aijmer 1985: 13–14; Traugott 1989: 38), Bulgarian (Toj šte:WANT.3SG diode ‘He will come’), Persian (xwāstan ‘want, wish’, verb > future tense marker); and, beyond the Indo-European domain, Swahili and Kibundu (Kuteva et al. 2019: 453–454). In a function-to-form perspective, Bybee et al. (1994: 240) summarise the possible lines of development of obligation and desire as in Figures 6 and 7 respectively. These figures are referred to a cross-linguistic scenario and also include other modal meanings in addition to those sketched in this section (which is a preamble to the cases discussed in the next sections and, therefore, deliberately limited in its scope). The two figures exhibits a map structure, which has diachronic value, but also represents possible synchronic relationships (as we will see better in 1.4.4). The second figure shows similar paths for ‘motion towards’ and ‘desire’, the latter, as seen, metaphorically referring to an oriented directionality and, like motion, inherently projective. Two other concepts deserve a quick mention, which underpin a modal sub-domain more marginal than the others for the work in the following sections, but nevertheless central to research on modality. They are knowledge and potency, both of which can be once 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 82 again traced back to the fundamental pattern of the transmission of energy (actual or potential) from one participant to another that we have already seen in Figure 5. The modal sub-domain emanating from them is that of ability and, hence, possibility. Figure 6. Path of development from obligation (Bybee et al. 1994: 240). Figure 7. Path of development from desire and movement (Bybee et al. 1994: 240). Knowledge is the most common source concept for the expression of ability. Bybee et al. (1994: 190) indicate a wide range of languages where mental ability can be traced back to 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 83 lexical verbs meaning ‘to know’ or ‘to know how to’. These are first restricted to mental ability and then generalise to physical ability, thus becoming general signals of ability (e.g., the Danish auxiliary kunne, cognate with English can; English, where to knows how refers to mental ability, as in I know how to speak French, and also to physical ability, as in I know how to shoot a crossbow). Bybee et al. do not mention Romance languages, where the continuations of the Latin SAPIO, SAPERE ‘to know’ also express mental and physical ability, along with the lexical sense of knowledge. It is worth noting en passant, however, that it is only in late Latin that SAPIO acquires the sense of knowledge, replacing SCIO and resigning the original sense of ‘to taste’; this sense remains in some marginal uses in Romance languages, e.g., It. quest’acqua sa di uovo ‘this water tastes like egg’ and in the noun sapore ‘taste’.31 On the other hand, from the concept of potency first physical ability develops which then expand to mental ability. This is the case with Latin POSSUM, POSSE (*POTERE) ‘to be able’, related to the adjective POTENS ‘powerful’ and from which French pouvoir, Spanish poder, Italian potere arose, which mean ‘can’ as auxiliaries, but retain the basic meaning of ‘power’ as nouns. Bybee et al. (1994: 90; Bybee 1988) proposed that “these verbs generalize in the same way as the mental ability verbs: since physical and mental ability often overlap, they would be used in many cases where both types of ability are intended and thus come to predicate both”. Both in the case of knowledge and in that of verbs meaning ‘to have the power’, ability easily shift to possibility. This happens when the enabling condition to perform a certain action no longer reside in the participant (as in the case of ability), but depend on the external world. A further step is permission, whereby the external conditions reside in the speaker (Bybee et al. 1994: 192–193). These shifts are summarised in Figure 8. Bybee et al. (1994: 194, 204–205) describe this path for English can, but also indicate other nonrelated languages where a similar development can be observed, e.g., Lao (aat), Basque (subjunctive form of the auxiliaries edin or ezan and the infix -ke), Cantonese (hó nàng), and Kui (duhpa). See also Narrog (2012: 122–124), who recontruct a path abili- 31 Whether or not this influences the development of the modal sense is yet to be established, as is the study of the modal network of SAPERE continuators in the different Italo-Romance varieties. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 84 ty>possibility>epistemic possibility for the American Spanish capaz ‘capable’. It would be interesting to verify the consistency of this path for the Sicilian marker capaci (ca), analysed by Cruschina only in synchrony (see 1.3.3). More generally, the internal shifts within this modal sub-domain in the Romance language would deserve further exploration. Figure 8. Paths of development from ability, 1 (Bybee et al. 1994: 194). Figure 9 shows the shift from possibility to epistemic possibility: Figure 9. Paths of development from ability, 2 (Bybee et al. 1994: 199). More central to the themes addressed in the next sections, we find ‘motion (from)’. We are dealing with an extremely basic schematic import, which has received ample attention in cognitivist literature and is here represented as in Figure 10. This involves a mobile entity 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 85 (a mover, tr) moving along a path (the oriented arrow) from a starting point (source, S) to an end point (E). Figure 10. Basic motion event. Verbs meaning ‘to go’ in world’s languages focus the displacement from the point of departure, i.e., they are source-oriented. In some languages they are also deictic, in that the speaker’s point of view (G) is located in the source (cf. I go home at midnight, where the temporal indication coincides with departure, cf. Fillmore 1975). I will not dwell on this last characterisation, as it will be discussed at length in the dedicated section (see 5.1.1). However, it is essential to note that subjectification is a central factor in the development of most abstract meanings of ‘go’ verbs (cf. Langacker 1990: 23 on the development of future meanings in English). Indeed, deictic motion verbs (not only the andative ‘to go’, but also the ventive ‘to come’) underlie many grammaticalisation paths in the world’s languages. Some of these paths are to date only marginally documented and studied; among, we find the development of directive markers from ‘go’ verbs, described by Craig (1991) for Rama (where bang ‘go’ > first person plural imperative suffix). Various French-based Creoles also developed a first person plural imperative marker from allons ‘we go, let’s go’ > anõ, anu, ãn, ãn (Kuteva et al. 2019: 209). More recently, by Mauri and Sansò (2014; also based on Nicolle 2007) on the basis of on a wide typological survey, proposed that the directive value develops when ‘go’ cooccurs with another verb, which is hence the content of the command (the event to be performed, so the speaker is asked to go to do tr S E ‘to go’ G 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 86 something). As will become clear in the next section, we find ourselves here on the periphery of modal values and indeed, according to some authors, beyond the modal domain and within the contiguous but distinct domain of illocution (what we referred to in 1.4.2 as mands). 1.4.4 Semantic topography. The descriptions of the various paths of development presented so far subtend, as we have seen, a spatial logic. They can in fact be represented by means of maps, i.e. geometric representations of meanings within a space. This is a description model developed since the 1980s in the field of linguistic typology to account for crosslinguistic regularities and to differentiate between what is general or universal and what is language specific. Semantic maps make it possible to describe the polysemy of linguistic categories, both grammatical and lexical. The different meanings (or rather functions, as suggested by Haspelmath 1997) are arranged in the map space according to the degree of semantic distance. As Georgakopoulos (2019) well summarises, “the main idea of the method is that the spatial arrangement of the various meanings reflects their degree of (dis)similarity: the more similar the meaning, the closer they are”, according to the so-called connectivity hypothesis formulated by Croft (2001). Although this is an admittedly synchronic approach, maps can also incorporate information on the directionality of changes, which is indicated by oriented arrows, as we have seen in the descriptions of Bybee et al. (1994) and as is also the case in other studies, where the diachronic dimension is inherent (cf. van der Auwera and Plungian 1998, to be discussed shortly). Fundamental reflections on the nature of maps are in Haspelmath (2003) and van der Auwera (2013); a fairly comprehensive review of studies on semantic maps is in Georgakopoulos (2019). The focus will now be on the representation of the semantic space of modality, as proposed by van der Auwera and Plungian (1998), which in addition to being one of the most systematic crosslinguistic descriptions of this domain (and certainly the most systematic within the map approach), has assumed an increasingly central role in my work, especially in terms of the internal organisation of the notional domain and, thus, also in terms 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 87 of labelling. van der Auwera and Plungian (1998: 79) are explicitly inspired by Bybee et al. (1994, to which I have repeatedly referred in the previous section), on whose typological data they base their generalisations and attempt to map out a way “to make predictions about the ways languages express modality.” However, compared to the proposals of Bybee et al., they mark the boundaries of the modal domain differently. First, they limit it to the meanings of necessity and possibility, which are in any case central – though not exclusive – values in all approaches. These two options are then divided into four dominies (van der Auwera and Plungian 1998: 80–81): (1) Participant internal modality, that is, a type of possibility or necessity internal to the participant involved in the state of affairs; these meaning are elsewhere called ability and necessity respectively (e.g., Boris can sleep ten hours a night; Boris needs to sleep ten hours a night). (2) Participant-external modality, which refers to the circumstances that are external to the participant involved in the state of affairs and that make this state of affairs possible or necessary; this is called root modality in Bybee et al. (1994) (e.g., to get the station, you can/have to take the bus 66). (3) Deontic modality, which is a subdomain or a specialisation of participant-external modality that identifies the enabling or constraining circumstance as a persons (often the speaker) or as some social or ethical norms that enable or oblige the participant to engage in the state of affairs); this is articulated in permission and obligation in Bybee et al. (1994) (e.g., John may/must leave now). (4) Epistemic modality, which is referred to the judgment of the speaker, who considers the proposition as uncertain or probable (John may/must have arrived). The options in (1) to (4) can be summarised as non-epistemic modality (1-3) vs epistemic modality (4), a bipartition that is generally accepted in modality studies. Volition and (noninferential) evidentiality remain outside the domain of modality. Directives, which pertain to illocution, are also outside (as we shall see in the following sections, we include volition, but accept the distinction between modality and illocution). In line with Bybee et al., 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 88 van der Auwera and Plungian (1998: 87) focus only on those modal markers that are verbal and that are at least partially grammaticalised; thus, they exclude nominal (e.g., necessity), adjectival (e.g., possible), adverbial (e.g., perhaps) or mental state predicates (e.g., I think) markers, although it is assumed that what applies to the verb also applies to the other classes. A central aspect of van der Auwera and Plungian’s analysis is the attempt to unify, by readjusting them, the minimaps of necessity and possibility already drawn by Bybee et al. (1994) into a single map. This unified representation must also explicitly include, outside the boundaries of the modal domain, on the one side premodal (lexical) meanings from which the modal ones originate and, on the other side, the postmodal ones (e.g., future, imperative mood etc.). The decision to unite the paths in a single representation follows from the fact that necessity and possibility share some postmodal developments, but also some modal developments: for instance, they observe, deontic necessity may come out of deontic possibility (e.g., in Engl. now you may (=must) tell a story), and vice versa. The results of this systematisation are shown in Figure 11. Not unlike the paths traced by Bybee et al. (1994), van der Auwera and Plungian’s unified map is intended to have universal relevance, both in diachrony and synchrony. On the diachronic level, all modal meanings depend on premodal meanings, in a developmental path that is unidirectional. This unidirectionality is represented along the horizontal axis, from left to right, and also concerns successive shifts, within the modal domain and beyond. In some cases, a premodal meaning may also end up in the postmodal domain (e.g., future). On the vertical axis, on the other hand, there is no directionality at all, so that, van der Auwera and Plungian note, there can be bidirectional switches, for example, in the deontic domain between necessity and possibility, as noted above. The importance of this representation is self-evident, as it represents a reference model for any discourse on modality in the world’s languages. It should be noted that it is not so much a matter of adhering point by point to the semantic-functional organisation proposed by van der Auwera and Plungian (and in fact the articles presented below deviate from this organisation in many cases, excluding or more often including other segments). 