Agouraki, Y. (1991). A Modern Greek complementizer and its significance for Universal Grammar. UCL Working Papers in Linguistics, 3, 1–24.

Βελούδης, Γ. (1985). Η δήλωση του χρόνου στα 'να' συμπληρώματα. Μελέτες για την Ελληνική Γλώσσα, 6, 183–198.

Βελούδης, Γ., & Φιλιππάκη-Warburton, Ε. (1983). Η υποτακτική στη νέα ελληνική. Μελέτες για την Ελληνική Γλώσσα, 4, 151–165.

Charitonidis, C. (2007). Towards a typological classification of Modern Greek. Studies in Greek Linguistics, 27, 52–64.

Giannakidou, Α., & Mari, Α. (2021). Truth and veridicality in grammar and thought: Mood, modality, and propositional attitudes. The University of Chicago Press.

Haberland, H. (2010). Mood in Greek. In Β. Rothstein & R. Thieroff (Eds.), Mood in the languages of Europe (pp. 473–491). Amsterdam/Philadelphia: John Benjamins.

Oikonomou, D. (2016). Covert modality in root contexts [Doctoral dissertation]. Massachusetts Institute of Technology.

Pavlidou, T. (1991). Cooperation and the choice of linguistic means: Some evidence from the use of the subjunctive in Modern Greek. Journal of Pragmatics, 15, 11–42.

Philippaki-Warburton, I. (1994). The subjunctive mood and the syntactic status of the particle na in Modern Greek. Folia Linguistica, XXVIII(3–4), 297–328.

Philippaki-Warburton, I., & Spyropoulos, V. (2004). A change of mood: the development of the Greek mood system. Linguistics, 42, 791–817.

Rivero, M.-L. (1994). Clause structure and V-movement in the languages of the Balkans. Natural Language and Linguistic Theory, 12, 63–120.

Rivero, M.-L., & Terzi, A. (1995). Imperatives, V-movement, and logical mood. Journal of Linguistics, 31, 301–332.

Roussou, A. (2000). On the left periphery: Modal particles and complementizers. Journal of Greek Linguistics, 1, 65–94.

Τσαγγαλίδης, Α. (2001). Κριτήρια τροπικότητας: Η κατηγορία των τροπικών μορίων στα νέα ελληνικά. Μελέτες για την Ελληνική Γλώσσα, 21, 759–770.

Tsoulas, G. (1993). Remarks on the structure and interpretation of νa-clauses. Studies in Greek Linguistics, 14, 191–206.

Anastasia Giannakidou
The University of Chicago
giannaki@uchicago.edu

While many propositional attitude verbs in Greek select their mood (e.g., modal verbs select subjunctive, belief and knowledge verbs the indicative) some attitudes are fluid with respect to the mood they allow in their complements, specifically verbs of semblance (SEEM) and of perception (SEE). As we can see, the choice of subjunctive or indicative correlates with the availability or not of a veridical inference (as observed in Giannakidou and Mari, 2021):

1. Ta paidia fenonde na ine kourasmena (ala bori ke na min ine). the children seem.3pl that.SUBJ be.3sg tired The children seem to be tired (but they might not be).
2. Ta paidia fenonde oti ine kourasmena (#ala bori ke na min ine) the children seem.3pl that.SUBJ be.3sg tired It is obvious that the children are tired (#but they might not be).
3. Ta paidia fenonde pu ine kourasmena (#ala bori ke na min ine). the children seem.3pl that.SUBJ be.3sg tired The children are tired, and it is apparent.

