Mind the Gap! Null Modals (and Other Functional Verbs) in Finite Complementation in Italo-Greek

Abstract

Although Italo-Greek is characterized by a general retreat of infinitival complementation, it partly preserves the infinitive in restructuring contexts: a handful of functional auxiliaries—in an overt or covert form—allow for infinitival complements, with which they enter in a monoclausal union. Such a preservation also triggers consequences for finite complementation. Indeed, those predicates that still select for infinitival complements may lack finite complementation, resulting in only the lexical embedded verb surfacing instead of the complex sentence AUX + na + finite verb: δen du ékame típote ‘he could not do anything to him’ (lit. ‘he did not do anything to him’). I claim that such an absence—which gives rise to a semantic or even a syntactic gap—depends on the effects of the restructuring rule, which creates a high level of dependency and interlacing between the matrix and embedded verbs. When finite complements gradually replaced infinitival ones, though only sporadically, some predicates stopped selecting for finite complements, ultimately depriving the sentence of modal and other functional specifications.

1. Introduction

Italo-Greek—on par with both other Greek varieties since Late Antiquity and the languages of the Balkan Sprachbund (Joseph 1983; Tomić 2006, pp. 413–655)—is characterized by the general avoidance of the infinitive in dependent clauses, when the main verb expresses will, wish, aim or intention (see, among others, Baldissera 2013; De Angelis 2013b, pp. 20–24; Katsoyannou 1992, 2001, pp. 47–52; Ledgeway 2013, 2016, 2023; Manzini and Savoia 2005, I, pp. 650ff.; Rohlfs [1958] 1972, 1977a). Both in Calabrian Greek (Greko Bovese, henceforth: Greko) and in the Greek spoken in Salento (Griko), the infinitive1 is replaced by a finite verb form (built on the aoristic stem), invariably inflected in the present tense. Such verbs are headed by a complementizer (=comp_sbjv), generally na, derived from Ancient Greek hína ‘so that, in order that’, which originally introduced purpose clauses (Joseph 1981; Tsangalidis 2004, pp. 198–200):

(1.a)	θeli	na	mini
	want.3sg	compsbjv	stay.3sg
	‘He/she wants to stay’ (Greko)
(1.b)	iχa		na	mino
	have.ind.impf.1sg		compsbjv	stay.1SG
	‘I had to stay’ (Griko)

As (1.a,b) show, the complementizer na2 is generally selected for “future-oriented” predicates, that is, predicates that encode «[…] an action or state which cannot be verified at the time of speaking» (Ledgeway 1998, p. 21). As such, it is traditionally labelled as irrealis and is used to introduce subjunctive clauses, in which information relating to time and space anaphorically depends on the main clause predicate. Similarly, due to Greek influence, in most Italo-Romance dialects spoken in Salento, central-southern Calabria and north-eastern Sicily a broad range of subordinate clauses—generally codified by the infinitive in Standard Italian and most Western Romance languages—are also replaced by finite clauses introduced by an irrealis complementizer, whose morpholexical shape is diatopically differentiated, realized as (m)i, (m)u, ma (<Lat. (quo)mŏdo) in north-western Sicily and central-southern Calabria, and as cu (<Lat. quod) in Salento (Calabrese 1993; Chillà and Citraro 2012; De Angelis 2013a; Ledgeway 1998, 2007, 2013; Loporcaro 1995; Roberts and Roussou 2003; Rohlfs 1922, 1961, among others).

However, it has been observed that—in both Italo-Greek and in extreme southern Italo-Romance varieties—not all the verbs share this innovation. Among those more resistant to change, CAN (Greko sonno, Griko sozo [see Rohlfs 1964 s.v. σώζω], Sic. and central-southern Calabr. putiri) stands out as preserving the embedded infinitive more than other predicates, though not categorically (Baldissera 2013; De Angelis 2013a, 2013b; Ledgeway 2013, 2023; Remberger 2011; Trumper and Rizzi 1985)3. More specifically, as far as Griko is concerned, CAN is the only predicate that licenses infinitival complements, whereas in Greko, HEAR, and, to a lesser extent, KNOW and MAKE, can also select for infinitives (Ledgeway 2013; 2016, pp. 1024–25; 2023). See the following examples from Greko:

(2)	δe	ssónnome	χórtai
	neg	can.1pl	satiate.inf
	‘We cannot satiate (ourselves)’ (Roccaforte, TNC 1.3)