1FRAMINGTHESTUDYOFMODALITYINSICILIAN:ANINTRODUCTION. Figure11.Modality’ssemanticmap(vanderAuweraandPlungian1998:111). 8 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. Rather, what is interesting and should be taken as a model is the logic of representation itself and its cross-linguistic value, against which the validity of individual and languagespecific representations should be measured. I will return to this in the concluding part of the volume, after examining the functional semantic networks of the various modal and modal-like markers covered in the studies presented. 1.5. Outline of the volume: paths of modality and postmodality. The works collected in the sections that make up the rest of the volume meet and develop the issues sketched so far in various ways, first of all setting them down in the specific reality of the Sicilian and experimenting their heuristic efficacy from time to time, in a process of fine-tuning the working tools. In a nutshell, it could be said that from the first article to the last, the approach remains the same: a usage-based view in which insights from cognitive linguistics and functionalism converge. However, the specific weight exerted by individual models gradually changes: from a clear prevalence of cognitive schematisations, which characterised the first proposals, to the increasing prevalence of a functional sensitivity and a growing focus on sociolinguistic and textual aspects. Along this theoretical path, the initial assumptions are not dismissed, but resized, if anything. We could say that the works collected operate between the two poles of the same theoretical continuum, held together by the common use of a stable interpretative notion, namely grammaticalisation. In what follows, an overview of the five articles is proposed, attempting to highlight the continuity among them and how each of them meet the various aspects of the framework. These studies are conceived as different paths to and through the modal domain, and beyond. The first path, “From possession and volition to modal meanings” (Section 2), is a reflection strongly rooted in the cognitivist view, but which incorporates several assumptions of functionalist approaches, in a common usage-based perspective. As the title itself reveals, the study takes as its starting point two source concepts, whose grammaticalisation towards modal and post-modal values is outlined. The first of these source concepts, possession, had already been investigated in the literature on Sicilian (cf. in particular Bent- 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 91 ley’s work discussed in 1.3.2) and is reconsidered here in the light of cognitive grammar and new data collected in the field. This reinterpretation thus provides an interpretative scheme for the analysis of volition, the other source concept investigated here for the first time in Sicilian (and, as far as I know, in Italo-Romance more generally). The two source concepts underlie two verbal constructions, respectively aviri a + infinitive and vuliri + infinitive, employed in an articulated network of modal, pre-modal and post-modal values, differently grammaticalised and functionally overlapping in some central (deonticnecessive) areas. The theoretical notions activated to analyse the data are the subjectivation à la Langacker (cf. 1.4.2) and the metaphorical-metonymical model, discussed in 1.4.3, according to which the grammaticalisation of entities is not produced in isolation, but is triggered by the context of use, through operations of a metonymical-inferential nature. The geometric representation of the semantic networks of the two modal constructions examined shows an early interest in spatial-topological logic (see 1.4.4), which will return in the final article (and, more systematically, in the concluding section of the volume). The interest in the natural datum and thus the sociolinguistic concern become more systematic in the second path, “From volition to modal-passive meanings” (Section 3). Originally published a few years after the first one, this new article in fact represents the irruption into the dimension of speakers, only foreshadowed in the cognitivist approach and here made operational through the in-depth investigation of (semi-spontaneous) speech data collected in the context of the Atlante Linguistico della Sicilia of the University of Palermo (see 1.2.2.3). This concrete plunge into the use of speakers allows us to formulate more precise hypotheses on the forms and degree of grammaticalisation of constructions that occur in specific linguistic uses, in discourse. In particular, the article explores the range of textual usages and morphosyntactic contexts of three constructions involving the volitional verb vuliri and the past participle of a lexical verb or a passive infinitive. These constructions convey a clear deontic meaning, alongside or as an alternative to the volitional one, which may interact with a different notional context, that is, the passive, according to a pattern crosslinguistically attested but never studied before (apart from sporadic observations in dialectology) for Sicilian. 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 92 The perspective shifts to ancient Sicilian in the third path, “From volition to directive meanings” (Section 4). The semantics of volition and its trajectories within and beyond modality are explored here, in search of the origins of the present values described in the previous contributions. The interest in the natural datum persists, naturally translating on the diamesic and textual level and, thus, drawing on an articulate corpus of ancient Sicilian, ARTESIA, developed at the University of Catania, which is representative of a wide range of genres and text types (see 1.2.2.1). Besides temporal and textual, the shift is also slightly theoretical, as the essay now stands more firmly in the functionalist dimension, which allows modal nuances, their continuities and discontinuities, to be sounded out more accurately. In particular, the centrality of Bybee’s and van der Auwera’s typologically informed proposals emerges, which, as we have seen, arrange meaning networks on topographical semantic spaces (see 1.4.3.2 and 1.4.4). The use of this theoretical model reveals a network of meanings already highly articulated in the 14th -16th centuries, in which vuliri has developed or is developing clear deontic values alongside the volitional sense. Vuliri, moreover, exceeds the boundaries of modality, expressing directive meanings, which develop from modal ones or, more frequently, emanate directly from the volitional one, intertwining with diverse needs of subjectification, including politeness (see 1.4.2.2). Some postmodal meanings are at this chronological stage only profiled, such as the proximativefutural meaning (see 1.4.3.2), which only shows up in a couple of examples and is directly linked to the volitional meaning. On the other hand, the semantic network of vuliri also includes other postvolitional and/or postmodal values, which were rather stable as early as the Middle Ages, i.e., optative formulae in the domain of illocution and various uses as discourse markers. Finally, the complex network of meanings conveyed by vuliri, as well as their selection constraints and mutual dependencies, are arranged on a comprehensive semantic table. The fourth path, “From (itive) motion to illocution” (Section 5), examines a different source concept, that of oriented motion and in particular displacement from the deictic perspective of the speaker. Like volition and possession, motion is a concept that underlies so many grammatical phenomena in the languages of the world, but is very little documented in this specific pathway to directive values (see 1.4.3.2). In Sicilian this function is 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 93 firmly associated with the construction employing an imperative marker based on ‘to go’ and the imperative of a lexical verb, in the structure [va IMP]. Given the crosslinguistic rarity of the pattern, the study calls for a closer dialogue with the typological literature and the patterns of change hypothesised in this field. The perspective adopted is mainly synchronic and is based, once again, on the data made available by the Atlante Linguistico della Sicilia. However, a preliminary diachronic survey is also proposed to assess the presence of the phenomenon at different chronological stages and to attempt to sketch a line of development. To this purpose, the ARTESIA corpus for the 14th -16th centuries and some representative examples of the rich but scattered existing literature for the later chronological phases, up to the threshold of the 20th century, were checked. Although partial, this chronological view allows us to follow the grammaticalisation of the construction at different moments, starting from a process of subjectivation rooted in the deictic dimension. By virtue of the presence of ‘to go’, the perspective from which the event denoted by the second verb is observed includes deixis: the perspective of the speaker, thus the order itself, becomes as central as the event (lexical verb). This process triggers certain changes that affect not only the morphosyntax of the forms involved, but also their interclausal organisation, investing the domain of subordination. This shift would originate precisely in directive contexts, to shape basic functions of dyadic interaction. The hypothesised paths of development undergone by the construction are then represented on a semantic map of primarily diachronic value. The topographical representation takes prominence in the last path, “A semantic map for modal verbs” (Section 6), devoted to an overview of the verbal marking of the nonepistemic (participant-oriented) domain encoded by different modal constructions and its areas of overlap with the directive domain. Constructions already dealt with in previous studies return in this new work (aviri a + infinito, vuliri + infinito) and some new modals are taken into account for the first time (putiri + infinito, sapiri + infinito). However, the perspective changes somewhat, as it is no longer exclusively semasiological as in previous works (focused on the semantic-functional networks of individual constructions), but includes an onomasiological view. In fact, here the focus shifts to the set of forms that encode a given semantic domain, that of the participant-oriented modality. Inspired by the 1 FRAMING THE STUDY OF MODALITY IN SICILIAN: AN INTRODUCTION. 94 literature on semantic maps (1.4.4), the functional spaces expressed by the single constructions are first sketched and then a unified map of the domain, as realised in contemporary Sicilian, is proposed. The interest is exclusively synchronic, so the representation accounts for connections among notionally contiguous functions, either within (e.g., necessive and deontic) or beyond modal boundaries (participant-oriented and directive), but does not systematically include assumptions about the directionality of changes and, therefore, does not assume the horizontal, unidirectional arrangement proposed in the relevant literature. The last path, “Towards a unified account: a semantic modality’s map for Sicilian” (Section 7), does not report on a previous study but serves to comment on and draw conclusions from what has been said before. At the same time, it imposes a renewed look at previous work as a whole, connecting more intimately with topographical literature and, in particular, van der Auwera and Plungian’s proposals for a unified map of synchronic and diachronic scope. After presenting a summary of the results obtained through the previous paths, now realigned in a more fully onomasiological perspective, a map of the modal domain is reconstructed that is highly consistent with the one proposed by van der Auwera and Plungian, both in terms of internal relations and in terms of the directionality of changes, but which makes some divergent choices with respect to the latter. First of all, volition is not relegated to the premodal domain, but penetrates, albeit marginally, the modal one. Secondly, areas of continuity with contiguous domains in the field of illocution are emphasised, which are encoded not only by basically modal constructions, but also by morphological means (verbal moods), lexical markers, other non-modal constructions. Although it exceeds the nuclear area of modality, the map remains fuzzy in several of its areas, especially at the margins, where numerous phenomena only marginally observed for Sicilian accumulate (e.g., instances of lexical expression of epistemic modality or syntactic expression of mands), thus representing clear research directions that will allow for the complexification of modality’s semantic map of Sicilian. 