We see the difference also in English. I will propose, following Giannakidou and Mari that APPEAR is a flexible attitude sometimes construed as forming a belief (using a doxastic base Dox the appearance of P as evidence for P) and sometimes as `suppositional’, given below:

4. SEEM as belief (fenete oti/ Greek): [[SEEM p ]]^{w,Dox,speaker} is = 1 iff ∀w^′[w^′ ∈ Dox(speaker)) → w^′ ∈ {w^′′ | p(w^′′)}]
5. Suppositional SEEM (fenome with subjunctive in Greek): [[i SEEMsup p]]^{K,Dox,speaker} is defined iff K(speaker) is nonveridical (partitioned into p and not p words). If defined, [[i SEEMsup p]]^{M,Dox,speaker} = 1 iff ∀w^′[w^′ ∈ Dox(speaker)) → w^′ ∈ {w^′′ | p(w^′′)}]

The suppositional complement gives nonveridicality because it does not entail belief of p. The knowledge modal base K(speaker), on the other hand, is partitioned for supposition, but not in the case of the pu complement which presupposes knowledge of p. What is crucial for the subjunctive choice is not entailing belief of p, which renders suppositional seem akin to an epistemic MUST modal. Exactly the same, I will argue, holds for perception verbs:

6. O Nicholas idhe ton Flavio na kleini tin porta, ala i porta dhen ine the Nicholas saw.3sg the Flavio that.SUBJ close.3sg the door, but the door not closed Nicholas saw Flavio closing the door, but the door is not closed.
7. O Nicholas idhe oti o Flavio eklise ton porta, #ala i the porta den ine the Nicholas saw.3sg that.IND the Flavio closed.3sg. the door, but door not is closed. #Nicholas saw that Flavio closed the door, #but the door is not closed.

Here too, I will argue, the lexical entry is flexible, and can associate with either a doxastic modal base (Dox) or a nonveridical one K. This flexibility affects veridicality as is reflected in the distinct mood pattern. Therefore, the mood flexibility does not necessitate ambiguity in the lexical entries of attitude verbs; rather it shows variability in being able to pick different modal bases. With an epistemic nonveridical base the subjunctive will be chosen. Overall, the big conclusion seems to be that propositional attitudes and modals are closer conceptually than what is usually assumed, and that mood choice is inextricably tied to belief formation.

Brian D. Joseph
The Ohio State University
joseph.1@osu.edu

Veloudis and Philippaki-Warburton (1983) offer an analysis of the subjunctive in Modern Greek in which a key argument for a category of “subjunctive” comes not from special morphology or special functions that distinguish a subjunctive mood from an indicative mood, but rather from the fact that there are distinct negation markers, δεν for indicative and μη(ν) for subjunctive. The distribution of these negators correlates further with the presence or absence of additional markers, specifically να and ας, which are negated with μη(ν), as opposed to δεν, which is impossible with να and ας, and occurs with verbs with no such marker.

These Greek facts are interesting and important in their own right, given the rather outsized role that a “little word” like να plays in Greek syntax. That is, no analysis of Greek complementation, including both subordination (embedded-clause usage) and insubordination (main-clause usage), can ignore να, as it is not only ubiquitous but also a key element in complement structures. At the same time, however, the Greek facts regarding να and the subjunctive mood more generally cannot be viewed in isolation, since there are important parallels to be noted between the Greek situation with να and similar sets of facts in various of the Balkan languages, most importantly Albanian, Aromanian, and Macedonian.

While Albanian and Aromanian (and Balkan Romance more broadly), unlike Greek and Macedonian, have special morphology for the subjunctive (e.g. Albanian endings 2sg -sh and 3sg -jë, along with stem changes for some verbs), it is more the behavior of the elements that are the counterpart to να that is of particular interest: Albanian të, Aromanian să, and Macedonian da. Moreover, the argument within Greek noted above, based on negation, for a subjunctive mood marked by να finds an exact parallel in the behavior of Albanian mos as a modal negator distinct from nuk or s’ as indicative negators. Furthermore, the parallel behavior extends beyond να and negation, since the construct consisting of the verb and various modifiers for tense, mood, negation, and argument structure has the various elements lined up exactly parallel in these languages, giving essentially a template for a multi-faceted unit serving as the “verbal complex”. Thus, for ‘I will not give it to him’, we find the following in Albanian (a), dialectal Greek (b), Romanian (c), and dialectal Macedonian (d):



Beyond this descriptive contribution to an understanding of Greek να, it is considered here whether this parallelism is of historical interest as being due to language contact within the Balkan Sprachbund or is just a matter of typological interest, being due to convergent language- particular developments.