(3)	egò	δen	ákua	mai	mentuvéspi	ettúndom	mágo
	I	neg	hear.ind.aor.1sg	ever	discuss.inf	this	magician
	‘I have never heard of this magician’ (Roccaforte, TNC 9.10)

(4)	Egò	éχo	mía	lleḍ̣ḍ̣̣á […]	pu	šéri	kámi	túnda prámata.
	I	have	a	sister	who	know.3sg	do.inf	these things
	‘I have a sister who knows how to do these things’ (Roccaforte, TNC 8.11)

(5)	m’ékame	jelísi
	me=make.ind.aor.3sg	laugh.inf
	‘He made me laugh’ (Roghudi; Rohlfs 1977a, p. 191)

The predicates that select infinitival complements fall under the category of functional verbs (including modal and aspectual, verbs of perception, and causative verbs), which undergo restructuring. Following the seminal work of Rizzi (1976), biclausal structures, consisting of a matrix clause followed by an infinitival one, are said to undergo a process of clause union, whereby a monoclausal structure is formed under certain syntactic conditions (Calabrese 1993, pp. 29–32; Cinque 1999, 2004, 2006; Manzini and Savoia 2005, I, p. 650ff.; Cruschina and Ledgeway 2016, pp. 563–65; Ledgeway 2016, p. 1022). The main predicates involved in this process lexicalize different functional (temporal, modal or aspectual) positions, and select for a (infinitival) reduced v-VP complement (Hill 2017; Ledgeway 2023).

The primary aim of the present paper is to analyze the effects of this (covert and overt) mandatory association of functional predicates with their embedded infinitival clauses in restructuring contexts. I would argue that the loss of infinitival complements, and, accordingly, of restructuring structures, leads to a semantic and even syntactic gap in licensing modal (as well as other functional) auxiliaries in finite complementation. For reasons of time and space, the analysis will be restricted to Greko, but its implications may also be tested for both Griko and Italo-Romance dialects in contact with Italo-Greek.

2. Silent Auxiliaries in Restructuring Contexts

Recently, Ledgeway (2023) traced two different structures to the same pattern exhibited by monoclausal, restructuring sentences with an overt functional predicate (see exx. 2–5). The first of these structures, which will be focused on here, consists of the so-called ‘infinitival relative’ (see exx. 6–8), which is a clause «[… in which the infinitive is preceded by a wh-relative, which is itself the complement of an existential predicate» (Ledgeway 2023, p. 29); see, e.g., It. non ho con chi parlare (Caponigro 2021). These headless relative clauses convey an existential meaning, as shown by the possibility of inserting an existentially quantified nominal expression or an indefinite noun. See a possible paraphrase of the above-mentioned Italian example, as well as its English translation, Non ho nessuno con cui parlare ‘I don’t have anyone to talk to’, which both employ a negative indefinite pronoun (nessuno/anybody). Moreover, these clausal complements may only be used with a highly restricted set of verbs, especially the existential predicates ‘be’ or ‘have’ (Grosu 2004, among others).

In such structures, an overt modal operator spells out the possibility/ability (or necessity) of reading (Šimík 2011). However, the operator can remain hidden, as shown by the possibility of inserting potere ‘can’ in the above-mentioned Italian sentence: Non ho con chi posso parlare ‘I don’t have anyone I can talk to’. The evidence supporting the covert occurrence of a modal operator comes from the assumption that these clauses generally require an irrealis marker on the embedded verb (morphologically realized through the infinitive or the subjunctive), which triggers a modal interpretation (Caponigro 2021, pp. 11–12, 34–37). In Italo-Greek—on a par with the languages of the Balkan Sprachbund (Grosu 2004)—they are realized either by na-clauses (exx. 8–9) or by the infinitive (exx. 6–7), in free variation (Ledgeway 2023, p. 30). See the following examples from Greko:

(6)	Ma δen íχai	ti	kámi
	but neg have.impf.ind.3pl	what	do.inf
	‘But there was nothing they could do’ (Roccaforte, TNC 18.5)

(7)	Egò	δen éχo	ti	aḍḍáši
	I	neg have.prs.ind.1sg	what	change.inf
	‘I have nothing to change’ (Roccaforte, TNC 19.21)

(8)	δen	éχi	χristyanúse	pu	na	kámusi ećíndese sonátese
	neg	have.ind.prs.3sg	men	rel	comp.sbjv	make.3pl those	songs
	‘There were no men who could perform those songs’ (Roccaforte, TNC 36.4)