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 95 2 First path. From possession and volition to modal meanings. “Possession and volition in the development of modal meanings: A case study from Sicilian”, 2019 [2009].32 2.1. Introduction. The Sicilian, an Italo-Romance variety spoken in the extreme South of Italy, has an articulated analytic modal system involving different modal verbs constructed with the infinitive form (INF) of a lexical verb. The attention will be here focused on two modal constructions, namely aviri a ‘to have to’ + INF and vuliri ‘to want’ + INF, which, beside modal values, are used to express futurity (Sicilian lacks a synthetic form of future). These constructions have already been described in the dialectological tradition (Leone 1995; Rohlfs 1969) and have also been individually analysed through diverse theoretical models (see Amenta 2004, 2010a; Amenta and Mocciaro 2016, 2018; Bentley 1997b, 1998a, 1998b, 1998c, 2000c; Brucale and Mocciaro 2009). We will refer to these previous works during the discussion. However, a comprehensive account of the semantic networks of aviri a and 32 Originally published as: Luisa Brucale and Egle Mocciaro (2019), Possession and volition in the development of modal meanings: a case study from Sicilian. Journal of Contemporary Philology 2(2): 27–47. This work is a revised version of an earlier article by the authors, namely: Polisemia e convergenze nel dominio dei modali in siciliano: una lettura funzional-cognitivista di vuliri e aviri a. In Luisa Amenta and Giuseppe Paternostro (a c. di) (2009), I parlanti e le loro storie. Competenze linguistiche, strategie comunicative, livelli di analisi, 195–206. Palermo: Centro di studi filologici e linguistici siciliani. Here, I report the second version (2019), in English in the original, placing it at the very beginning of the proposed collection. This is because, although extensively revised, it retains the original theoretical framework and, therefore, represents well the first approach to the phenomenon of modality that has characterised my research. As reported in the source article, “[t]he whole paper results from close collaboration of both authors. However, for academic purposes, Luisa Brucale is responsible for 1 and 3 [here, 2.1 and 2.4] and Egle Mocciaro for 2 and 4 [2.2, 2.3 and 2.5, that is, the theoretical description and the section dedicated to volition]. The final section represents a joint effort on the part of both authors. We use a strict alphabetical order which does not reflect any priority of contribution.” 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 96 vuliri + INF and their semantic overlap is still lacking, as is the description of their interaction with temporal-aspectual inflection. The corpus is built on data elicited from semi-guided interviews with speakers of different local varieties of Sicilian (Agrigento, Caltanissetta, Catania, Palermo, Siracusa). We have also compared these data with not yet published data from the Atlante Linguistico della Sicilia ‘Linguistic Atlas of Sicily’ (ALS) (cf. D’Agostino and Ruffino 2005; Ruffino 1995; Sottile 2019; see 1.2.2.3).33 Data analysis is largely based on Cognitive grammar (Langacker 1991). While original at the theoretical level, this approach does not substantially deviate, however, from the traditional definition of modality as a grammatical category expressing the speaker’s attitude towards her utterance or towards the event contained in the proposition (Jespersen 1924; Lyons 1968; Palmer 1986 [2001]). The vagueness of this definition reflects the notional extent of the modal domain, which is in fact frequently analysed according to more specific parameters which further structure its internal space. In the reference literature, in fact, there is no agreement on number, types and limits of the features which define the modal domain. Notional pairs such as realis/irrealis, possibility/necessity, objective/subjective, volitionality/non-volitionality have been repeatedly included, rejected or redefined depending on the theoretical assumptions (cf. Bybee et al. 1994; Givón 1990; Hengeveld 2004; Langacker 1991; Lyons 1977; Narrog 2005; Nuyts 2001; Traugott 1989; van der Auwera and Plungian 1998). Probably the most neutral labels are epistemic vs non-epistemic. Epistemic modality refers to the speaker’s commitment to the truth of the proposition, along a continuum including possibility, probability, inferred certainty (and/or evidentiality) and prediction. Non-epistemic modality, in turn, is oriented on the participant, that is, it signals a condition on the participant (e.g., an agent) expressed as permission, obligation, necessity, ability, possibility and so forth. Also in this case, however, the organisation and the internal relationships within the two sub-domains are anything but homogeneous, especially in the case of non-epistemic modality (cf. Bybee et al. 1994; 33 We are grateful to Giovanni Ruffino and Mari D’Agostino for making the transcribed materials accessible to our research. 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 97 Hengeveld 2004; van der Auwera and Plungian 1998). Starting from the observation that both modal sub-domains display not only an internal structure but also a high degree of interrelation and overlap, we will adopt a scalar approach as the basis of our reasoning. Modals, in fact, express different semantic nuances depending on their various interaction with different parameters and contextual coordinates (aspect, actionality, as well as transitivity features, such as agentivity). The paper is organised as follows: in section 2.2 we illustrate the theoretical framework within which the analysis is placed; in sections 2.3 and 2.4 we analyse the nonepistemic and epistemic uses of aviri a and vuliri and summarise the results in two modal maps; finally, in section 2.5, we discuss the impact of the temporal-aspectual features on the modal meanings. 2.2 Theoretical assumptions. Langacker (1991: 548–549) defines modality as a strategy of grounding which locates the described event vis-à-vis the Ground, that is, “the speech event, its participants, and its immediate circumstances”. The basic distinction between epistemic and non-epistemic is maintained, although reinterpreted on the basis of the assumptions of the cognitive framework. In the cognitive approaches, non-epistemic (participant-oriented or root or deontic) values are generally claimed to be more basic than epistemic ones. According to Talmy (1988), deontic modality expresses a force-dynamic relationship between the speaker and the described event; in other words, since the speaker is portrayed as the locus of the potency directed towards the realisation of the event, the deontically modalised utterance (e.g., an order or a necessity) expresses a more concrete meaning than a judgement of truth. Following Talmy (1988), Sweetser (1990: 49–75) claims that the deontic→epistemic shifts depend on a metaphorical mapping from the concrete domain of socio-physical forces to the abstract domain of reasoning. This mapping accounts for the difference between deontic and epistemic meanings of the same modal, such as in You must go (which can be 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 98 analysed as ‘An authority compels you to go’) and You must be tired (‘A direct evidence compels me to the conclusion that you are tired’). Nevertheless, while the metaphorical interpretation is undoubtedly useful to explain the notional link between the source and the target domains, the inherent discrete character of this notion fails to account for the scalarity of modal meanings. Other scholars, in fact, reinterpret the role of metaphor intertwining it with a metonymical interpretation which accounts for the gradual character of process. This process is context-dependent, that is, it originates within the linguistic context through metonymical and inferential operations. In particular, metonymical operations include: (1) Emphasising a secondary or implicational feature of the modal meaning. (2) Defocusing a primary semantic meaning (cf. Bybee et al. 1994: 196–197; Goossens 1995; Heine 1993; Hopper and Traugott 1993 [2003]). 2.2.1 Subjectification. Langacker (1991) analyses modal sub-domains as different results of subjectification. This consists in the realignment of (a facet of) the relationship between the entities involved in the event. In a non-modalised utterance (e.g., Egle goes to Paris), these entities are: 1. The Trajector (tr), that is, a salient, moveable and active entity (in the example re ported, the subject, Egle). 2. The Landmark (lm), that is, the components of the event to be realised by the tr (in the example, goes to Paris). Different results of the realignment of the relationship between tr and lm are schematically represented in Figure 12 and in Figure 13. f 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 99 Figure 12. Objectively construed relation (Langacker 1991: 216, adapted). Figure 13. Subjectively construed relation (Langacker 1991: 216, adapted). 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 100 G can be focused or defocused from the representation of the event (i.e., the profile of the event), becoming progressively less salient, along the following continuum: S (i.e., an objectively-construed participant) → the speaker → some other individual → an unspecified facet of the physical or social world → the logical structure of reality (see Langacker 1991: 271–272). Figure 14 (elaborated on the basis of Langacker 1991: 215–220) schematically represents the different degrees of salience of G, which occupies different positions with respect to the stage where the event is represented. Figure 14. Degrees of salience of the Ground (G). This interpretation highlights the gradualness of modal shifts and, therefore, it is consistent with the historical development of modals (Traugott 1989)34 and, in particular, with the 34 In her seminal (1989) paper on the rise of epistemicity in English, E. C. Traugott identifies one of the main tendencies in semantic change in the increasing subjectification of meanings. More specifically, Traugott (1989: 31–43) analyses the development of epistemic modals from an initial stage in which they were main verbs; crossing the stage of “pre-modals” these verbs have gradually acquired a deontic, weakly subjective, 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 101 metonymical or metaphorical-metonymical models of description (Goossens 1995; Heine et al. 1991; Heine 1993; Hopper and Traugott 1993 [2003]). 2.3 Aviri a + INF. The analysis has shown a great degree of polysemy for aviri a + INF, whose meanings range around two prototypical cores, one necessive and one epistemic. In a cross-linguistic perspective, Bybee et al. (1994: 184–186) have shown that the ‘have to’ + INF constructions are frequently associated to an obligational meaning which may derive both from the finite (‘have to’) and the infinitive forms of the verb (lexical verb). Indeed, ‘have’ indicates that an agent metaphorically possesses (has) an activity that, being expressed by an INF, lacks temporal grounding and may be conceived of as not yet actual one and, in fact, as a projected one. Consequently, the periphrasis conveys both modal meaning and temporal information (i.e., a non-past and a non-present sense).35 The presence of the preposition a ‘to’ contributes to make explicit the directional/projective character of the construction. We can represent the schematic import of the periphrasis as in Figure 15: modal meaning. The diachronic investigation on English shows that when epistemic meanings arise, they concern weakly subjective meanings (such as the evidential uses of *sculan ‘to owe’ which do not require a strong involvement of the speaker), which, only later, develop a more strongly subjective epistemicity. 35 This also explains why the ‘have to’ + INF constructions are linked with a future meaning. As Bybee et al. (1994: 261–262) pointed out, future can develop from the obligational meaning, or (based on a proposal by Benveniste 1968) it can arise directly from the notion of ‘projection’ (towards the content of the infinitive). It should be observed that in Romance languages obligational constructions built on markers of possession always involve a prepositional marker. This can be a directional preposition, as in the case of Sicilian a, as well as French à and Italian da, or another marker of possession, such as Spanish and Portuguese de ‘of’. At least in a few local varieties of Sicilian, the preposition di ‘of‘ coexists with a in the ‘have to’ constructions. Although we have not yet investigated their functional distribution, a cursory look at the phenomenon in question suggests that, while a is generally employed both in the deontic and in the epistemic domain, di is univocally linked with the former, as in Aiu di fari ‘na cosa ‘I have something to do’. This is, however, a less stable periphrasis if compared with aviri a + INF, in that the possessive value of aviri is still transparent, as the presence of the direct object shows (Aiu di fari ‘na cosa / Aiu‘na cosa di fari). 