Fortune and Decay of Lexical Expletives in Germanic and Romance along the Adige River

Abstract

Lexical expletives can be divided into two main classes: (i) CP expletives required by the V2 constraint and, hence, by the necessity to lexicalize the position on the left of the inflected verb and (ii) TP expletives connected with the negative value of the pro-drop parameter and, therefore, with the necessity to lexicalize the ’structural‘ subject position, specifically, [Spec, TP]. The latter can, in turn, be divided into two subclasses: impersonal subjects and positional expletives, which occur with postverbal/low subjects and extraposed subject clauses. While CP expletives only appear in Germanic varieties that maintain V2, the subclassification of TP expletives yields interesting results when comparing Cimbrian and the Venetan varieties in Nord-East Italy, where the gradual disappearance of the positional expletive in free inversion structures and the residual maintenance of impersonal subjects from North to South along the Adige River confirms the distinction between two classes of subject expletives; furthermore, the resilience of impersonal subjects and their distribution in the northwestern part of the area under consideration sheds light on the role of language contact which is confirmed along the same axis—but crucially in the opposite direction—by the increasing employment of cleft constructions in WH-clauses replacing enclisis (i.e.,: pronominal subject inversion with the finite verb).

1. Introduction

Since the principles-and-parameters approach was introduced in the generative grammar framework (see Chomsky 1981, 1982, 1986), two descriptive correlations have never been challenged1: (a) the negative value of the null subject parameter (henceforth NSP) correlates with the occurrence of lexical expletives; (b) finite verb movement in the C-domain correlates with subject inversion (Vfin-Subj). A third correlation, originally introduced by Hulk and van Kemenade (1995) and recently revisited by Bidese and Tomaselli (2018), Tomaselli and Bidese (2019), and Bidese et al. (2020), maintains that V2 is incompatible with a positive value of the NSP, that is, a specific type of V2 (mandatory Vfin-to-C movement in the root context) is incompatible with a specific type of pro-drop (referential null subject of the third person singular).2 In this paper, we will provide further arguments to confirm the robustness of these assumptions based on the syntax of lexical expletives in two groups of languages in the northeastern part of Italy: a (still) V2 Germanic variety such as Cimbrian3 and the Italian dialectal varieties spoken in Trentino and in the western part of Veneto (the provinces of Verona and Vicenza), on the one hand, in comparison with the eastern Venetian varieties spoken in the provinces of Treviso, Padova, and Venice, on the other hand.

This paper is organized into four sections. The first part (Section 2) offers a description and analysis of the syntax of expletives in Cimbrian. In the second section (Section 3), we revise the data from Trentino (from the seminal work by Brandi and Cordin (1981, 1989) to incorporate recent refinements proposed by Cordin (2021), based on the data collected by VinKo4). Next, the third section (Section 4) explores the data on Venetan dialects that are not on the linguistic Germanic–Romance border. The last section (Section 5) compares the data discussed in the previous sections which show a decreasing rate of expletive realization along the Adige River (from NW to SE) with the different rates of verb–subject inversion in interrogatives (i.e., the maintenance of enclisis to the finite verb). The comparison of these two aspects of pronominal syntax leads us to propose a double conclusion. First, the phenomenon of pronominal subject inversion (enclisis to the finite verb in C) in the root interrogative context is fully maintained in the Trentino dialects, possibly due to the pressure of the Germanic varieties on the border which robustly show both expletive realization and generalized verb–subject inversion in the root clause (not only in WH contexts). Second, the maintenance of V2 in Cimbrian preserves this variety from a resetting of the NSP, as it reinforces the occurrence of lexical expletives instead. These two distinct developmental lines strongly support the idea that the role of language contact is limited to the speed of a given language change determined by the possible choices of the inherent system of the language (see Padovan et al. 2016).

2. The Typology of Lexical Expletives in Cimbrian

Cimbrian is characterized by two classes of lexical expletives that correlate with both the negative value of the NSP and the V2 phenomenon (see Bidese and Tomaselli 2018 for details):

(i)

′z/-z, which corresponds to German es (English it);

(ii)

-da, which approximately corresponds to English there.