Eleni Karantzola¹ & Konstantinos Sampanis^2
1University of the Aegean, ²University of Vienna
ekarantzola@gmail.com, konstantinos.sampanis@univie.ac.at

While the seminal work of Joseph (1983) traced the gradual decline of infinitival constructions in Greek and other (Balkan) languages there is no detailed diachronic account for the rise of the “na-subjunctive” (cf. e.g. Markopoulos, 2005 for a brief approach). In this paper we firstly present a thorough diachronic corpora investigation (mainly of TLG and a corpus of Late Middle and Early Modern Greek texts) that covers the period from late Koine onwards to the 18^th c. which focuses a) on the gradual verb-affixation of the conjunction ἵνα (h)ína in complement clauses and b) on the phonological attrition of (h)ína that occurred due to its affixation (cf. Trypanis, 1960) and, probably, under the influence of the particle ας as. We use the term affix by following Haspelmath´s (2021, p. 18) definition that “[a]n affix is a bound morph that is not a root, that must occur on a root, and that cannot occur on roots of different root classes”. On the grounds of our corpora analysis and Haspelmath’s definition, we argue that:

1. Diachronically, the subjunctive marker (h)ína/na took over the semantic features of the old subjunctive in a language change process that we describe as cyclical (in the sense of van Gelderen, 2011).
2. Synchronically, we analyze the na-marker as an affix, i.e. an element bound to a verb and therefore part of its paradigm in conjunction with the rest of preverbal affixes, namely the concessive marker ας as and the future marker θα θa. This approach rules out theories which consider na to be a syntactic element beyond the verbal phrase.
3. Typologically, Modern Greek is contrasted to cross-linguistic cases of preverbal affixation beyond the Balkansprachbund (e.g. to Farsi). We postulate that preverbal affixation is a “strategy” to which languages resort in order to develop novel mood verbal forms when the existing verbal suffixes are solely exponents of After a person ending further suffixation is not allowed for the expression of mood, therefore preverbal mood markers emerge.

References

Gelderen, E. van (2011). The linguistic cycle: Language change and the language faculty. Oxford: Oxford University Press.

Haspelmath, M. (2021). Bound forms, welded forms, and affixes: Basic concepts for morphological comparison. Voprosy Jazykoznanija, 1, 7–28. https://doi.org/10.31857/0373-658X.2021.1.7-28

Joseph, B. D. (1983). The synchrony and diachrony of the Balkan Infinitive, a study in areal, general, and historical linguistics. Cambridge: Cambridge University Press.

Markopoulos, T. (2005). Categorial features and grammaticalization: The case of Medieval Greek ‘na’. In Proceedings of the 7th International Conference on Greek Linguistics, York, 8th–10th September 2005.

Trypanis, C. A. (1960). Early medieval Greek ἵνα. Glotta ,38, 312–13.

Adam Ledgeway¹, Norma Schifano² & Giuseppina Silvestri^3
1University of Cambridge, ²University of Birmingham, ³University of California Los Angeles
anl21@cam.ac.uk, N.Schifano@bham.ac.uk, g.silvestri@ucla.edu

We examine in the Greek and Romance dialects of the extreme south of Italy the core dual contrast in the complementizer system between realis and irrealis propositional modality, investigating not only the opposition between modally differentiated complementizers, but also the principal formal and structural correlates of this complementizer opposition within the sentential core in relation to mood, variation in verb placement, tense, subject positions, and negation. In particular, although both linguistic groups display a dual complementizer system, both linguistic groups have also lost an original formal contrast on the verb between indicative and subjunctive moods, with the generalization of an invariant form un(der)specified for tense in irrealis clauses which contrasts with the deictic temporal specification of realis verb forms. Yet, this does not imply that modality is only marked on the C-system, since in both Italo-Greek and in Romance the contrast continues to surface syntactically through an asymmetry in the position of the verb in accordance with the realis- irrealis opposition. In particular, we shall highlight a lower position of the finite verb in realis clauses than in irrealis clauses, where the verb occurs in the highest available position within the sentential core. In light of these facts, it will be argued that the higher verb placement observed in Romance irrealis clauses is a consequence of contact with Italo-Greek. Thus, the Romance dialects of this area borrowed not only the superficial finite subordination pattern, but they also borrowed the syntax (namely, the exceptionally high verb placement) associated with the introduction of an irrealis complementizer. By contrast, root clauses and realis complement clauses, which do not typically replace the infinitive, were left unaffected and continue to display the default low(er) verb placement typical of all Romance dialects of the south of Italy (Ledgeway and Lombardi 2014; Ledgeway 2020). The result is a hybrid grammar which combines southern Romance low(er) verb placement in realis contexts with Greek-style high verb placement in irrealis contexts.