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 102 Figure 15. Schematic import of aviri a. This explanation is consistent with the diachronic analysis proposed by Bentley (1998b, 2000c), who claimed that the primary chronological sense of aviri a is deontic. It is linked with the notion of necessity of an action to be performed by a morally responsible agent and, therefore, it contains an element of ‘will’ (see also Heine 1995). In Sicilian, aviri a + INF means compelling necessity imposed on an agent by a certain state of affairs or, in other contexts, an obligation imposed by an authoritative speaker to the hearer. In other words, the construction expresses both a strictly modal (necessive) sense and a directive one.36 The inherent features acting on these domains can be formalised as (external) Necessity (N) and Intentionality (I), which represent forces pushing the tr towards the realisation of the lm (cf. Brucale and Mocciaro 2009). These forces differ in the alignment. Necessity is, in this case, an externally imposed force (i.e., it lies in G), while Intentionality is internal to the participant who carries out the event expressed in the proposition.37 Necessity and Intentionality are scalar and interacting values. The structure of the locus of potency ranks the meaning of the construction along a continuum from compelling necessity to obligation, interacting with the degree of Intentionality of the tr. 36 In the context of the research on modality, there are different interpretations of the relationship between modality and directivity. In particular, Bybee et al. (1994: 178–179) interpret directivity as a specific type of modality, that is, a speaker-oriented modality, which is aimed at provoking a certain state of affairs. On the other hand, van der Auwera and Plungian (1998: 83) keep the two notions distinct. 37 Bentley (1998b: 128–129) lists among the deontic uses of aviri a also some illocutionary types such as M’av’a ddiri paroli di me matri ‘(s)he just has to diss my mother’; this use, expressing an accusation or a complaint, can be classified as agent-oriented. Nevertheless, she points out that in the literature such uses – in which the aviri a construction has the role of clarifying the communicative purpose of the utterance – are not always numbered among the modal categories (cf. Narrog 2005: 680; van der Auwera and Plungian 1998 inter al.). 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 103 In (1), an obligation is imposed by the speaker on an agentive hearer (Tu), thus the source of the modal value is the speaker and the obligation is bound to the specific speech act. In (2) and (3), the locus of potency lies in cultural or physical conditions. Since these conditions are neither speaker-dependent nor directly linked to the speech situation, (2) and (3) express a general and external necessity rather than directive obligation (i.e., an order). Moreover, in (3), the obligational reading is also blocked because the physical necessity acts on an inanimate and non-intentional subject (U pummaroru). (1) Tu a’ (a) ffari comu ti dicu iu. you have:PRS.2SG (to) do:INF as you:DAT say:PRS.1SG I ‘You must/have to do as I say.’ (2) T’ a’ (a) maritari ‘n chiesa manzinnò ti REFL.2SG have:PRS.2SG (to) marry:INF in church otherwise REFL.2SG nni va u ‘mfernu. away go:PRS.2SG to.the hell ‘You must/have to get married in church otherwise you’ll go to hell.’ (3) U pummaroru bbonu av’38 a stari o suli. the.M tomato good have:PRS.3SG to stay:INF at.the sun ‘Good tomatoes must (need to) stay in the sun.’ Since inanimacy means a non-intentional subject, Bentley (1998b: 130) excludes the sentences with inanimate subjects from the deontic domain; rather, if an originally deontic or intentional structure occurs with an inanimate subject, this fact represents the threshold beyond which the periphrasis can be understood as shifting towards a different level (namely, epistemic or future-like). While we fully agree with the non-obligational interpretation she proposes, our scalar approach allows us to range both agentive and non-agentive constructions within the same continuum. In particular, while the highest degree of Intentionality is found in constructions with animate subjects, this value is faded or only meta- 38 Depending on the varieties, the third person of the present of aviri is av(a) or a. 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 104 phorically expressed in constructions with inanimate subjects. On the other hand, Necessity persists in constructions with inanimate subjects, as well as the forward-push dynamics. Via a metonymical shift, the compelling necessity is translated into a logicaldeductive (inferential) dimension, which can be inferred from the context, as in (4), due to the presence of the verb pinzari ‘to think’; in this case, the force compelling the event is the logic of the reasoning (‘I’m forced to think this because of the logical coherence of reasoning’), which produces an epistemic interpretation (‘it is highly probable that…’): (4) Si ddici d’accussi aiu a pinzari ca if say:PRS.2SG so have:PRS.1SG to thinf:INF that si babbu. be:PRS.1SG stupid ‘If you say something like that, I have to think that you’re stupid.’ It is likely that these uses have paved the way for a proper epistemic interpretation, as in (5) and (6), in which the speaker clearly expresses her opinion about the degree of probability and reliability of the proposition’s content. This is a complex operation which consists in: 1) triggering a range of presuppositions depending on the extra-linguistic situation; 2) inferring from these presuppositions one or more virtual beliefs; 3) evaluating these beliefs and rejecting all the possible conclusions but one, which is then presented as the only valid conclusion at the moment of the speech act (cf. Dendale 1994: 28). Therefore, in this use aviri a expresses not only the epistemic quality of the information, but also an epistemic/inferential operation, which is evidential in nature according to Dendale (1994). (5) Annu a essiri i tri. have:PRS.3PL to be:INF the.PL three ‘It must be three (o’clock).’ (6) Av’ a manciari quant’ un porcu ppi have:PRS.3SG to eat:INF like a pig for 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 105 iddu essiri accussì grossu. he be:INF so fat ‘He must eat like a pig for him to be so fat.’ It is worth noting that the epistemic reading is strongly linked to the actional value of the verb (Langacker 1991: 275–281; see also Pietrandrea 2005): it is allowed only in atelic situations, i.e., with stative verbs, typically essere ‘to be’, as in (5), and avere ‘to have’, or activities, the latter adding a habitual interpretation, as manciari ‘to eat’ in (6). A similar inference forces the periphrasis to signify a prediction, putting aviri a in a future-like domain. While the previous uses pertain to opinions about present situations “that are not yet part of the reality known by the speaker” (i.e., a potential reality, cf. Langacker 1991: 278), in the prediction domain we are dealing with states of affairs which are not yet part of the actual reality, i.e., with a future and projected reality. Especially in the case of 1st person subjects, which are obviously co-referent with the speaker, prediction is a direct implication of Intentionality, rather than of the speaker’s opinion. This is the case in (7), in which the statement of obligation strongly implies that the speaker intends to do something (cf. Bybee et al. 1994: 264). On the other hand, when a 3rd person subject is involved, the statement of her intention implies a prediction which is based on Necessity. In other words, (8) and (9) portray a reality which is expected by the speaker, in that a compelling logical necessity determines the certain and reliable realisation of a future event:39 (7) T’ aiu a rregalari un palluni. you:DAT have:PRS.1SG to give:INF a ball ‘I have to give you a ball / I’m going to give you a ball.’ 39 It is not surprising that 2nd person subjects are less suitable candidate to be involved in a construction expressing prediction, since they are naturally linked with the speech act, thus with the obligational reading. 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 106 (8) St’ annu u me zzitu m’ av’ a this year the.M my fiancé:M I:DAT have:PRS.3SG to ffari u bberloccu. do:INF the.M diamond ‘This year my fiancé will give me a diamond ring.’ (9) C’è Pino? EXS.PRS.3SG Pino No, av’ a bbeniri. no have:PRS.3SG to come:INF ‘Is Pino there? No, he’s going to come.’ It is worth noting that present-day Sicilian, unlike Italian, lacks a synthetic future; hence, the periphrasis is widely employed as a marker of future (beside the present tense). In these cases, Necessity interacts with Intentionality, due to the presence of an agentive tr, who is realising a non-stative situation, typically a telic action, as in (7) to (9). On the other hand, (10) expresses a non-agentive activity, i.e., ‘to rain’, which (together with the other natural events) represents the most typical instance of future-like value in Sicilian: (10) Av’ a chioviri. have:PRS.3SG to rain:INF ‘It’s going to rain.’ Since a non-agentive activity is involved, the feature of I is faded (or only metaphorically attributed to the event)40 and the only modal feature at work is Necessity, i.e., the logical necessity of the speaker’s reasoning, as in the other non-agentive case exemplified in (3). 40 Note, however, that meteorological predicates imply the existence of a natural force which acts on the realisation of the meteorological event. Luraghi (1995) analyses natural forces (together with emotions and diseases) as less prototypical agents in respect to the human beings, in that they actually control the event and, differently from the inanimate objects, they cannot be manipulated by another human entity. 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 107 In Figure 16, we propose a synopsis of the modal values of aviri a + INF, which highlights the synchronic gradual shift from deontic to epistemic, as well as the focality of Necessity and Intentionality both in the deontic and in the epistemic domains: Figure 16. Semantic map of aviri a. 2.4. Vuliri + INF. The features Intentionality and Necessity play a crucial role also in the modal values of vuliri ‘want’. As a modal verb, it covers a range of meanings from necessity to prediction but, differently from aviri a, it maintains its original lexical value. Specifically, as a content verb, vuliri conveys a transitive volitional meaning, i.e., the desire (an internal condition) of an agent (tr) towards an object, which constitutes its lm. The schematic import of volition (cf. Heine 1993: 31) can be represented as in Figure 17. This configuration can be easily transferred to an activity or an action, which will be a projected (non-actual) one, due to the directional nature of a transitive event (Voli manciari ‘He wants to eat’). This shift marks the transition towards the (non-epistemic) modal domain and is, in fact, crosslinguistically frequent (Bybee et al. 1994: 178). The schematic (i.e., basic and general) import of volition is broad enough to allow the selection of different implications or 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 108 aspects of the overall meaning, depending on the context. Following Heine (1995), we claim that the most significant contextual coordinate which triggers the specific modal readings is agentivity, understood as a prototypical, hence, scalar category (cf. Hopper and Thompson 1980). Figure 17. Schematic import of vuliri. Modal volition expresses an internal and self-oriented condition of an intentional and agentive participant. The subjectification of such meaning depends on the degree of agentivity of the tr: a high degree of agentivity triggers intentional values, while a low degree triggers necessity values and fades intentional ones (cf. Brucale and Mocciaro 2009). Thus, in contrast with aviri a, Intentionality and Necessity are mutually exclusive and give rise to different meanings. This also explains the extension of the volitional meaning to inanimate subjects. This extension represents a metaphorical shift, namely a person to object one (cf. Heine and Claudi 1986), which also involves a metonymical constraint on the grammatical person – the 3rd , i.e., the non-person (cf. Benveniste 1946 [1966]) – and the selection of an implication of the overall modal meaning, i.e., Necessity.41 It is worth noting that this extension is not exclusive of modal use, rather it also concerns transitive situations in which a 3rd person inanimate subject is metaphorically endowed with will. Leaving aside a few contexts in which the subject is a clear case of metonymy,42 we now focus on some instances in which a proper inanimate subject is involved. In (11), the selection of an inanimate subject 41 The relationship between volition and necessity is anything but surprising and is well testified in the diachrony of different languages. Cf. Engl. want, which derives from an Old Norse verb meaning ‘to lack, miss’, from which it developed the sense of ‘need’ (cf. Bybee et al. 1994: 178). 