(9)	en	éχo	ti	na	su	δóso
	neg	have.1sg	what	comp.sbjv	you.gen	give.1sg
	‘I have nothing to give you’ (Bova, TNC 1.2)

The same silent modal operator is also licensed in indirect interrogative sentences codified by infinitival complements (see e.g., I don’t know what to do), which share both formal and semantic properties with the infinitival relatives:

(10)	δen	éχo	pu	pái
	neg	have.1sg	where	go.inf
	‘I have nowhere to go’ (Rohlfs 1977a, p. 191)

(11)	δen	izzéro	ti	grázzi
	neg	know.1sg	what	write.inf
	‘I don’t know what to write’ (Rohlfs 1977a, p. 191)

Indeed, in these sentences, too, a modal operator taking an infinitival, indirect interrogative complement, which codifies deontic necessity (as well as a possibility/availability reading) is licensed; witness the possible paraphrasis of the above-mentioned English example: I don’t know what I have to/can do. Similar readings point in favour of a structure involving an underlying covert modal auxiliary, which—unlike the infinitival complements of restructuring predicates—is not phonologically realized:

…[_CP wh- [_TP AUX(_can/must)…[v-VP V_INF]]]

However, Italo-Greek presents indirect interrogative infinitival and subjunctive complements very rarely. This is in line with the general avoidance of the infinitive with MUST, and, consequently, with the silent deontic operator MUST (Ledgeway 2023, p. 31).

Another structure that Ledgeway (2023) considers to be monoclausal is the second-person negative imperative, which in both Romanian and southern Italo-Romance (but not Italo-Greek) is codified by the negator followed by either the infinitive or a finite irrealis clause, which are in free variation:

(12a)	Non ghiri   aa	casa
	neg go.inf to=the	house
	‘Don’t go home!’ (southern Calabrese, Reggio Calabria)

(12b)	Nommi	trasi!
	neg=comp.sbjv	enter.2sg
	‘May he not enter!’ (southern Calabrese, Montebello Ionico, Rohlfs 1977b)

Although scholars generally analyze the occurrence of the infinitive in these structures as a phenomenon of suppletivism, where the infinitive replaces a dedicated finite imperative form, Ledgeway (2023) traces such constructions back to a monoclausal structure with a silent modal. Under this view, the sentential negator licenses a functional head, by forcing the projection of the T-domain, that is the possibility of licensing inflectional features that are otherwise absent in positive imperatives. Indeed, negative imperatives are said to instantiate reduced infinitival complements headed by a silent modal auxiliary, according to the following underlying structure:

[_cp [_tp NEG (AUX)… [v-_vp V_inf]]] (Ledgeway 2023, p. 33).

3. Null Modals (and Other Functional Predicates) in Finite Complementation

To sum up, the stronghold of the infinitive involves a series of (covert and overt) functional predicates. This constitutes a highly limited subset of functional verbs, which undergo restructuring, at least until finite structures (almost entirely) replace the infinitival ones.

The association of a handful of functional auxiliaries with monoclausal structures selecting for infinitival complements seems to have also produced some effects on finite complementation. Indeed, albeit very sporadically, some modal verbs, as well as other functional predicates involved in restructuring contexts, are optionally absent from the sentence, alongside the overwhelming majority of cases in which they head a na-clause. If they are absent, the restructuring complex sentence [FUNCTIONAL AUX + infinitive] can optionally be replaced by a simplex one, consisting of the finite embedded verb only (which in the complex sentence corresponds to the verb in the infinitive), inflected for the same mood and tense as the (silent) auxiliary, as shown by the following almost-minimal pair4 (Mallimaci, forthcoming):

(13.a)	o	kórakose	típote	ísoe	kámi	tu
	the	crow	nothing	can.aor.ind.3sg	make.inf	the.gen
	kaććavendúlu
	carrion crow
	‘The crow could not do anything to the carrion crow’
	‘Il corvo non potè far nulla alla cornacchia’ (Roccaforte, TNC 21.19)

(13.b)	O	kaććavéndulose	apétae	spilà će	δen	du
	The	carrion crow	fly.aor.ind.3sg	high and	neg	him.gen
		ékame	típote
	Ø	make.ind.aor.3sg	nothing
	‘The carrion crow flew high and he could not do anything to him’
	‘La cornacchia volò in alto e quello non poté fargli nulla’ (Roccaforte, TNC 21.19)

I would like to propose that the absence of the modal is best interpreted as the consequence of a(n albeit residual) constraint, which originally prevented modal and functional verbs from surfacing in finite na-clauses.