At the same time, the observed absence of temporal marking on irrealis verb forms offers us an elegant explanation for the ungrammaticality of immediately preverbal subjects in irrealis clauses. Irrealis clauses lack the tense-related projections found in realis clauses which license the canonical preverbal subject position (in accordance with the so-called EPP requirement), an observation which we argue explains the typical postverbal position of the subject in irrealis clauses.

Mood marking also surfaces in the area of negation. While Italo-Greek distinguishes lexically between two negators (den/min), no such lexical distinction is found in Romance. Nonetheless, it will be demonstrated that the same realis-irrealis structural distinction surfaces in the variable position of reflexes of Lat. NON in these Romance dialects which can lexicalize positions both within the sentential core and the left periphery, highlighting a significant difference between MAT(ter) and PAT(tern) borrowing (Matras & Sakel 2004; 2007).

In summary, although our examination of the fine structure of the left periphery reveals a remarkable degree of structural and semantic parallelism in the distribution and functions of the complementizer systems of Italo-Greek and their neighbouring Romance varieties which cannot but be the product of centuries of intense and extensive language contact from Greek on Romance, there are also some cases of Romance structural influence on Italo-Greek modal marking. For example, of the two complementizers, the realis complementizer is arguably the least marked, since it may substitute the irrealis complementizer in its core functions in accordance with a widespread southern Italo-Romance pattern. In such cases irrealis modality is no longer marked explicity in the C-system, but is marked in the sententail core through an expletive deontic modal MUST, which, in the absence of the irrealis complementizer, is pressed into service to overtly mark the clause as irrealis.

Marika Lekakou¹ & Josep Quer Villanueva^2
1University of Ioannina, ²Universitat Pompeu Fabra
mlekakou@uoi.gr, josep.quer@upf.edu

A long-standing controversy in Greek syntax concerns the status of na: is it an inflectional element heading MoodP (Philipaki-Warburton, 1994, 1998; Rivero, 1994 among others) or is it a complementizer (Agouraki, 1991; Tsoulas, 1993)? Roussou (2000) proposes that it is both. On her proposal, the CP-domain includes three distinct, hierarchically ordered C-positions: C, COp and CM, the lower one of which bears modal features. Na merges in CM and obligatorily moves to the intermediate COp. Oti is base-generated in COp and optionally moves to C. C hosts pu.

While the proposal effectively captures the dual nature of na and goes a long way in accounting for the co-occurrence restrictions among the complementizers and particles of Greek, it faces some problems. One of them is that, despite the proliferation of C-heads, the empirical goals are not fully attained. For instance, the sequence pu oti is not ruled out, since oti only optionally moves to C. Another set of complications arises in connection to the lack of true negative imperatives in Modern Greek (but not universally): negation evidently does not intervene for the purported movement of na, but it does for V-movement to COp.