42 E.g., “the Law”, “the Legend”, etc. which embody shared human values or beliefs and, hence, do not change the overall conditions of agentivity, just pushing the interpretation towards an authoritative imposition/permission or cultural belief. 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 109 determines a necessive interpretation even outside the proper modal construction, that is when vuliri is used as a mere transitive verb; this also applies to (12), where, however, voli can be considered elliptical of both a noun (i.e., time, cooking) and an infinitive form (i.e., cociri): (11) U pummaroru bbonu voli assai luci. the.M tomato good want:PRS.3SG much light ‘A good tomato needs plenty of light.’ (12) Com’ è a pasta? how be:PRS.3SG the pasta Ancora voli. still want:PRS.3SG ‘How’s pasta (Is it ready)? It needs more time.’ This metaphorical-metonymical shift, thus, also gave rise to the deontic value of vuliri with inanimate subjects. Examples in (13) and (14) show that this deontic value crosses the domain of the deontic aviri a, specifically in the instances in which the inanimacy of the subject reduces the room for an obligational reading (cf. ex. 3): (13) U pummaroru bbonu voli stari o’ suli. the.M tomato good want:PRS.3SG much at.the light ‘A good tomato needs to stay in the sun.’ (14) A pasta voli cociri (ppi gghiessiri bbona). the pasta want:PRS.3SG cook:INF for be:INF good ‘The pasta needs to cook (to be ready).’ Also in these cases, Intentionality is faded and only metaphorically attributed to the subject. Since both intentional and obligational values are barred, Necessity is the focal sense of the modal use. The reoriented relation of potency is no longer internally agentive nor externally speaker-dependent, but originates from a compelling, yet generic and less prom- 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 110 inent force (a physical, natural condition). In other words, the locus of potency is defocused by the ‘want’-metaphor and represented as a sort of inherent necessity of the S.43 Sometimes, vuliri + INF is used to attenuate the force of an obligation. In this case, agentivity is implied, although it can be defocused through an impersonal si-construction. In (15), the agentive responsibility is in fact addressed to the second speech act participant, as the speaker indirectly exhorts the hearer to perform an action: (15) A quarara si voli inchiri. the.F pot IMPS want:PRS.3SG fill:INF (S’ a (a) gghinchiri.) IMPS have:PRS.3SG (to) fill:INF ‘The pot needs to / must be filled” (=you have to fill it up).’ In other words, even if compelling necessity is the most probable reading when the subject is inanimate, directive obligation and necessity are linked by a relation of continuity as they belong to different but contiguous domains. In (15), the sentence in brackets shows the overlap with aviri a also in this secondary obligational reading. However, differently from aviri a, the presence of agentive subjects does not produce an unambiguous obligational reading, although it re-establishes the inherent intentional value. Volition may shift to a mental aptitude, namely an ability, as in (16), which reports an internal enabling force acting on the subject (Bybee et al. 1994: 177). This meaning, however, appears to be accessible in particular non-realis conditions, such as in negative sen- tences.44 Moreover, it is contextually motivated by the presence of a mental state predicate, while a proper volitional reading is selected with other predicates. 43 Note, in fact, that this interpretation is more easily accessible if a gnomic situation is represented, in which a general (non-individual) necessity is involved, as in (14). 44 This constraint is consistent with Heine’s (1995) claim, according to which a deontic meaning is more easily understood if the modal occurs in a negative or interrogative sentence – i.e., non-realis conditions. 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 111 (16) N’ o voli capìri. not it.ACC.SG want:PRS.3SG understand:INF ‘He doesn’t want to get it / he can’t understand it.’ On the other hand, the ability reading may acquire an obligational nuance in another nonrealis condition, namely in interrogative sentences. As in the case of aviri a (see, in particular, ex. 1), the periphrasis here conveys a directive sense, which is addressed to the second speech act participant encoded as the subject, who is asked to intentionally realise a condition imposed by the speaker. This circumstance is exemplified in (17): (17) Ti vo stari mutu? REFL.2SG want:PRS.2SG stay:INF silent ‘Would you be silent (→ Could you stay silent)?’ Ability can involve inanimate subjects, as in (18), where the periphrasis expresses a merely metaphorical resistance to a human will, as the tr is not endowed with intentionality: (18) A macchina non mi voli partiri. the.F car NEG to.me want:PRS.3SG start:INF ‘The car won’t start (for me) / It’s not starting.’ As for the epistemic domain, vuliri is not involved in the expression of a potential reality concerning the present, rather it is linked to the future-like domain, i.e., to the expression of the prediction or projected reality, both with animate and inanimate subjects. This value does not originate from a deontic-to-epistemic shift and rather derives from an extension of the feature of Intentionality; in other words, it represents a direct implication of the schematic import of volition. This is due to the co-reference between subject and speaker in (19), while in (20), where we find a 3rd person subject, the speaker’s subjective evaluation is represented as an objective one, i.e., as the Tr’s objective volition (cf. examples 8 and 9 for aviri a). In other words, in (20), the speaker expresses confidence about the next realisation of an event and, hence, the construction also conveys an epistemic nuance about a 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 112 future event. Differently from aviri a (see example 9), the predictive vuliri + INF does not seem to be sensitive to the telicity of the verb, as (19) and (20) show: (19) Vogghiu iri a ffari a spisa rumani. want:PRS.1SG go:INF to do:INF the.F shopping tomorrow ‘I’m going to the shops tomorrow.’ (20) Voli iri a curriri rumani. want:PRS.3SG go:INF to run:INF tomorrow ‘He is going to jog tomorrow.’ Vuliri is also employed with meteorological predicates, which – as already pointed out (see example 6) – imply a less prototypical agentivity, i.e., natural forces. In this case, Intentionality is not completely faded, rather it behaves as a dynamic force, i.e., a sort of natural volition metaphorically attributed to the event and represented as an objective feature. This metaphorical volition acts as an external necessity, affecting the degree of certainty of the speaker. Thus, (21) unequivocally expresses a prediction, based on the speaker’s awareness of external (physical, natural) conditions, and conveying a future-like value: (21) È niuru, voli chioviri. be:PRS.3SG black want:PRS.3SG rain:INF ‘It’s black outside: it’s going to rain.’ In conclusion, differently from aviri a, the epistemic and future-like values are more context-dependent and the modal vuliri is, on the whole, strongly rooted in the non-epistemic domain. Since the meanings associated to this domain are more basic and concrete than the epistemic ones, we claim that vuliri is less desemanticised than aviri a and, as it maintains its lexical value, shows a higher degree of “verbness” (Heine 1995). The semantic range of vuliri is represented in Figure 18. The map suggests the convergence of vuliri and aviri a in the expression of the semantic cores, i.e., necessity and prediction/future. These semantic areas, however, are linked to non-identical features: 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 113 Figure 18. Semantic map of vuliri. 1. In aviri a, Necessity and Intentionality are focal senses in both modal domains, while in vuliri they are mutually exclusive: Necessity is active in the deontic area only, whereas the other semantic areas depend on Intentionality. 2. The non-necessive meanings arise from different paths: in aviri a, the value of pre diction/future mainly originates from a deontic-to-epistemic shift, while in vuliri, from an extension of the basic feature of Intentionality. Moreover, vuliri lacks both a proper obligational value and the potential-epistemic one. The analysis carried out thus far brings to the light a difference in the grammaticalisation status of the two modals: A. Aviri a + INF is a fully grammaticalised construction, which shows a high degree of versatility, both in the non-epistemic and in the more abstract epistemic domain. The deontic-to-epistemic readings depend on the contextual coordinates, namely on the actionality of the verb and the agentivity of the Tr, therefore on the 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 114 feature of Intentionality. In other words, Intentionality variously interacts with the modal feature of Necessity within both modal domains. B. Vuliri exhibits a different sensitivity to Intentionality. The basic “verbness”, that is, the lexical semantics remains untouched and only when Intentionality is faded (or metaphorically represented), vuliri conveys a non-volitional, namely deontic modal meaning, where the notion of Necessity is in action. This feature does not develop other stable potential-epistemic meanings nor prediction ones, the latter represent ing a temporary inferential extension of Intentionality. Interestingly, also in the latter case, a proper modal (epistemic-like) meaning, involving the degree of certainty of the speaker, is elicited by a reduced degree of Intentionality. 2.5. Modal meanings and temporal-aspectual features. The different degree of grammaticalisation of aviri a and vuliri is further shown by the influence that temporal inflection wields over modal meanings. According to Langacker (1991), for modals, past inflection pertains to the degree of likelihood rather than to a simple temporal location. As Figure 19 shows, distal morphemes move away the event from the locus of the speech-act (G) and locate it in a non-immediate reality, both known (past meaning) and unknown (potential), or in the non-reality. As for Sicilian, however, this analysis needs further observations on the aspectual features of past tenses. The imperfective past correlates of the examples thus far proposed for aviri a convey a counterfactual reading when the sentence implies forward-push dynamics towards a not yet realised event, i.e., in the cases of deontic or prediction meanings. This is the case in (22) and (23): (22) T’ avi(ev)’ a rregalari un palluni. you.DAT have:PST.IPFV.1SG to give:INF a ball ‘I should have given you a ball (but I didn’t).’ 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 115 Figure 19. Epistemic model (Langacker 1991: 243-244, adapted). Example (22) implies a ‘but I didn’t’ implication, thus expressing a non-realised possibility and maintaining a projective/future-like orientation (a future possibility in the past, cf. Bertinetto 1986, 1991; Condoravdi 2002). An analogous consideration arises from the analysis of the imperfective past of vuliri, although in this case also a non-counterfactual reading is accessible both in the necessity and in the prediction domain, depending on the context, as in (23): (23) A pasta vuliva cociri: è crura. the pasta want:PST.IPFV.3SG cook:INF be:PRS.3SG raw ‘The pasta needed (to take more time) to cook: it’s still raw.’ The epistemic shifts conveyed by imperfetto depend on the aspectual nature of this tense, irrespective of the presence of a modal: imperfetto is indefinite with respect to the actual completion of the event, thus easily shifts towards virtuality, i.e., towards epistemic values (cf. Bertinetto 1986, 1991). On the other hand, the perfective past does not naturally con- 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 116 vey a modal value and generally denotes the completeness of a past event. Only when the distal morphemes interact with a modal verb, the perfective past expresses modal nonimmediacy, specifically an epistemic meaning concerning the degree of likelihood of the event. This circumstance is crucial in our analysis of the modal status of aviri a and vuliri, which manifest in fact a different behaviour in perfective past tenses: (24) T’ app’ a rregalari un palluni. you:DAT have:PST.PFV.3SG to give:INF a ball ‘S(he) most likely gave you a ball/S(he) just had to give you a ball.’ (25) Assai vosi cociri a pasta. much want:PST.PFV.3SG cook:INF the pasta ‘The pasta needed a lot of cooking.’ (26) St’ annu nun vosi chioviri this year not want:PST.PFV.3SG rain:INF mancu na vota. not.even one time ‘This year it didn’t rain even once.’ In the simple perfective past (Sicilian passato remoto), aviri a sometimes receives the same deontic reading as in the present tense, but it more likely conveys an epistemic interpretation (‘It’s very likely that...’), as in (24). Moreover, the future-orientation reading is entirely barred, since the utterance is set in the dimension of completeness. On the other hand, perfective past inflection does not wield any epistemic influence on the modal value of vuliri, neither in the necessity meaning, as in (25) (which maintains the same deontic value of the present, although set in the past), nor in the meanings based on Intentionality, as in (26) (which furthermore loses both the prediction value and the epistemic nuance involving the degree of certainty about a meteorological event). In a word, these examples convey a temporal meaning only. To sum up, the effects produced by the temporal inflection validate our analysis in terms of different degrees of grammaticalisation. Apparently, the epistemic values of vuliri are not only peripheral but also less steady and strictly depending on the modal features of 2 FIRST PATH. FROM POSSESSION AND VOLITION TO MODAL MEANINGS. 