As example (13.b) shows, the replacement of the monoclausal structure in (13.a) with only the finite verb produces a gap at the semantic, and even syntactic level, due to the absence of the functional auxiliary, which can only be recovered from the context. It is not by chance that most occurrences of these silent modal (and other functional) verbs are represented by interrogative, exclamative, negative, conditional and future-oriented sentences (see exx. 14–17), where the modal (deontic) reading originates from unrealized, potential or other counterfactual events5.

Such a gap mainly affects those predicates that (in an overt or covert form) preserve the infinitive more than others; in contrast, verbs that no longer take the infinitive only rarely surface as null auxiliaries, if at all.

To test the occurrences of finite verbs used to replace restructuring constructions, I selected two different written corpora: the short stories and fairy tales collected by Rossi Taibbi and Caracausi ([1959] 1994) from the villages of Roccaforte (pp. 10–266), Condofuri (pp. 277–78), Roghudi (pp. 300–6) and Bova (pp. 394–418), and Aesop’s fables translated into Greko by Crupi (2021).

In a survey of the corpora, I recorded the optional absence of CAN, KNOW, and (very rarely) MUST as the main verbs in complex finite structures, alongside the overwhelming majority of cases in which they are overtly realized, followed by na-clauses (as regards MAKE, see Section 3.1). Then, with the exception of MUST (see below), these very auxiliaries are the same which still partially retain infinitival complementation in Greko, though in alternation with an irrealis finite clause (see § 1)6.

Conversely, those predicates which over time favoured the total replacement of infinitival clauses with finite ones (LET, WANT, COME, GO), surface exclusively as overt heads of na-finite clauses:

• CAN

(14.a)	Ma	púpote	tin		ívre
	but	anywhere	her	Ø	find.ind.aor.3sg
	‘But he couldn’t find her anywhere’
	‘Ma non la poté trovare in nessun luogo’ (Roccaforte, TNC 11.10)

(14.b)	Ti	egò	δen	dim		bérro	meθému apikátu χumátu
	because	me.nom	neg	her	Ø	take.1sg	with=me under ground
	‘Because I cannot take her with me underground’
	‘Perché io non la posso portare con me sotto terra’ (Roccaforte, TNC 12.3)

(14.c)	Óδe m’éstile				o	téftose	principínose,	ti
	here me=sent.ind.aor.3sg				the	such	little prince	because
	θéli,	an		éχi		tósso onúri	na	tu
	want.3sg	if	Ø	have.3sg		much honour	comp.sbjv	him.gen
	δóite	ti	rigginótta		ya yinéka
	give.2pl	the	big queen		as wife
	‘The little prince sent me here, because he hopes that you will give him the queen as a wife, if he may have the honour’
	‘[…] se può avere tanto onore […]’ (Roccaforte, TNC 12.10)

(14.d)	An		éχo	tósson onúri,	eláste	sto
	if	Ø	have.1sg	much honour	come.2pl	to=the
	spitímmu
	house=me.gen
	‘If I may be so honoured, please come to my house’
	‘Se posso avere tanto onore, venite a casa mia’ (Roccaforte, TNC 40.21)

(14.e)	Ka	póse		fégwome […]?
	comp	how	Ø	escape.1pl
	‘How can we escape?’
	‘Come possiamo fuggire […]?’ (Roccaforte, TNC 14.7)

(14.f)	Aš’ óla		ton ebúrlespa,		ma	me	túndim
	all the times		him fool.ind.aor.1sg		but	with	this.acc.f
	máttsa	δen		arriparégwo
	club	neg	Ø	remedy.ind.prs.1sg
	‘I managed to fool him every time. As far as this club is concerned, I cannot find a solution’
	‘[…], ma per questa mazza non posso rimediare […]’ (Roccaforte, TNC 27.16)

(14.g)	ca	pion	ammialó		iche	tundo	nimali!
	comp	what	brain	Ø	have.impf.ind.3sg	this	animal
	‘What brain this animal could have had!’
	‘Che quale cervello poteva avere questo animale!’ (“Glossa”, 61)