Even if the lexical forms etymologically coincide (es = it; da = tho/there), their syntax is clearly divergent if required by either one (pro-drop) or the other (V2) syntactic phenomenon.

2.1. Expletives Required by the Negative Value of the NSP

2.1.1. The Expletive ′z/-z

In connection to subject syntax, the lexical expletive ‘z obligatorily occurs on either the left (1a) or the right of the finite verbal element (1b) in the root declarative sentence. Thus, it behaves quite similar to its German counterpart es (cf. 2):

(1)	a.	′Z	snaibet	haüt	atz	Lusérn.
		it	snows	today	at	Luserna
	b.	Haüt	snaibet=z	atz	Lusérn.
		today	snows=it.cl	at	Luserna
		‘It is snowing today in Luserna.’

(2)	a.	Es	schneit	heute	in	Lusérn.
		it	snows	today	at	Luserna
	b.	Heute	schneit	es	in	Lusérn.
		today	snows	it	at	Luserna
		‘It is snowing today in Luserna.’

Furthermore, this expletive occurs to the right of the lexical complementizer in the embedded clause, for instance, bal ‘when’ (cf. 3), az ‘that’, or be ‘if’ (cf. 4):5

(3)	a.	Bal=z	snaibet	starch	atz Lusérn,		stea=bar	dahuam.
		when=it.cl	snows	hard	at	Luserna,	stay=we.cl	at=home
		‘When it snows heavily in Luserna, we stay home.’
	b.	I sperar,	az=(z)6	snaibe	starch		haüt	atz Lusérn.
		I hope	that=it.cl	snow.conj	hard		today	at	Luserna
		‘I hope that it will snow heavily today in Luserna.’
	c.	I vors=mar,	be=z	snaibet	starch		haüt	atz Lusérn.
		I ask=me.dat.cl	if=it.cl	snows	hard		today	at	Luserna
		‘I’m wondering if it is snowing heavily today in Luserna.’

These examples reveal the same structural position as in German. Compare (3a–b) with (4a–b) modulo OV and the position of the adverb ‘heavily’:

(4)	a.	Wenn es	in			Lusérn	stark	schneit,	bleiben wir	zu Hause.
		when=it.cl	in			Luserna	hard	snows,	remain we	at	home
		‘When it snows heavily in Luserna, we remain home.’
	b.	Ich	hoffe, dass			es	heute	in Lusern	stark	schneit.
		I	hope	that		it	today	in Luserna	hard	snows
		‘I hope that it will snow heavily today in Luserna.’
	c.	Ich	frage	mich, ob		es	heute	in Lusern	stark	schneit.
		I	ask	me	if	it	today	in Luserna	hard	snows
		‘I’m wondering if it is snowing heavily today in Luserna.’

Lastly, the lexical expletive -z obligatorily occurs to the right of the finite verb in the direct interrogative context (cf. 5), as in the German equivalent sentence examples (cf. 6):

(5)	a.	Snaibet=z	haüt	atz Lusérn?
		Snows=it.cl	today	at	Luserna
		‘Is it snowing at Lusern today?’
	b.	Bo	snaibet=z	haüt?
		where	snows=it.cl	today
		‘Where is it snowing today?’

(6)	a.	Schneit	es	heute	in Lusérn?
		Snows	it	today	in	Luserna
		‘Is it snowing in Lusern today?’
	b.	Wo	schneit	es	heute?
		where	snows	it	today
		‘Where is it snowing today?’

The same lexical item occurs with the same distributional pattern in co-occurrence with subject clauses (cf. 7) as happens in German (cf. 8):

(7)	a.	‘Z	hatt=en	gevallt,	[az=ta	dar Luca	sai(be)	khent	afn	vairta]
		it	has=him.cl	pleased	that=da	the Luca	is.conj	come	to=the	party
		‘He was pleased that Luca came to the party.’
	b.	Gestarn	hatt=z=en	gevallt,	[az=ta	dar Luca	sai(be)	khent	afn	vairta]
		Yesterday	has=it=him.cl	pleased	that=da	the Luca	is.conj	come	to=the	party
		‘Yesterday, he was pleased that Luca came to the party.’
	c.	Hatt=z=en	gevallt	gestarn,	[az=ta	dar Luca	sai(be)	khent	afn	vairta]?
		has=it=him.cl	pleased	yesterday	that=da	the Luca	is.conj	come	to=the	party
		‘Yesterday, did he like yesterday that Luca came to the party?’