117 the imperfective inflection (both in the present and in the imperfective past). On the other hand, aviri a is a stable means of modal expression and the complex interaction of temporal, aspectual, actional and, finally, transitivity features does not alter but further elaborates the modal status of the construction, thus strongly suggesting the multi-factorial nature of the category of modality. 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 118 3 Second path. From volition to modal-passive meanings. “Vuliri + PP in the data of the Atlante Linguistico della Sicilia”, 2016.45 3.1 Introduction. The article aims at describing some contemporary Sicilian periphrases in which an inflected form of vuliri ‘want’ is constructed with the past participle (PP) of a lexical verb, as in (1)-(3): (1) U scaluni è lordu: voli essiri puliziatu. the.M step.M be.PRS 3SG dirty.M want:3SG be.INF clean:PST.PTCP.M ‘The step is dirty: it must be cleaned.’ (2) U picciriddu voli essiri accattata na cosa duci. the.M child.M want:PRS.3SG be.INF buy: PST.PTCP.M a.F thing.F sweet ‘The child wants to be bought something sweet.’ (3) U picciriddu voli accattata na cosa duci. the.M child.M want:PRS.3SG buy: PST.PTCP.M a.F thing.F sweet ‘The child wants to be bought something sweet.’ Although similar, the types in (1)-(3) differ in semantic-functional and morphosyntactic terms: 45 Originally published as: Luisa Amenta and Egle Mocciaro (2016), Vuliri + PP nei dati dell’Atlante linguistico della Sicilia. In Éva Buchi, Jean-Paul Chauveau, Jean-Marie Pierrel (éd.), Actes du XVIIe Congrès international de linguistique et de philologie romanes (Nancy, 15-20 juillet 2013), Vol. 1, 933–944. Strasbourg: ÉLiPhi. As reported in the source paper, “[t]he article is the result of close collaboration between the authors; however, for academic purposes, Luisa Amenta is responsible for 1 and 3 [here, 3.1 and 3.3], and Egle Mocciaro for 2 and 4 [here, 3.2 and 3.4].” 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 119 − In (1), vuliri, a basically volitional verb, conveys deontic meaning, absent in the other cases. − Both (1) and (2) are formally passive structures, whose subjects (u scaluni, u picciriddu) agree with the past participle (PP) of the verb; however, they differ seman tically, because only in (1) the subject embodies the role of patient (the object of the transitive PP), while in (2) it maintains the function of first (animate) argument of the volitional verb (therefore the role of experiencer that corresponds to the potential beneficiary of the event denoted by the PP). − Example (1) allows for inanimate subjects, while in (2) and (3), where the volition al value is intact, the subject is characterised by the trait of animacy. Since animacy is an implication of volitionality, the presence of inanimate subjects in type (1) rep resents an extension related to the shift of the function of vuliri from volition to deontic necessity and suggests a process of grammaticalisation within the domain of modality. − (2) and (3) are similar on the semantic-functional level, but differ on the formal level because of the presence of the passive auxiliary essiri ‘to be’ in (2). − (2) and (3) also differ in diaphasic terms because volere + PP is also possible in written Italian of high (literary and non-literary) register.46 In the light of these observations, we conducted a systematic analysis of three constructions, as they occur in the data of the socio-variational section of the Atlante Linguistico della Sicilia (‘Linguistic Atlas of Sicily’, henceforth ALS).47 In particular, our corpus consists of: a) the answers to the questions of the morphosyntactic section of the questionnaire, containing two groups of translation questions: Italian to Sicilian and Sicilian to Italian; b) the examples produced by the informants during the socio-variational interviews and in the ethno-dialectal section, which allow us to observe freer contexts than the input- 46 Cf. Salvi (1988) and Serianni (1988, Section 65). It is possible that the construction is acceptable to southern speakers because it is supported by the Sicilian periphrasis. 47 On the objectives of the Atlante Linguistico della Sicilia and the characteristics of the socio-variational surveys, cf. D’Agostino and Ruffino (2005, 2013), Ruffino (1995) and see Section 1.2.2.3. 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 120 based translations. The sample of informants is divided into family types differentiated by age, first language (Italian or Sicilian) and education background. The analysis, conducted against the existing reference literature (Section 3.2), focused on the following aspects: 1. Areal diffusion and productivity in contemporary Sicilian and Italian spoken in Sicily (regional Italian) according to a variationist approach; in particular, we tried to verify: a) the role of sociolinguistic variables in the degree of productivity of the three types, also with reference to different textual types; b) whether the occurrence es of the phenomenon can be found in different areas (Section 3.3). 2. Relative frequency of the three types and their semantic and formal characteristics (animacy of the subject; temporal selection; PP agreement; syntactic contexts of occurrence) (Section 3.3). 3. Degree of grammaticalisation of the three types (Section 3.4). 3.2 The state of the art. The reference literature offers important reports on the constructions with ‘want’ in various dialects of Italy. As far as Sicilian is concerned, although the construction is still vital and transits also in regional Italian, little space is devoted to it in studies dealing with the morphosyntax of the dialect or regional varieties. Based on De Mauro (1963: 403), Telmon (1993: 126) discussed the “infinitive construction of the completive after ‘verba voluntatis’ as generically southern”, with a possible epicentre in Campania and Sicily. However, most of the specific contributions on this topic base their analysis mainly on data from Neapolitan and other southern varieties, giving Sicilian only partial attention. Therefore, the construction of Sicilian remains substantially unexplored as regards its morphosyntactic and semantic characteristics and its geographical distribution. Nevertheless, the construction has not failed to attract the attention of scholars. Rohlfs (1969: 131) described type (3) as the southern extension of an independent proposition with conjunction deletion (voglio fatta giustizia ‘I want justice done’ < voglio che sia fatta 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 121 giustizia ‘I want justice to be done [literally: I want that justice is done]’) and gave examples from Pitrè’s (1875) collection of tales and short stories (voli tinciuta sta tila ‘he wants this cloth (to be) painted’). Rohlfs pointed out that the construction is common to other southern varieties, but while in the case of Sicilian “the participle is still grammatically linked to the object, in Calabria and Salento it is strictly referred to the subject” (cf. Calabrian vue pagatu ‘you want/have to be paid’, a cammisa vole lavata ‘the shirt wants/needs to be washed’). Leone (1995: 42) analysed type (2) as an implicit proposition with a subject different from the main one and considered the construction as an example of passive, which would compensate for the rarity in Sicilian of the canonical passive essere ‘to be’ + PP (e.g., Maria vosi èssiri ccattata a casa ‘Maria wanted them to buy the house [literally: Maria wanted to be bought the house]’). Similar considerations are in Varvaro (1988: 725), who described sentences of the type (3) as passive examples (vosi dittu tuttu ‘he wanted to be told everything [literally: he wanted told everything]’; vogghiu mannatu un paccu ‘I want them to send me a parcel [literally: I want sent a parcel]’). The unpopularity of the analytical passive essere + PP has been indicated by Cennamo (1997: 146-153) as one of the most salient features of Southern dialects. In particular, Sicilian prefers active sentences with left dislocations with pronominal resumptions or the periphrasis avere avuto ‘to have had’ + PP.48 Based on existing studies, Cennamo (1997: 151) also reported the deontic uses of ‘want’ (type 1) and the passive or ‘patient-oriented’ uses (types 2 and 3) for several Southern dialects, highlighting their main aspects (agent defocusing, topicalisation and subjectification of a non-agent). In particular on modal use, an important contribution is in Loporcaro (1988), who analysed the deontic construction of ‘want’ in the Altamura dialect (lu pesce ulia mangiatu 48 Cf. the case cited by Varvaro (1988: 725), Pippinu, comu lu vittiru, u chiamaru ‘Peppino, as soon as they saw him, they called him’. 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 122 stammane ‘the fish had to be eaten this morning [literally: the fish wanted eaten this morning]’; this type, however, diverges from type (1) because it lacks the auxiliary)49 . An extensive analysis of ‘want’-passives, conducted within the minimalist framework, is in Ledgeway (2000), who examined mainly data from Neapolitan and Cosentine and identified three types of construction, largely corresponding to the possibilities of Sicilian. According to Ledgeway (2000: 236), these three types are different in terms of semantics and morphosyntax, but all of them are “complex complementation structures in which the embedded participial clause represents a non-finite canonical passive”. Ledgeway (2000: 244) observes that, unlike in other southern passive structures, the ‘want’passive can have an animate argument – technically an experiencer – in subject position. This subject can easily correspond to the dative subject – technically a beneficiary or a recipient – of ditransitive participles. The deontic type in (1), on the other hand, seems to impose no restrictions on the trait of animacy and thus admits inanimate subjects. Bentley (2003) proposed an analysis of type (1) in some southern dialects against the theoretical background of the Role and Reference Grammar. She considered deontic modality as a subgroup of the “modalité extérieure au participant”, a type of possibility or necessity that depends on circumstances external to the participant (see also van der Auwera and Plungian 1998). In contrast to Ledgeway (2000), Bentley (2003: 63) noted that the meanings ‘want’ and ‘must, have to’ are not equipollent: while the latter has an identical active parallel on the modal plane (the letter must be written/John must write the letter), ‘want’ does not (the letter wants to be written [modal]/John wants to write the letter [volitional]). Therefore, in the case of ‘want’, the correspondence is not derivational in nature. The modal construction of ‘want’ consists of a “syntactic knot” between an active predicate (‘want’) and a passive one (‘write’), which have a common argument (‘the letter’). This argument is the semantically affected object of the PP predicate, which is encoded as the subject of ‘want’ (the first argument of ‘write’ is instead defocused, as is 49 But see the examples reported in Loporcaro et al. (2018: 294), about the presence of the past participle in the variety of Pantelleria: “However, the past participle occurs in volitional and deontic periphrases formed with vuˈliːrɪ ‘want’: e.g., ‘vɔɟːʊ kanˈtaːta sta kanˈtsuːni ‘I want this song to be sung to me’; sta ˈpinːa ˈvɔːli jiˈtːata aa muˈnːitːsa ‘this pen should be thrown in the bin.” 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 123 normally the case in passive constructions). The other subject of ‘want’ is the predication containing ‘write’, which is thus subordinated (“nidifiée”) to the first, although this subordination is not marked canonically, but as a co-subordination, with a common argument between the two predicates. In the light of the existing reference literature, in the following paragraphs we will provide a description of the three types of ‘want’-constructions. This description will allow us to propose some hypotheses on their state of grammaticalisation. In the conviction that grammatical forms are strictly connected to the concrete use that speakers make of them, the following paragraph will also take into account the vitality of vuliri + PP in contemporary Sicilian and the sociolinguistic variables that seem to influence their selection. 3.3 Data analysis. 