• KNOW

(15.a)	Egò	éχo	mía	lleḍ̣ḍ̣̣á […]	pu	šéri	kámi túnda prámata.
	I	have	a	sister	who	know.3sg	do.inf these things
	Ta			kánni esú?
	them		Ø	make.2sg	you
	[…]	Dóppu ti tróvume,			pettónnome		na	ívro
		After	that eat.1pl		let’s.go.1pl		compsbjv	see.1sg
		an da		kámise
		if them	Ø	make.2sg
	‘I have a sister who knows how to do these things. Can you do them? […]
	After we eat, let’s go and see if you can’
	‘Io ho una sorella […] che sa fare queste cose. Sai farle tu? […] Dopo che mangiamo, andiamo a vedere se sai farle’ (Roccaforte, TNC 8.11)7

(15.b)	Árte θoríse	ećíno	pu	su		kánno
	now see.ind.prs.2sg	that	rel	you	Ø	make.1sg
	egó
	me.nom
	‘Now you’ll see what I can do’
	‘Ora vedrai che so fare io’ (Roccaforte, TNC 26.12)

(15.c)	ta	emáθiese	pu	ekrúnnai		ólese	tes
	them	instruct.ind.aor.3sg	so that	Ø play.ind.impf.3pl		all	the
	sonátese	pu ekánnai		me	ton	gózmo
	songs	rel make.ind.impf.3pl		with	the	world
	‘He instructed them so they learned how to play all the songs in the world’ (Roccaforte, TNC 36.2)

• MUST

(16.a)	Će o ríga		tu	ípe	ti	i	θiχatératu
	and the king		him.gen	tell. ind.aor.3sg	comp	the	daughter=him.gen
		éne	i	yinékatu
	Ø	be.prs.ind.3sg	the	wife=him.gen
	‘The king told him that his daughter was to be his wife’
	‘E il re gli disse che la figlia doveva essere sua moglie’ (Roccaforte, TNC 25.19)

(16.b)	Egò		éχi	tósso	pu patégwo	ya esséna,
	me.nom		have.3sg	so long	rel suffer.1sg	for you
	će	árte		aḍḍizmonáo	ya esséna?
	and	now	Ø	forget.1sg	for you
	‘I have suffered for you for so long, and now I am supposed to forget you?’
	‘È tanto tempo che soffro per te ed ora (dovrei) dimenticarti?’ (Roccaforte, TNC 9.24)

(16.c)	će	épyae	tóssa	ta	δinérya
	and	make.ind.aor.3sg	so much	the.pl	money.pl
	pu	ta		épire			me	éna	tráino
	rel	them.pl	Ø	take.ind.aor.3sg			with	a	pulling
	‘And he made so much money that he had to haul it’
	‘Ed egli ricavò tanti denari che dovette portarli con un traino’ (Roccaforte, TNC 11.14)

In this scenario, the (albeit rare) occurrences of silent MUST would seem to contradict the hypothesis presented here. Indeed, the almost mandatory association of MUST with finite complementation is already observed in the texts collected in Rossi Taibbi and Caracausi ([1959] 1994) (=TNC), in comparison with other auxiliaries such as, among others, HEAR and MAKE, which still overwhelmingly occur with the infinitive (Ledgeway 2013). Indeed, a constraint on MUST surfacing in finite structures would not be expected.

However, that this predicate presents behaviour different from that of CAN can be found in the following evidence. In very few instances, an overt form of MUST replaces CAN, with the meaning of the latter, followed by a na-clause:

(17.a)	éχi	na	mu	kámise	áše	módo	pu	éχo
	must.2sg	comp.sbjv	it	make.2sg	of	such a way	comp must.1sg
	na	mbéo	ećóssu
	comp.sbjv	go.1sg	inside
	‘You have to build it in such a way that I can pass through it’
	‘[…] e devi farlo in modo che io possa entrarvi dentro’ (Roccaforte, TNC 19.14)

(17.b)	ma egò		tósso	éχo	na	porpatío	pu
	but me.nom		far enough	must.1sg	comp.sbjv	walk.1sg	comp
	éχo	na	tin	drovéspo
	must.1sg	comp.sbjv	her	find.1sg
	‘But I have to walk a great deal in order to be able to find her’
	‘Ma io devo camminare tanto da poterla trovare’ (Roccaforte, TNC 33.22)

(17.c)	pemu		po	ccrázese	t’	echo	na
	tell.2sg=me		how	croak.2sg	so that	must.1sg	comp.sbjv
	tos	ipo		oló […]
	them.acc	tell.1sg		all
	‘Tell me how you croak so that I can tell everyone […]’
	‘Dimmi come gracchi affinché io possa dire a tutti […]’ (“Glossa”, 7)

A reviewer rightly observes that examples (17.a–c) all involve the use of multiple modals: there is always a modal element in the matrix clause (specifically, an obligation modal in (17.a) and (17.b) and an imperative in (17.c)), followed by éχo na in the subordinate clause. The obligation/imperative modality in the higher clause could have triggered the selection of MUST in lieu of CAN in the lower clause, in a process of analogical modal replacement.