(8)

a.

Es

hat

Hans

gefallen,

[dass

Lukas

zum

Fest

gekommen

ist]

it

has

Hans

pleased

that

Lukas

to=the

party

come

is

‘Hans was pleased that Lukas came to the party.’

b.

Gestern

hat

(es)

Hans

gefallen,

[dass

Lukas

zum

Fest

gekommen

ist]

yesterday

has

it

Hans

pleased

that

Lukas

to=the

party

come

is

‘Yesterday, Hans was pleased that Lukas came to the party.’

c.

Hat

(es)

Hans

gestern

gefallen,

[dass

Lukas

zum

Fest

gekommen

ist]?

Has

it

Hans

yesterday

pleased

that

Lukas

to=the

party

come

is

‘Was Hans pleased yesterday that Lukas came to the party?’

When the subject clause is topicalized before the finite verb, there is, of course, no possibility for the lexical expletives ‘z/-z and es to show up in either Cimbrian (cf. 9) or German (cf. 10):

(9)	[Az=ta	dar Luca	sai(be)	khent	afn	vairta],	hatt=(*z)=eni	gevallt	in	Hånsi.
	that=da	the Luca	is	come	to=the	party	has=(it)=him.cl	pleased	the.dat	John
	‘That Luke came to the party pleased Hans.’

(10)	[Dass	Lukas	zum	Fest	gekommen	ist],	hat	(*es)	Hans	gefallen.
	that	Lukas	to=the	party	come	is	has	(it)	Hans	pleased
	‘That Luke came to the party pleased Hans.’

2.1.2. The Expletive -da

The expletive particle -da is excluded in all the preceding contexts, and it will never play the role of an impersonal subject. It is always realized as an enclitic form to either the finite verb in C/Fin or the lexical complementizer bal, be, or az. The occurrence of the enclitic particle -da is required whenever the nominal subject is not fronted to the left of the finite verb (compare 11a with 11b–d):

(11)	a.	Moi nono	iz		khent		atz Lusérn	haüt.
		my grandpa	is		come		to Luserna	today
		‘My grandpa came to Luserna today.’
	b.	Haüt	iz=ta		khent		atz Lusérn	moi nono.
		today	is=da		come		to Luserna	my grandpa
		‘Today, my grandpa came to Luserna.’
	c.	I sperar,	az=ta	sai(be)	khent		atz Lusérn	moi nono	haüt.
		I hope,	that=da	is	come		in Luserna	my grandpa	today
		‘I hope that my grandpa came to Luserna today.’
	d.	Iz=ta	khent		atz Lusérn		moi nono	haüt?
		Is=da	come		in Luserna		my grandpa	today
		‘Did my grandpa come to Luserna today?’

The German translation of the preceding examples reveals that Cimbrian -da does not find a lexical counterpart in Standard German (cf. 12):

(12)	a.	Mein Großvater	ist	(*es)	heute		nach Lusérn	gekommen.
		my grandpa	is	(it)	today		to Luserna	come
		‘My grandpa came to Luserna today.’
	b.	Heute	ist	(*es)	nach		Lusérn	mein Großvater	gekommen.
		today	is	(it)	to		Luserna	my grandpa	come
		‘Today, my grandpa came to Luserna.’
	c.	Ich hoffe,	dass	(*es)	heute	nach	Lusérn	mein Großvater	gekommen	ist.
		I hope,	that	(it)	today	to	Luserna	my grandpa	come	is
		‘I hope that my grandpa came to Luserna today.’
	d.	Ist		(*es)	heute	nach	Lusérn	mein Großvater	gekommen?
		is		(it)	today	to	Luserna	my grandpa	come
		‘Did my grandpa come to Luserna today?’

The expletive particle -da partially corresponds to English there in subject inversion constructions, as in the following:

(13)	a.	There arrives a man.
	b.	I know that there lived a lonely old man in the forest.