In this talk, we follow Roussou in spirit, in that na is not exclusively a complementizer or a modal element. However, it is not always both such things, but rather either one, depending on the sentential context. In other words, na is sometimes lexicalizing C and other times a lower head in the clausal spine. To do this, we exploit the framework set out in Wurmbrand & Lohninger (2020) (henceforth W & L), based on Givón’s (1980) universal hierarchy of clausal domains. The gist of their proposal is that, across languages, there are three main types of clausal complements which are distinct in terms of independence, transparency and integration with respect to the main predicate. These complement types correlate with the semantic entity they encode: propositions, situations and events. A verb like say embeds a propositional complement, corresponding to CP; a verb like promise combines with a situation-type complement, corresponding to TP; a verb like begin embeds an event-type complement, corresponding to vP. The three complement types form part of an implicational hierarchy of ‘clausehood’. Key semantic and syntactic differences among the three types concern the kind of control they (dis)allow, and in terms of (in)dependence of temporal reference.

Following W & L (2020), oti lexicalizes the proposition-type complement head. We depart from W & L, however, concerning na and argue that it may lexicalize the head related to either one of the three types of complements. Concretely, na may head a situation-complement (under e.g. iposxome ‘promise’), an event-complement (under e.g. arxizo ‘begin’), but also a propositional complement. We argue that na-clauses are of the propositional type in at least two cases not considered by W & L: complements to epistemic modals, as in (1), and polarity subjunctives, as in (2). In both these cases, the na-clause is temporally independent from the matrix tense, and an overt subject distinct from the matrix one is allowed. The entire complement clause, we argue, is propositional in nature.

(1)  Prepi na irthe o Andreas.                                          (2)   Les na irthe o Andreas?
must SUBJ came-3SG the Andreas                                say-2SG SUBJ came-3SG the Andreas
‘Andreas must have come.’                                             ‘Do you reckon Andreas has come?’

We observe a parallelism between (1) and (2) and proposition infinitives in English according to W & L (Clara believes Danny to be eating salad right now). “Subjunctive” thus finds a comfortable home within this system, where categories such as infinitives have no unique semantic correlate, and there is no single syntactic locus of finiteness.

Key References

Roussou, A. (2000). On the left periphery: Modal particles and complementizers. JoGL, 1, 65–94.

Wurmbrand, S., & Lohnlinger, M. (2020). An implicational universal in complementation: Theoretical insights and empirical progress. In J. Hartmann & A. Wöllstein (Eds.), Propositionale Argumente im Sprachvergleich: Theorie und Empirie. Tübingen: Gunter Narr Verlag.

Despina Oikonomou
University of Crete
despina.oikonomou.ling@gmail.com

In this paper, we investigate subjunctive questions in Greek providing evidence in favor of an analysis of subjunctive mood as a modal element (Giannakidou, 2015; Philippaki-Warburton, 1992; Rivero, 1994; Roussou, 2000). Subjunctive Questions (SQs) as in (1) are differentiated from Indicative Questions (IQs) (e.g. Ti tha fai/troi to moro gia vradino?) by the presence of the subjunctive particle na (Veloudis & Philippaki-Warburton, 1983). In terms of their interpretation, SQs query the questionee’s prioritizing state (i.e. desires, duties, goals) whereas IQs are about the questionee’s doxastic/epistemic state (i.e. beliefs about the actual world) (Giannakidou, 2009; Oikonomou 2016; Rouchota, 1994).

1. Ti na fai/troi to moro gia vradino? ‘What should the baby eat for dinner?’

We notice that, although subjunctive in non-interrogative environments is restricted under certain predicates, subjunctive interrogative complements have wider distribution. Predicates like siniditopio ‘realize’, anakalipto ‘discover’, katalaveno ‘understand’, ime siguros/veveos ‘I’m certain’, cannot combine with a non-interrogative subjunctive (2b) but can embed an SQ (2a).

2. Molis siniditopiisa pos na tu milao gia na min pareksigite, i shesi mas veltiothike.
3. *Molis siniditopiisa na tu milao me evgenia, i shesi mas veltiothike.

The presence of prioritizing modality in matrix SQs and in sentences as in (2), can be explained under a modal analysis of subjunctive (SUBJ). Following a decompositional analysis to attitude predicates (Kratzer, 2006; Moulton, 2009), SUBJ in (3) is analysed as a possibility modal with dual background, i.e. a modal base f and an ordering source g (Kratzer, 2016; Portner & Rubinstein, 2020). We argue for existential semantics, since a subjunctive can be used to request or provide permission, without imposing a requirement (see also Giannakidou, 2009, 2015).