3.3.1 Restrictions on use. The analysis of the ALS data allows us to draw a picture of the current vitality of the different types of construction and the characteristics they exhibit in informants’ productions. To start with, the ‘want’-constructions especially occur in ethnotexts, in which the speakers report on food practices or festivals in their countries, and only sparingly in the interviews in Sicilian, where the informant is asked to recount an episode from her own life. This suggests that, compared to the numerous occurrences in Pitrè’s texts referred to by Rohlfs (1969: 131), the ‘want’-types under consideration are no longer among the expressive potentialities of informants, at least in the narrative textual genre exemplified in the interviews. In this genre, the periphrasis aviri a ‘have to’ + infinitive is preferred for the expression of the deontic modality (me soru ruminca av’a gghiri a Montemaggiore ‘my sister must go to Montemaggiore on Sunday’)50 and vuliri + infinitive for volitional sentences (io un vògghiu bballari ‘I do not want to dance’). This is also the case in the 50 A highly productive form that has also spread into the functional domain of vuliri + PP because it allows both the expression of the deontic modality and the future tense (cf. Amenta 2004, 2010a; Brucale and Mocciaro 2009). 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 124 productions of speakers with low education, for whom we can assume a good competence of Sicilian. Therefore, vuliri + PP seems to have narrowed its range of use and remains confined only to certain contexts, tending to fossilise in specific groups of speakers and/or with particular predicates. A case in point is represented by the ethnotexts related to food practices, where we find occurrences of constructions with vuliri that are produced by elderly dialect speakers with a low level of education, mostly the elementary diploma. This suggests that vuliri + PP of the type (1) can be linked to the prescriptive textual type, where it is used in the description of food preparation procedures or ritual festivities. From the diatopic point of view, no relevant local differences could be observed in the distribution of the constructions, which can therefore be considered pan-Sicilian. 3.3.2 Formal and semantic properties. The corpus shows occurrences of types (1) and, above all, (3), here exemplified respectively in (4) and (5). Type (2) does not occur, except in the case in (6), where, however, the interpretation of ‘St. Joseph’ as a volitional animate subject is questionable, since it more likely refers metonymically to the celebration and, therefore, the periphrasis shifts from the volitional to the deontic modality: (4) U pummaroru voli èssiri accutturatu. the.M.SG tomato:M.SG want:PRS.1SG be:INF cook:PST.PTCP.M.SG ‘You must/have to do as I say.’ (5) Pigghi la farina, la ‘mpasti take:PRS.2SG the.F.SG flour:F.SG it.ACC.F.SG knead:PRS.2sg ci metti ‘n pocu di lèvitu – a CLIT.LOC put:PRS.2sg a.bit.of yeast the.F.SG pasta voli èssiri fatta dough want:PRS.3SG be:INF make:PST.PTCP.F.SG 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 125 né modda né ddura. neither soft nor hard ‘Take the flour, knead it, put a bit of yeast in it – the dough must be made neither soft nor hard.’ [literally: wants be made] (Maria, Gibellina, TP, 70 years old, elementary school diploma) (6) Quindi si facìa la festa ad so IMPS do:PST.IPFV.3SG the.F.SG festival:F.SG at agosto non perché era san Giuseppe August not because be:PST.IPFV.3SG saint Giuseppe ma picchì puru san Giuseppe but because also saint Giuseppe vulìa èssiri festeggiatu. want:PST.IPFV.3SG be:INF celebrate:PST.PTCP.M.SG ‘So the festival was celebrated in August not because it was St. Joseph, but because St. Joseph also had to be celebrated.’ [literally: wanted be celebrated] (Gregorio, Menfi, AG, 50 years old, high school diploma) In all the examples, the object agrees with PP, occupies a postverbal position and the preverbal position occurs only when the object is a personal pronoun. To continue with the level of morphosyntax, the construction can occur in dependence on existential or presentative sentences (c’è ‘there is’, c’era ‘there was’), as in (7) and (8), both with a pronominal subject (cu/chi ‘who’): (7) C’era cu ‘n ci vulìa EXS.PST.IPFV.3SG who not CLIT.LOC want:PST.IPFV.3SG misu u tumazzu, cu put:PST.PTCP.M.SG the.M.SG cheese.M.SG who ‘n ci vulìa a cipudda. not CLIT.LOC want:PST.IPFV.3SG the.F.SG onion:F.SG ‘There were those who did not want cheese added, those who did not want onion added.’ [literally: who wanted put] (Maria, Alcamo, TP, 80 years old, high school diploma) (8) C’era chi ci voli spaimmatu EXS.PST.IPFV.3SG who CLIT.LOC want:PRS.3SG smear:PST.PTCP.M.SG 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 126 bbellu paru paru u pumidoro. INT entirely the.M.SG tomato:M.SG ‘Some wanted the tomato to be smeared entirely.’ [literally: who wanted smeared] (Benedetta, Villagrazia di Carini, PA, 74 years old, elementary school diploma) In (7) the second member of the sentence (cu un ci vulia a cipudda ‘who did not want onion added’) is elliptical of the PP, maybe due to the symmetry of the two coordinated sentences. In (8), although agreed to the PP, the object is separated from it by the interposition of other lexical material (bbellu paru paru ‘entirely’). In (9), the construction is in a sentence with right dislocation of the object, anticipated by the pronoun a in preverbal position. The presence of the dislocation does not entail any modification in the PP agreement: (9) Si tàgghia nna lu menzu e ppo IMPS cut:PRS.3SG in the middle and then tu la conzi comu vo, comu you it.ACC.F.SG season:PRS.3SG as want:PRS.2SG as a vo cunzata ti la it.ACC.F.SG want:PRS.2SG season:PST.PTCP.F.SG REFL it.ACC.F.SG conzi la focaccia. season:PRS.2SG the.F.SG focaccia:F.SG ‘You cut it in the middle and then you season it as you wish, season the focaccia as you wish it to be seasoned.’ [literally: you want seasoned] (Rosa, Lercara, PA, 88 years old, high school diploma) In (10), the construction occurs in a subordinate introduced by ca ‘in that, because’, where the second person subject refers to the antecedent animate subject, corresponding to the indirect object (beneficiary) of the participial predicate fatti ‘made, prepared’: (10) Me patri era vizziusu ca my:M.SG father be:PST.IPFV.3SG pampered in.that vulìa fatti i scacciati. want:PST.IPFV.3SG make:PST.PTCP.PL the.PL scacciati.PL ‘My father was a gourmand and he wanted them to make him the scacciate.’ [literally: he wanted made] (Giuseppa, Resuttano, AG, 72 years old, high school diploma) 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 127 As regards the temporal selection, vuliri is often in the present tense, but the selection of the imperfect is also frequent and focuses on an open time span in which what is expressed by the construction can occur, as in (11) and (12): (11) Vulìa fattu ddu favuri ma want:PST.IPFV.1SG make:PST.PTCP.F.SG that.M.SG favour but iddu un ci vosi sèntiri. he not CLIT want:PST.PFV.3SG listen ‘I wanted him to do me that favour, but he wouldn’t listen to reason.’ [literally: I wanted made] (Maria, PA, 59 years old, junior high school diploma) (12) Tutti nzèmmula ci vulìanu all together us.ACC want:PST.IPFV.3PL misi ddocu ma nui make:PST.PTCP.M.PL there but we un ci misimu. NEG REFL put:PST.PFV.1PL ‘They wanted to put us all together there, but we didn’t.’ (Anna, Menfi, AG, 69 years old, elementary school diploma) 3.3.3 From Sicilian to Italian. As regard regional Italian, we analysed the translations from Sicilian to Italian and the interviews in Italian. The translational questions contain an example of the type (3), A vo accattata na cosa duci? ‘Do you want me to buy you a cake? [literally: want bought]’, with an unexpressed animate subject and right dislocation of the object with clitic climbing (A ‘it’, coreferential with cosa duci [literally: ‘something sweet’]). The Sicilian verb accattari ‘to buy’ is frequently selected in the construction with the verb vuliri, as in (13): (13) Io per esempio vìu picciotti â I for example see:PRS.1SG boy:PL at.the šcuola ogni annu vonnu accattatu school every year want:PRS.3PL buy:PST.PTCP.M.SG 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 128 u zzàinu novu. the.M.SG schoolbag new ‘I, for example, see the boys who, every year at school, want to be bought a new schoolbag.’ [literally: want bought] (elderly speaker, male, Bagheria, PA, L151 Sicilian, junior high school diploma) As Ledgeway (2000) observed, passive-volitional constructions are often connected to ditransitive verbs, as accattari is, whose direct object may or may not agree with the PP,52 while the dative object (semantically, the beneficiary) is encoded as a grammatical subject. The answers given by the informants are reported in (14a), (14b) and (14c): Almost all informants, irrespective of age, level of education and point of enquiry, gave the answers (14a) and (14b). Those who attempted to move away from the input sentence replaced ‘a sweet thing’ with ‘cake’, perceived as more acceptable in Italian (where dolce is both an adjective meaning ‘sweet’ and a noun for ‘cake’), but still used the periphrasis as the preferred answer. The sentence in (14c) is a translation option chosen only by speakers aged 25-50 and the younger generation aged 18-25, with high education and mostly from the central-western area. However, these answers represent sporadic occurrences. On the other hand, informants consider the calque of the input sentence, i.e., the 51 L1 is the abbreviation for language 1, that is, first or primary or native langage. 52 As observed, the agreement between object and PP is the rule in the cases under investigation. However, in other varieties there may also be agreement with the grammatical subject (in this case, *i picciotti vonnu accattati u zzainu). (14a) La vuoi comprata una cosa dolce? It.ACC.F.SG want:PRS.2SG buy:PST.PTCP.F.SG a:F.SG thing:F.SG sweet:SG (14b) Lo vuoi comprato un dolce? It.ACC.M.SG want:PRS.2SG buy:PST.PTCP.M.SG a:M.SG cake:M.SG (14c) Vuoi che ti compri una cosa dolce? want:PRS.2SG that to.you buy:SBJV.1SG a:F.SG thing:F.SG sweet:SG ‘Do you want me to buy you something sweet?’ 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 129 option (a), as an acceptable option and do not hesitate in answering this translation ques- tion. Finally, the corpus shows an example, reported in (15), in which the grammatical subject does not correspond to the indirect object of the PP, but to the direct object of a change of state. However, based on the data at our disposal (as well as our competence as speakers), this construction appears rather atypical (although widely attested in other southern varieties), as it would imply, as a rule, the presence of the infinitive form of the passive auxiliary ‘to be’, which does not occur instead in (15): (15) È mmorto uno, e gli be:PRS.3SG die:PST.PTCP.M.SG one and to.them ha scritto che qquesto voleva have:PRS.3SG write:PST.PTCP that this want:PST.IPFV.3SG seppellito nel commune di bury:PST.PTCP.M.SG in.the municipality of Capo d’Orlando. Capo d’Orlando ‘A guy died and wrote that he wanted to be buried in the town of Capo D’Orlando.’ [literally: he wanted buried] (elderly speaker, male, Capo D’Orlando, ME, L1 Sicilian, junior high school diploma) Since we are dealing with an isolated occurrence, and since there is no evidence in the Sicilian data, we must conclude that the absence of the auxiliary is barely due to a change of planning in the speech of the elderly informant. Only the finding of other cases, possibly elicited by ad hoc questionnaires, would show the actual presence of this type, present in other southern varieties but so far not attested in Sicilian (but in Pantesco, see fn. 49). 3.4 The grammaticalisation of vuliri + PP: some considerations. From the perspective of grammaticalisation theory (see Bybee et al. 1994; Hopper and Traugott 1993 [2003]; Heine 2003, inter al.), the deontic value expressed by vuliri is to be considered as a semantic extension, therefore temporally secondary, of the basic value of agentive intentionality. This extension can be interpreted in metaphorical terms, as a pro- 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 130 jection of the volitional meaning to inanimate subjects (a ‘person-to-object’ metaphor, in terms of Heine and Claudi 1986). The extension implies a partial semantic emptying of the verb, consisting in the loss of the trait of agentive intentionality, which allows its generalisation to non-intentional and non-animate subjects. It also produces a restriction on the grammatical person, since the non-animate subjects of the deontic construction are always in the third person: in fact, the presence of the first and second person would re-establish the intentional value. It is therefore possible to claim that the deontic interpretation of the construction is strongly dependent on the context of occurrence: the deontic value is included within the semantic network of vuliri but is only activated in certain contexts. Some remarks are needed here. First, the link between volition and necessity is far from surprising and, indeed, well attested in the diachrony of several languages.53 Secondly, it should be noted that the deontic value of Sicilian vuliri is not limited to passive constructions only, but also covers other uses (cf. the lexical use in u pummaroru voli luci ‘the tomato needs light [literally: wants light]’ or the construction with inactive predicates in the infinitive form, such as u pummaroru voli stari o suli ‘the tomato must be in the sun [literally: wants to stay]’, discussed in Brucale and Mocciaro 2009). In this perspective, the extension to passive contexts could be read as a simple phenomenon of generalisation: by grammaticalising, an entity tends to widen its range of use. On the other hand, the convergence between deontic necessity and passive meaning54 also produces other effects: the necessity of the event (which, in the case of the passive, is a basically agentive event) is metaphorically represented as an intentionality of the subject, who “wants” to become the object of a given change of state. In other words, although “extérieure au participant”, the deontic necessity is represented as internal to it. The agentive responsibility for the event, normally defocused in the passive, is thus further concealed (and, indeed, syntactically unrecoverable). Whether this circumstance underlies the 53 Cf. Bybee et al. (1994: 178). However, the directionality of the shift in Sicilian deserves further investiga- tion. 54 In Italo-Romance, this is not limited to ‘want’; suffice it to mention the passive modal construction andare ‘go’+ PP, analysed by Giacalone Ramat (2000) as expressing “modulation of deontic force”. However, this construction is formally different from those under investigation and, furthermore, not attested in Sicilian. 