Two main differences can be seen between Cimbrian -da and English there: (i) -da is not sensible to the definiteness effect as in English (cf. 14a with 14b), and (ii) it is not limited to a specific verbal class (i.e., unaccusative verbs) (cf. 15a with 15b). The following comparison of the same sentences in English and in Cimbrian demonstrates the two main differences:

(14)	a.	*There arrives Mary today		(definiteness effect)
	b.	okHaüt	khint=(t)a	di Maria.
		today	comes=da	the Mary
		‘Mary is arriving today.’

(15)	a.	*There has read a book my mother/my mother a book			(transitive verbs)
	b.	okHaüt	hatt=(t)a	gelest	an libar	moi mamma
		today	has=da	read	a book	my mother
		‘Today, my mother has read a book.’

The occurrence of -da represents one of the most peculiar aspects of Cimbrian subject syntax; it is used for nominative case assignment (NOM) whenever the nominal subject is not raised to the structural subject position (i.e., the subject does not occur on the left of the finite verb), which is assumed to be the specifier of the lowest CP projection (i.e., [Spec, Fin]):

(16)

The basic idea proposed by Bidese et al. (2020) and represented in (16) above is that C/Fin is responsible for case assignment in both German and Cimbrian (which are both V2 languages) but that only in Cimbrian Fin provides NOM on the left (i.e., to its specifier position [Spec, Fin], much like Italian T).

2.2. V2 Expletives

Cimbrian is considered a structural V2 language that has lost the well-known linear restriction characterizing all other Germanic languages with the exception of Modern English. This means that Cimbrian is characterized by mandatory finite V-to-C movement in a Split-CP configuration. Therefore, the following word order pattern (cf. 17a) should be analyzed as proposed in (17b) (see Bidese and Tomaselli 2016, pp. 69–71):

(17)	a.	Haüt	in	balt	hatt=ar	gesek	in	has.
		today	in=the	wood	has=he.cl	seen	the.acc	rabbit
		‘Today, he saw the rabbit in the wood.’
	b.	[CP [Haüt] [in balt] [FinP hatt=ar …. [VP gesek [dp in has]]]]

If the position on the left of the finite verb is not lexicalized by at least one constituent, either the subject or a complement, the sentence becomes grammatical only as a Yes/No question (cf. 18a for Cimbrian and 18b for German):

(18)	a.	Hatt=ar		gesek	in	has	haüt	in	balt?
		has=he.cl		seen	the.acc rabbit		today	in=the	wood
		‘Did he see the rabbit in the wood today?’
	b.	Hat	er	heute	im Wald		den Hasen		gesehen?
		has=he.cl		today	in=the wood		the.acc rabbit		seen
		‘Did he see the rabbit in the wood today?’

In order to maintain the declarative/assertive modality, a V2 positional expletive can occur in the first position on the left of the finite verb in the CP domain (cf. 19a for Cimbrian and 19b for German):

(19)	a.	‘Z	hatt=(t)a	gesek	in	has	dar djunge	in	balt	haüt.
		it	has=da	seen	the.acc rabbit		the boy	in=the	wood	heute
		‘Today, the boy saw the rabbit in the wood.’
	a.	Es	hat	der Junge		heute	im Wald	den Hasen	gesehen.
		it	has	the boy		today	in=the	the.acc rabbit	seen
		‘Today, the boy saw the rabbit in the wood.’

V2 positional expletive ‘z (German es) lexically corresponds to the subject expletive ‘z, with two main differences:

(i)

The morphological realization: it only occurs on the left of the finite verb (i.e., it refuses enclisis);

(ii)

The co-occurrence with enclitic -da: since ’z implies a postverbal (i.e., not raised subject), the enclitic particle -da is always required (cf. 20):

(20)	‘Z	laütan=da	di klokkn.
	it	ring.pl=da	the bells
	‘The bells are ringing.’

V2 expletive ‘z should not be confused with locative da (cf. also Grewendorf and Poletto 2011: 315, footnote 8), which satisfies the V2 constraint as well as other unaccented, monosyllabic adverbials (for instance, est ‘now’) in a presentative construction (cf. 21).

(21)	a.	Da	laütan=da	di klokkn.
		there	ring.pl=da	the bells
		‘There the bells are ringing.’
	b.	Da	hatt=(d)a	gesek	in	has	dar	djunge.
		there	has=da	seen	the.acc hare		the.nom	boy
		‘There the boy saw the hare.’