4. 

Crucially, under (3), the flavor of SUBJ is not lexically determined in the entry of SUBJ. Rather, it is defined by the content of the embedding event. The dependency between SUBJ and the matrix event is modeled under Hacquard’s (2006, 2010) event relativity approach. Since SUBJ requires a dual modal background, we can explain why it combines primarily with prioritizing predicates as opposed to doxastic/epistemic predicates (e.g. siniditopio ‘realize’) which are typically characterized by a single doxastic/epistemic content (Portner & Rubinstein, 2020). For matrix non-interrogative subjunctives (e.g. Na erthis), we assume that they are embedded under a covert speech act operator with directive illocutionary force, defining the flavor of SUBJ (Rivero & Terzi, 1995). Under this approach, a crucial question arises as to how the flavor of SUBJ is defined in matrix and embedded interrogative environments as in (1) and (2a). The lexical meaning of siniditopio ‘realize’ does not have a prioritizing component, it is epistemic. Likewise, a question speech act event asks for information, without defining whether it is about the questionee’s prioritizing or epistemic state (Giannakidou & Mari, 2021; Grano, 2018). Following Bhatt’s (2006) analysis of infinitival questions, prioritizing semantics in SQs arise due to the question operator and not due to the embedding predicate, which should otherwise be ambiguous. Finally, in addition to the modal flavor of SQs, we investigate the variable quantificational force of SUBJ, proposing an Exhaustication-based approach (Chierchia et al., 2012).

Selected references

Βελούδης, Γ., & Φιλιππάκη-Warburton, Ε. (1983). Η υποτακτική στη νέα ελληνική. Μελέτες για την Ελληνική Γλώσσα, 4, 151–165.

Giannakidou, A. (2009). On the temporal properties of mood: The subjunctive revisited. Lingua, 119, 1883–1908.

Portner, P., & Rubinstein, A. (2020). Desire, belief, and semantic composition: Variation in mood selection with desire predicates. Natural Language Semantics, 28, 343–393.

Vassilis Spyropoulos
National and Kapodistrian University of Athens
vspyrop@phil.uoa.gr

Veloudis & Philippaki-Warburton’s approach to Greek subjunctive in their (1984) seminal paper made a very significant assumption, which was further developed and asserted in Philippaki-Warburton’s subsequent work (1992, 1994, 1998), namely that mood should not be understood and defined with reference to modality, but it is rather a purely grammatical category defined by means of morphosyntactic marking and distribution (see also Philippaki- Warburton & Veloudis, 1985; Philippaki-Warburton & Spyropoulos, 1999, 2004, 2006; also Tsangalidis, 2001a,b, 2002). Such a proposal severs mood from modality, in the sense that modality does not select for a mood and mood is not defined by means of the modality it may express. This, of course, does not mean that there is no (prototypical) mapping between a mood and certain modalities; however, this mapping should not be considered a defining and grammatical property of mood. The significant implication of such an assumption is that mood is more related to clause type than to modality.

In this paper, I will attempt to further entertain this implication and claim that mood is a clause-type category and not a modality marking device. I will propose that the major evidence for such a claim comes from the distribution of subjunctive in embedded clauses. In particular, the appearance of subjunctive in complement clauses of verbs that denote (sensory) perception, a phase of the event, a psychological state, or a certain way of acting or being/becoming (e.g., akuo ‘I hear’, ksero ‘I know’, arxizo ‘I start’, simveni ‘it happens’, fenome ‘I have the appearance of’, θimame ‘I remember’, etc.) cannot be explained by an approach in which subjunctive is defined by being dependent on an intentional, negative or, generally, modal operator (Manzini 2000). In fact, I will show that the diverse distribution of subjunctive in complement clauses, which is the result of the diachronic development of complementation in Greek, and, more specifically, of the substitution of non-finite complement clauses involving the infinitive and the ‘supplementary’ participle by finite clauses, can only be understood as the result of mere lexical selection by the matrix predicate, which does not necessitate semantic features. Any modality associations may result from the semantic properties of the matrix predicate, but, what is crucial, mood selection does not make reference to them. I will, then, explore syntactic formalizations of how such a selection works, including the probing approach discussed in Angelopoulos (2019), and I will address some issues that arise with respect to the locality of this selection. Finally, the modality-free approach to mood will allow me to draw a distinction between mood particles, such as na, and modal particles such as θa, which implies that these two particles may not occupy the same syntactic position at any stage the derivation (see also Roussou & Tsangalidis, 2010; cf. Roussou, 2000).