A different possibility lies in the restriction on licensing CAN—the true stronghold of infinitival complementation—as an overt functional auxiliary in finite complementation. Under this view, MUST—insofar as it is systematically associated with finite complementation—is allowed to surface as an overt modal, despite its deontic meaning. Thus, although MUST still shows very few remnants of its aversion to finite complementation—on a par with those auxiliaries that partially preserve infinitival complementation—it differs from these insofar as it behaves as a suppletive modal form in place of the operator CAN. As such, it shows a greater predisposition to licensing finite dependent clauses, whereas CAN is evidently constrained (and barred) due to its near-categorical association with the infinitive.

Crucial evidence in favour of the presence of null modal predicates in (14–16) is provided by the almost-minimal pairs in the examples below (18.a,b, repeated from 13.a,b). Indeed—although in both sentences the same possibility operator is required—only in the former is CAN overt, followed by the infinitive of the lexical verb ‘make’. In the latter CAN is silent, and the lexical verb ‘make’ (ékame) surfaces in the finite form:

(18.a)	o	kórakose	típote	ísoe	kámi	tu
	the	crow	nothing	can.ind.aor.3sg	make.inf	the.gen
	kaććavendúlu
	carrion crow
	‘The crow could not do anything to the carrion crow’
	‘Il corvo non poté far nulla alla cornacchia’ (Roccaforte, TNC 21.19)

(18.b)	O	kaććavéndulose	apétae	spilà će	δen	du
	The	carrion crow	fly.aor.ind.3sg	high and	neg	him.gen
	ékame		típote
	make.ind.aor.3sg		nothing
	‘The carrion crow flew high and he could not do anything to him’
	‘La cornacchia volò in alto e quello non poté fargli nulla’ (Roccaforte, TNC 21.19)

3.1. The Case of MAKE

The case of MAKE seems quite different. Indeed, the contexts that are otherwise predicted to trigger the recovery of silent causative heads generally present intransitive motion predicates as finite verbs. In the corpora under examination, these predicates include prototypical motion verbs, such as pettónno ‘ascend’, or other semantically related verbs, such as apetáo ‘fly’ and spuntéggwo, spundégwo ‘sprout’:

(19.a)	Túnda peδía […]		espundéspai	éna	ćentinari će	mía
	These children	Ø	sprout.ind.aor.3pl	a	hundred and	one
	llaranghía
	oranges
	‘These children grew a hundred and one oranges’ (lit. ‘made them sprout’)
	‘Questi bambini […] fecero spuntare cento ed un arancio’ (Roccaforte, TNC 18.11)

(19.b)	[…] i	θilići		espúndespe	éna	póδi
	[…] the	female	Ø	sprout.ind.aor.3sg	a	plant
	milinǵána
	aubergine
	‘The female grew an aubergine plant’ (lit. ‘made it sprout’)
	‘[…] la femmina fece spuntare una pianta di melanzana (Roccaforte, TNC 18.15)

(19.c)	éχome	túndo	piććuni će	t’	apetúme
	have.1pl	this.acc	pidgeon and	it.acc	fly.1pl
	‘We have this pigeon and we’re making it fly’
	‘Abbiamo questo piccione e lo facciamo volare’ (Roccaforte, TNC 40.20)

(19.d)	Póse	apetásai	to	piććúni […]
	when	fly.ind.aor.3pl	the	pigeon
	‘When they made the pigeon fly’
	‘Quando fecero volare il piccione’ (TNC 40.20)

(19.e)	pósso	ívrai	ti	pettónnusi	ećindim
	and	see.ind.aor.3pl	comp	ascend.ind.prs.3pl	that
	máňi	ǵúveno
	beautiful	woman
	‘And they saw that they were inviting that beautiful young woman to come inside’
	‘Ed ecco che videro che facevano salire quella bella giovane’ (Roccaforte, TNC 38.9)

(19.f)	epyásai	ećíndin	ghinéka,	tin	pettóai	apánu
	take.ind.impf.3pl	that	woman	her	ascend.ind.impf.3pl	over
	‘They took the woman and brought her upstairs’
	‘Presero quella donna, la portarono sopra’ (Roccaforte, TNC 19.13)