This kind of adverb represents the only choice whenever the subject is realized by an enclitic pronoun (cf. 22 for Cimbrian):

(22)	a.	Da	laütan=sa.
		there	ring.pl=they.cl
		‘There they are ringing.’
	b.	Da	hatt=ar	gesek	in has	haüt	in	balt.
		there	has=he.cl	seen	the.acc hare	today	in-the	wood
		‘There he saw the hare in the wood today.’

This is also true for Standard German (cf. 23a) unless the postverbal pronominal subject is accented (cf. 23b):

(23)	a.	*Es	ist	er		angekommen.
		it	is	he		arrived
	b.	Es	ist	(auch) ER/DER		gekommen.
		it	is	also	he	arrived
		‘He also arrived.’

The syntax of lexical expletives represents a strong argument for the assumption that Cimbrian is characterized by both structural V2 (mandatory finite verb movement to C) and the negative value of the NSP. Hence, Cimbrian syntax reinforces the divide between COMP-dominant languages and INFL-dominant languages introduced by Hulk and van Kemenade (1995): V2 languages are COMP-dominant languages that do not allow referential null subjects; in contrast, null-subject languages are INFL-dominant languages, which do not allow V2.7

2.3. First Summary

The data discussed in the previous paragraphs are summarized in the following

Table 1. Typology of lexical expletives8.

		ENGLISH	CIMBRIAN	GERMAN	ITALIAN
(24)		it	‘z/-z	es	pro	(impersonal subjects)
	TP expletives
		there	-da	pro	pro	(with postverbal subjects)
	CP expletives	/	‘z	es	/	(V2 expletive)

(see Bidese and Tomaselli 2018):

The comparison between Cimbrian and German underlines the maintenance and even the reinforcement of lexical expletives in both impersonal constructions and non-subject initial V2 clauses. Moreover, the comparison with the Italian dialects of the northeastern region should provide us with new insight concerning the syntax of subject clitics. Although we would expect the absence of V2 expletives, we would not exclude a potential distinction between positional expletives that co-occur with postverbal subjects and subject clitics linked to impersonal verbs.

3. The Typology of Lexical Expletives in the Trentino Dialects

It is well known in the literature that expletive clitic subjects with weather verbs like nevegar/fiocar (‘to snow’) are still well attested in the Trentino dialects (see Casalicchio and Cordin 2020, p. 109: “Meteorological verbs [in Central Trentino] require a 3rd singular person subject clitic, which has no argumental value” and 264: “Semiargumental verbs, such as meteorological verbs, take an obligatory expletive subject that morphologically corresponds to a 3rd person masculine singular subject clitic”):

(25)	a.	El	piòve.			(Casalicchio and Cordin 2020, p. 109)
		it	rains
		‘It’s raining.’
	b.	L’a	nevegà	tuta	la	nòt.
		it.cl-has	snowed	all	the	night
		‘It snowed all night.’

A glance at the data recently collected by VinKo, a platform based on crowdsourcing (cf. footnote 4), strongly confirms this observation. The translation of the Italian sentence n. 136 in the VinKo platform (cf. here 26), as follows,

(26)	È	nevicato.
	is	snowed
	‘It snowed.’

has been recorded by 14 informants from 11 different locations in Trentino, with the following results:

Twelve maintained the subject clitic ‘l ‘it’ (third person singular masculine) independently of both lexical choice (nevegar versus fiocar, both ‘to snow’) and auxiliary selection:9

(27)	a.	L’a	fiocà.	(Lavis, S0136_tre_U0409)10
		it.cl-has	snowed
		‘It snowed.’
	b.	L’è	nevegà.	(Cinte Tesino, S0136_tre_U0411)
		it.cl-is	snowed
		‘It snowed.’
	c.	L’a	nevegà.	(Trento, S0136_tre_U0451)
		it.cl-has	snowed
		‘It snowed.’

Only two young informants translated the sentence without the subject clitic:11

(28)	a.	È	nevegà.	(Levico, S0136_tre_U0565)
		is	snowed
		‘It snowed.’
	b.	A	fiocà.	(Mori, S0136_tre_U0469)
		has	snowed
		‘It snowed.’