Selected References

Philippaki-Warburton, I. (1994). The subjunctive mood and the syntactic status of "na" in Modern Greek. Folia Linguistica, 28, 297–328.

Veloudis, I., & Philippaki-Warburton, I. (1984). Subjunctive in Modern Greek [In Greek]. Studies in Greek Linguistics, 4, 151–168.

Roussou, A., & Tsangalidis, A. (2010). Reconsidering the ‘Modal Particles’ in Modern Greek. Journal of Greek Linguistics, 10, 45–73.

Anastasios Tsangalidis
Aristotle University of Thessaloniki
atsangal@enl.auth.gr

The semantics of the subjunctive has been notoriously difficult to describe in both cross- linguistic (e.g. Givón, 1994; Thieroff, 2010) and Greek-specific terms (e.g. Haberland, 2010; Veloudis, 2010). It is often contrasted to the indicative as a means of referring to a situation in a non-assertive way. As described in Holton et al. (2012, p. 266), “the indicative is normally used for describing facts or states, the subjunctive for talking in general about non-facts”.

The paper focuses on some cases in which speakers apparently use the subjunctive to describe facts rather than non-facts. The relevant data involve four distinct cases in which the presence of na seems to be compatible with the speaker’s belief that the propositions introduced by na are actually true. Thus, as is generally accepted, speakers will generally use an indicative sentence to assert that a proposition p is the case. However, there are clear instances of subjunctive sentences in which the speaker assumes that p is the case and, further, comments upon it (a) by referring to the source of their evidence (mostly sensory evidence), e.g. Ακούω να παίζουν μουσική or Τον βλέπω να έρχεται – Delveroudi et al 1994; (b) by expressing surprise (= apparently asserting both p and that p is unbelievable), e.g. Να μη γνωρίζει τον άντρα της or Να τους κοροϊδεύει κι αυτοί να τον χειροκροτούν – Veloudis 2010; (c) by maximizing the juxtaposition with another proposition for rhetorical purposes, e.g. Μπορεί να είναι μικρός, αλλά δεν είναι χαζός and (d) by embedding it under an explicitly evaluative matrix predicate, e.g. Έχω την τιμή να εκπροσωπώ τον Πρύτανή μας or Έκανα το λάθος να τους ψηφίσω.

In formal terms, it is interesting to note that there are various restrictions on tense and aspect choices affecting the availability of particular interpretations (as also shown in Giannakidou & Mari, 2021; Roussou, 2006; Veloudis, 1985). Moreover, interpretation seems to relate to specific properties of the selecting predicates in the case of embedded na-clauses, while in the case of matrix na-clauses the particular interpretations are dependent on both ordering and contextual restrictions.

In all these cases, the presence of na does not question the actuality of the situations described, in the sense that the speaker clearly considers them to be part of known reality. Indeed, to the extent that these subjunctives can alternate with indicatives (i.e. either zero marked forms or forms accompanied by tha rather than na) the mood distinction is apparently not truth-conditional: both the indicative and the subjunctive seem to be available options for describing facts, rather than non-facts. Overall, it seems necessary to draw a distinction between (non-)factivity and (non-)assertion: in view of the data reviewed, it is argued that the subjunctive can be used to refer to facts (as well as non-facts); however, that the subjunctive is never used to assert a proposition remains a valid generalization.