Causative constructions codified by overt MAKE + na + motion predicate (or a pseudo-coordinated finite structure, as in 20.a), are attested only sporadically:

(20.a)	[…]	se	kánni	će	pettónnise	apánu
		you.acc	make.3sg	and	ascend.2sg	above
	‘He will bring you upstairs’ (=‘He will invite you inside’)
	‘[…] ti farà salire sopra’ (Roccaforte, TNC 25.30)

(20.b)	Kámeto	na	to	pettói	apánu
	make.impv.2sg	comp.sbjv	it	climb.3sg	over
	‘Have them take it (= the mirror) upstairs!’
	‘Fai che lo salgano sopra (scil. lo specchio)’ (Roccaforte, TNC 19.15)

In some cases (21.a,c), the causative construction with overt MAKE is attested in free variation with the construction presenting only the finite verb (21.b,d):

(21.a)	Káme	na	pettói	apánu
	make.impv.2sg	comp.sbjv	climb.3sg	over
	‘Have her come upstairs!’
	‘Falla salire sopra’ (Roccaforte, TNC 41.9)

(21.b)	Dóppu	ti	tin	epéttoe	apánu […]
	after	comp	her.acc	climb.ind.aor.3sg	over
	‘After that he made her climb over […]’
	‘Dopo che la fece salire sopra […]’ (Roccaforte, TNC 41.9)

(21.c)	Pái	će	kánni	kámi	tría	míla
	go.ind.prs.3sg	and	make.prs.ind.3sg make.inf		three	apples
	‘He is going to have three apples made’
	‘Va a farsi fare tre pomi’ (Roccaforte, TNC 36.24)

(21.d)	Će	ékame	tría	mila	áše	χrisáfi
	and	make.ind.aor.3sg	three	apples	of	gold
	‘He had three golden apples made (= built/fabricated)’
	‘E fece fare tre pomi d’oro’ (Roccaforte, TNC 36.21)

The examples (19.a–f) show the transitive use of intransitive motion predicates with causative value. Although Greek varieties generally do not present this syntactic pattern, it is nevertheless attested in Italo-Greek and southern Italo-Romance (Ledgeway 2009, pp. 850–52)8.

It is highly probable that the transitive use of these verbs in Greko was triggered by interference with the Romance varieties of southern Calabria9. This change (i.e., the intransitive > transitive shift) may be connected with the resistance of MAKE to surfacing with finite complements, where the latter is replaced by the former. That is, newly transitivized intransitives (arising from contact-induced change) substituted the original causative structure encoded by the complex sentence MAKE + na-clause.

The few silent functional verbs in the corpora suggests that this represent a very conservative pattern, in which, at least in an earlier stage, modal and other functional predicates underwent restructuring and were prevented from surfacing in finite completive structures. In such contexts, only the lexical (embedded) verb can occur, ultimately replacing the complex sentence AUX + irrealis COMP + embedded finite predicate.

In this respect, the cases illustrated in the examples above could represent the last remnant of a diachronic stage in which some predicates could act exclusively as functional heads in monoclausal structures, whence the ban on surfacing as auxiliaries heading na-structures.

One may wonder, in these examples featuring silent modals, whether the modal meaning (as well as other functional values) is instantiated through the raising of the lexical verb into the I-domain, where it realizes its inflectional features. Indeed, in finite clauses, the verb phrase can raise into the Inflectional domain:

[Infl [VP]] > [Infl VP [VP]]

A well-known example is represented by the epistemic use of the future in substandard Catalan, which can codify supposition even in the absence of the modal auxiliary; see, e.g., (colloquial) Cat. Tindrá raó ‘He must be right’ (Cruschina and Ledgeway 2016, p. 560). However, lexical verbs involved in the Greko examples do not display special TAM features, as the future tense in Catalan. Instead of verb movement to the IP, I argue that the Greko examples present a “modality gap”, likely triggered by the once mandatory union of functional verbs with infinitival complements in restructuring contexts.

4. How to Explain the Gap: Some Conclusions

In the corpus analyzed, some occurrences present the replacement of monoclausal structures with the lexical verb only, in its finite form (corresponding to the embedded verb in the infinitival complement), instead of a complex structure consisting of the functional predicate followed by a na-clause. Such a replacement is probably linked to the spread of finite subjunctive complements at the expense of infinitival ones. When finite na-clauses have been generalized (though not exclusively) to the detriment of infinitival complementation, the auxiliaries that prove more resistant to the use of finite complementation can remain silent, hence the optional replacement of complex restructuring structures by means of a simplex clause, made up of only the lexical finite verb (instead of the complex sentence AUX + na + finite verb).