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 131 extension of vuliri to passive contexts or is just an after effect is a question that only diachronic investigation can answer. What is relevant here is that, in such contexts, vuliri has a fully modal behaviour. The grammaticalisation of vuliri as a modal verb is suggested not only by the semantic bleaching and the restrictions on the grammatical person, but also by the loss of argument structure: vuliri is indeed devoid of arguments of its own, since its grammatical subject is the affected and unintentional argument of the action denoted by the passive segment ‘be’ + PP. On the other hand, the first argument of this passive segment is the defocused agent. However, this modal value is unstable because strictly anchored to the passive context. At the same time, as Bentley (2003) observed, it lacks an active counterpart (‘the letter wants to be written’ is modally different from ‘someone wants to write the letter’). The defocused agent of the passive is, in fact, typically animate and intentional and, as we have seen, in an active structure, where it is encoded as a subject, volitional interpretation is inevitably restored. The same development in the grammaticalisation process cannot be observed in the volitional passive of types (2) and (3), where vuliri is endowed with its own argument structure: a first intentional argument, that is, the animate subject of vuliri (which maintains its role even when it coincides with the dative argument of the PP), and the object of volition corresponding to the embedded passive clause (PP or ‘be’ + PP). The passive value of these structures thus remains largely dependent on the presence of the passive clauses, which are inherently passive or passive-oriented clauses (i.e., transitive PPs, cf. Haspelmath 1990). The very limited vitality of the passive-deontic type in contemporary Sicilian is not theoretically surprising: since its use is unstable and highly context-dependent, we can hypothesise a phenomenon of retraction, a notion introduced by Haspelmath (2004) to describe an arrest of the grammaticalisation process of a construction, which drastically reduces its range of use; in the case under investigation, this could also be due to the competition of aviri a ‘have to’ + infinitive. Even more interestingly, the contraction actually affects all the ‘want’-passive constructions, which, as already noted, seem to be confined to specific textual types. In addi- 3 SECOND PATH. FROM VOLITION TO MODAL-PASSIVE MEANINGS. 132 tion, these constructions reflect the linguistic competence of elderly and only Sicilian speakers with a limited educational background, suggesting that their usage may have “frozen” and in fact persist as a fossil of the language. As a matter of fact, it does not appear in the more innovative usages of young people, largely oriented towards Italian, e.g., the passive construction venire ‘come’ + PP or the aforementioned aviri a + infinitive, generally considered to be a strongly characteristic feature of Sicilian. Precisely because it is a fossil, the ‘want’-construction transits into regional Italian, where it is accepted and proposed by the speakers, irrespective of their age and level of education, as a potential translation from Sicilian. However, even in regional Italian, it is characterised by a limited range of use, since it occurs essentially with ditransitive predicates while retaining both volitional and modal values. This first survey of the passive uses of vuliri leaves open many questions, indicating as many directions for further investigation. First of all, the direction of change must be ascertained, that is, whether the volitional type constitutes the basis on which the deontic type has developed or whether, rather, the latter is based on the deontic values conveyed by vuliri even outside the domain of the passive. Secondly, we will need to evaluate the times, contexts and possible areas of irradiation of these new values, through the examination of the data available for medieval and modern Sicilian. This is what we will try to verify in the near future. 4 THIRD PATH. FROM VOLITION TO DIRECTIVE MEANINGS. 173 On the level of sociolinguistic variation, the richness and variety of the corpus, which has various textual genres and various diastratic and diaphasic connotations, makes it possible to formulate hypotheses about the intertwining of sociolinguistic and textual variables and investigated phenomena (for instance, in the case of directive construction, particularly frequent in texts of a regulatory type, such as monastic rules, in which, according to the principles of obedience, the monk takes charge of the will of another in the accomplishment of the precepts). Other meanings, of the futural type, perhaps lie on the same sociolinguistic dimension, notably on the diamesic specificity of the corpus, or otherwise on that of temporal development. These meanings, whose occurrences have not been traced in the concordance of the Sicilian School, are, on the other hand, frequent in the Sicilian of later centuries, especially in the dimension of orality:77 the diachrony of these forms deserves further investigation, even by relating constructions with vuliri to other constructions that are particularly vital in contemporary Sicilian. 77 The reference to the diamesic dimension could also help to explain the absence of other constructions that, in later centuries, will involve vuliri, in particular the passive construction ‘vuliri + past participle’ (u picciriddu voli (essiri) accattata na cosa duci ‘the child wants you to buy a cake’ , cf. Amenta and Mocciaro 2016), which, based on a preliminary survey, seems to emerge only in the 19th century. 278 COMMENTARY TO HABILITATION THESIS1 I present five previously published articles, written either as an individual scholar or in collaboration with two other colleagues. I have framed the articles in a theoretical discourse comprising Chapter 1 “Framing the study of modality in Sicilian: an introduction” (pp. 13-94, where I present the corpora on which the analyses proposed are based, the existing literature, the theoretical background adopted, an overview of the articles and how they link to each other in a coherent manner) and Chapter 7 “Towards a unified account: a modality’s semantic map for Sicilian” (pp. 241-248, where I summarise the content of the articles, highlighting points of convergence and also reviewing some theoretical points and conclusions, including an original semantic map). Chapters 2 to 6 contain the articles presented. They fully reflect the originals in their content and structure. However, for the sake of consistency, they are presented in English, with minor adaptations from the Italian originals of most of them, which mainly facilitate the linguistic transition. In addition, to facilitate understanding by non-Sicilian readers, interlinear glosses are consistently provided for all examples in ancient and contemporary Sicilian (i.e., the second grey line between the example in Sicilian and its translation in English, where grammar is explained, as usual in linguistics). Finally, the reference list has been updated where necessary and, in order to avoid repetition and facilitate consultation, it is emancipated from the individual contributions and listed in one place, at the end of the work. Other minor changes compared to the originals are the addition of internal cross-references and the narrowing down of some notes, in particular, those relating to the corpora used, where repeated from one article to the next. Otherwise, everything new in the volume is contained in the theoretical introduction (Ch. 1) and in the final synthesis (Ch. 7). Chapters 1 and 7 are my own original work (Experimental work, Supervision, Manuscript, Research direction: 100%) The five articles collected here, as well as the theoretical chapters that frame them (1, 7), address the issue of modal and directive encoding in ancient and contemporary Sicilian, focusing on some specific grammaticalisation paths and their interactions and overlaps. They are inspired by a cognitivist and functionalist theoretical approach, which is intrinsically usage-based and thus takes as an interpretative key the assumption that grammatical forms and constructions emerge from the communicative needs that speakers manifest in discourse. In this context, a complex system of operations of a metaphorical-metonymic nature is activated, through which fundamental concepts such as movement, possession or volition are reworked at the linguistic level and provide raw material for the construction of grammatical forms, which become increasingly abstract. In each article, the data analysis is based on a broad sampling of linguistic data, which consist, for contemporary Sicilian, in the spontaneous speech collected in the Atlante Linguistico della Sicilia (led by the University of Palermo), and in data extracted from a textual archive of 14th- and 15th-century Sicilian (Archivio Testuale del Siciliano Antico, led by the University of Catania). The reference to the original articles is made explicit in the Abstract preceding the work, the Preface, the last section of Chapter 1 and at the beginning of each chapter, by means of a footnote where all information is reported in detail. The original articles, whose bibliographic records are listed below, are collected in the Appendix that is introduced by the present Commentary and constitutes the second volume of the Habilitation thesis. 1 The commentary must correspond to standard expectations in the field and must include a brief characteristic of the investigated matter, objectives of the work, employed methodologies, obtained results and, in case of coauthored works, a passage characterising the applicant’s contribution in terms of both quality and content. 279 [1 ]2 E. Mocciaro (& L. Brucale), 2019. Possession and volition in the development of modal meanings: a case-study from Sicilian. Journal of Contemporary Philology 2(2): 27-47. I am the author of Sections 2 and 4. (This article, reported in the volume, is a revised version of Brucale & Mocciaro 2009, see record 2.) Experimental work (%) Supervision (%) Manuscript (%) Research direction (%) 50% 50% 50% 50% [2] E. Mocciaro (with L. Brucale), 2009. Polisemia e convergenze nel dominio dei modali in siciliano: una lettura funzional-cognitivista di vuliri e aviri a. L. Amenta, G. Paternostro (a c. di), I parlanti e le loro storie. Competenze linguistiche, strategie comunicative, livelli di analisi, 195-206. Palermo: CSFLS. I am the author of Sections 2 and 3. Experimental work (%) Supervision (%) Manuscript (%) Research direction (%) 50% 50% 50% 50% [3] E. Mocciaro (with L. Amenta), 2016. Vuliri + PP nei dati dell’Atlante linguistico della Sicilia. É. Buchi, J.-P. Chauveau, J.-M. Pierrel (éd.), Actes du XVIIe Congrès international de linguistique et de philologie romanes (Nancy, 15-20/7/2013), 1, 933-944. Strasbourg, ÉliPhi. I am the author of Sections 2 and 4. Experimental work (%) Supervision (%) Manuscript (%) Research direction (%) 50% 50% 50% 50% [4] E. Mocciaro (& L. Amenta), 2018. Il verbo vuliri in siciliano antico tra volizione e modalità, Bollettino del Centro di Studi filologici e linguistici siciliani 29: 147-176. I am the author of Sections 2, 3, 5. Experimental work (%) Supervision (%) Manuscript (%) Research direction (%) 50% 50% 50% 50% [5] E. Mocciaro, 2019. La grammaticalizzazione dei verbi di movimento in siciliano: il caso di iri ‘andare’ in funzione direttiva. Bollettino del Centro di studi filologici e linguistici siciliani 30: 181-215. Experimental work (%) Supervision (%) Manuscript (%) Research direction (%) 100% 100% 100% 100% [6] E. Mocciaro (& L. Amenta, L. Brucale), 2021. Modalità orientata sul partecipante in siciliano contemporaneo: un’analisi dei dati dell’Atlante linguistico della Sicilia (ALS). L. Schøsler, J. Härmä, J. Lindschouw (eds), Actes du XXIXe Congrès international de linguistique et de philologie romanes (Copenhague, 1-6/7/2019), 2, 1029-1042. Strasbourg, SLR/ELiPhi. I am the author of Sections 4 and 5. Experimental work (%) Supervision (%) Manuscript (%) Research direction (%) 33% 33% 33% 33% 2 Bibliographic record of a published scientific result, which is part of the habilitation thesis.