Selected References

Delveroudi, R., Tsamadou, I., & Vassilaki, S. (1994). Mood and modality in Modern Greek: The particle na. In I. Philippaki-Warburton, K. Nicolaidis, & Μ. Sifianou (Eds.), Themes in Greek Linguistics (pp. 185–192). Amsterdam/Philadelphia: Benjamins.

Giannakidou, Α.. & Mari, Α. (2021). Truth and veridicality in grammar and thought: Mood, modality, and propositional attitudes. The University of Chicago Press.

Givón, T. (1994). Irrealis and the subjunctive. Studies in Language, 18(2), 265–337.

Haberland, H. (2010). Mood in Greek. In Β. Rothstein & R. Thieroff (Eds.), Mood in the languages of Europe (pp. 473–491). Amsterdam/Philadelphia: John Benjamins.

Holton, D., Mackridge, P., & Philippaki- Warburton, I. (2012). Greek: A comprehensive grammar (2^nd ed.). Revised by V. Spyropoulos. London and New York: Routledge.

Ρούσσου, Α. (2011). Συμπληρωματικοί δείκτες. Αθήνα: Πατάκης.

Thieroff, R. (2010). Moods, moods, moods. In B. Rothstein & R. Thieroff (Eds.), Mood in the languages of Europe (pp. 1–29). John Benjamins Publishing Company.

Βελούδης, Γ. (1985). Η δήλωση του χρόνου στα 'να' συμπληρώματα. Μελέτες για την Ελληνική Γλώσσα, 6, 183–198.

Βελούδης, Γ. (2010). Από τη σημασιολογία της ελληνικής γλώσσας: Όψεις της επιστημικής τροπικότητας. Θεσσαλονίκη: ΙΝΣ (Ίδρυμα Μανόλη Τριανταφυλλίδη).

George Tsoulas
University of York
george.tsoulas@york.ac.uk

It is well known that verbal and nominal (extended) projections display a set of common properties that sets them apart from, say, A(djectival/dverbial) or prepositional projections, (amongst many others: Abney, 1987; Ogawa, 2001; Szabolcsi, 1996). The parallelism in question has been reinforced in recent years in the so-called exoskeletal models of the grammar (Borer, 2005a,b; et seq). Within this setting the question that arises is that of the correspondence between functional elements on the extended projection of nominal and that of verbal elements. Most attention in the literature so far has focused on the variable perhaps correspondence between CP/TP and DP and, roughly speaking, the idea is that the argumental nature of Cps and DPs puts them on a par, with variation regarding clauses of smaller size (TPs). This picture is, however, too coarse: One question that arises concerns the degree to which the two domains must parallel one another. Assuming that they do and that they do so closely, it follows that languages that lack overt Ds, which have been independently argued to have bare NP arguments, would present a direct challenge to the idea that CPs qua arguments must, as a rule, always parallel DPs. In this respect, the issue to be settled concerns the relationship between the argument status of bare nouns and the (necessary) projection of the upper layers of the clause. More specifically concerning Greek, Alexopoulou et al. (2013) have argued that nominal arguments in Greek, rather than DPs, are Number phrases. If this is so - and still in keeping with the idea that there is a strong parallelism - then it follows that CPs will be equivalent to NumPs in Greek. This has two natural consequences, the first is that the nominal equivalent of mood will be lower than Num, and that D will have no obvious counterpart in the verbal domain. Setting the second question aside, we will develop further the idea that the modal heads not only have a corresponding element in the nominal projection but that head is the definiteness head. Splitting the functions traditionally associated with D, we propose that definiteness and argumentization are represented by Def and Num respectively. The correspondence is as follows:

(1)
CP – NumP
MP – DefP
TP – ClP

To substantiate the correspondences, we will develop further the arguments for subjunctives representing verbal/clausal (in)definiteness given originally in Tsoulas (1994), Manzini (1994), Iatridou (2013) etc. The second aspect of the classification in (1) that we will zoom in is the analysis of tense as a classifier over events. We finally argue for a separation of the argumentizing function associated with one head of the extended projection from its natural semantics.