The constraint—though anything but categorical—of these verbs on selecting na-clauses probably represents the last remnant of a diachronic stage in which some predicates (in our corpus, mainly modal and causative verbs) began to leave their previous restructuring contexts.

As a result, monoclausality is lost, giving rise to several phenomena all pointing to a decreased level of dependency between the matrix and embedded clauses. These include the cliticization of the pronoun to the dependent verb—without the option of climbing to the main predicate—in addition to the possibility of the embedded verb licensing a subject. This may have led modal and other functional predicates to acquire a greater lexical status, partially freeing themselves from their otherwise (exclusively) functional role. Should this be the case, their resistance to surfacing in finite structures could represent a step in the degrammaticalization process, which leads them to progressively realize more lexical features, and at the same time reduces their ability to act (exclusively) as functional heads. Their (albeit sporadic) absence in selecting finite complements could be a signal of functional instability, in which they still oscillate between a purely functional role and a lexical one.

This view is further supported by the finding that some of the predicates that select for na-clause over infinitival complements show a lexical value that is missing in restructuring contexts. This applies to CAN. Although rare, several occurences of CAN followed by a na-clause are attested in the oral texts collected by Rossi Taibbi and Caracausi ([1959] 1994). In these cases, CAN takes on a series of meanings that diverge from its prototypically modal value, e.g., ‘to be (physically) able to, to have the physical strength to, etc.’, by encoding an ability (or, in negative sentences, an inability), which depends on the subject (De Angelis 2013a).

The following examples are particularly revealing:

(22.a)	Egò	δe	ssónno	pléo	[na]	staθò	arrássu
	and	neg	can.1sg	(any)more comp.sbjv		stay.1sg	away
	‘I can no longer stay away’ (Rochudi, TNC 281.1)

(22.b)	[…]	ećinose	δen do	ísonne	mánko	na	to
	and	he	neg it	can.ind.impf.3sg	even	comp.sbjv	it
	movéspi
	move.3sg
	‘He could not even move it (= the stone)’ (Roccaforte, TNC 27.8)

(22.c)	O póverose prevíterose δen			ísonne		mánko	δéka	rótula,
	the poor	priest	neg	carry.ind.impf.3sg		even	ten	rolls
	će ísonne		na	féri	ećindo	šiní yomáto?
	and can.ind.impf.3sg		comp.sbjv	carry.3sg	that.n	rope full.n (of wood)
	‘The poor priest could not even carry ten rolls, and you think he could he carry that rope (with the wood)?’ (Roccaforte, TNC 27.11)

(22.d)	épyae	ta	áše metáši	će	tá	vale	óssu
	take.ind.aor.3sg	the	clothes of silk	and	them	put.ind.aor.3sg	inside
	sto vrastári.
	in=the boiler
	Poi δen	ésonne		na	ta	gwálise óšu
	then neg	can.ind.impf.3sg comp.sbjv		them take.3sg out
	‘He had taken the silk clothes and put them inside the boiler, but then he was not able to take them back out’ (Roccaforte, TNC 2.5)

In almost all these examples—as revealed by the different contexts they represent—in negative sentences, CAN is used to convey an impossibility arising from constraints inherent to the state of the subject (psychological, as in 22.a, or physical, as in 22.b–d). The use of the first occurrence of ísonne in (22.c), with the value of ‘carry’, is noteworthy.

It should be noted that this also applies to Griko, where CAN may select for na-clauses with the same ability reading, see e.g., en sòzo na pao ‘I don’t have the strength to go’ (Castrignano [Lecce]) (Baldissera 2013, pp. 57–58)10.

In conclusion, not only does Greko prove to be more refractory in losing infinitival complementation in comparison with Griko and with the southern Romance varieties affected by the Greek sub-/adstrate (Ledgeway 2013), but it also displays a major deficiency, albeit at a very residual level, in the realization of verbal modality (as well as other functional values) in finite complementation. This state of affairs should be taken into account when the spread of finite subjunctive complementation is investigated. When infinitival complementation is partially preserved, its stronghold, i.e., restructuring contexts, can produce noteworthy effects (i.e., modal and functional deficiencies in the verbal system) in the finite complementation system, too.