Following the seminal work by Brandi and Cordin (1981, 1989), we will consider subject clitics of the Trentino dialects (and, more generally, of the area we are focusing on) as the manifestation of subject agreement (see also Casalicchio and Cordin 2020, p. 258). From this perspective, the maintenance of expletive subject clitics with weather verbs corresponds to the lexical manifestation of expletive subject agreement of the third person singular.12

According to this point of view, it is interesting to underline two relevant points already discussed in the literature:

(i)

Inflected verbs in the third person singular always involve a clitic that doubles the argumental preverbal subject even when the last one was not left-dislocated (cf. 29) (see Brandi and Cordin 1989 and Casalicchio and Cordin 2020, p. 110: “Subject clitics must also be expressed when the preverbal subject is a noun phrase, or a pronoun”):13

(29)	a.	La	maestra de matematica	la		va	via.
		the	teacher of math	she		goes	away
		‘The math teacher is going away.’
	b.	*La	maestra de matematica	va		via.
		the	teacher of math	goes		away
	c.	La	maestra de matematica	l’	è	nada	via.
		the	teacher of math	she	is	gone	away
		‘The math teacher went away.’
	d.	*La	maestra de matematica	è		nada via.
		the	teacher of math	is		gone	away

(ii)

As already noted by Brandi and Cordin (1989) and confirmed by Casalicchio and Cordin (2020, pp. 111, 261), clitic reduplication is excluded in three contexts: (a) when the subject occurs in the postverbal position without being right-dislocated (cf. 30) (see also Poletto 1993); (b) in both restrictive relative clauses (cf. 31) and wh-interrogatives on the subject (cf. 32); and (c) with the negative quantifier nissun/nessun/nissuni/nessuni ‘nobody’ (cf. 33).14

(30)	a.	Va	via	la maestra	de matematica.
		goes	away	the teacher	of math
		‘The math teacher is going away.’
	b.	È	nà	via	la maestra	de matematica.
		is	gone	away	the teacher	of math
		‘The math teacher went away.’

(31)	Le putele	che	(*le)	ha	parlà	con ti …	(Brandi and Cordin 1989, p. 126)
	the girls	that	they.cl	have	talked	with you
	‘The girls who talked to you …’

(32)	Quante	putèle	(*le)	ha	parlà	con ti?	(Brandi and Cordin 1989, p. 125)
	how-many	girls	they.cl	have	talked	with you
	‘How many girls talked to you?’

(33)	Nesun	(*l’)	è	vegnù	en	tèmp.	(adapted from Casalicchio and Cordin 2020, p. 111)
	nobody	he.cl	is	come	in	time
	‘Nobody was on time.’

However, systematic results do not always confirm the generalization presented in (ii), as noted by Casalicchio and Cordin (2020, p. 111, footnote 22) and Kruijt et al. (n.d.), among others. Nevertheless, several sentences in ASIt, the online Atlante Sintattico d’Italia, which collects written dialectal data, exhibit the co-occurrence ‘clitic – postverbal subject’, especially in the Trentino area.15 This phenomenon has been recently confirmed by data collected within projects conceived with a different methodology, including AThEME and VinKo, which are based on oral recordings.

Following the analysis proposed by Kruijt et al. (n.d., Section 4.2, Trentino dialects: agreement with a postverbal subject), whose main aim was to compare the data collected by AThEME (based on fieldwork) with the data collected by VinKo (based on crowdsourcing), two tendencies clearly emerge:

(a)

Postverbal [-hum] subjects in declarative sentences do not to realize the clitic (36 cases versus 8 in VinKo) (cf. 34) (data taken from Kruijt et al., n.d.):

(34)	È	scominzià	la	scola.	(Brentonico, S0134_tre_U0049)
	is	started	the	school
	‘School has started.’

(b)

Postverbal [+hum] subjects in indirect interrogative sentences almost always require a clitic (18 cases vs. 1):

(35)	Non	so	miga	ndo	che=’l	sia	nà	el	Marco. (Brentonico, S0115_tre_U0088)
	Not	know	not	where	that=he.cl	is.conj	gone	the	Mark
	‘I don’t know where Mark has gone.’s

The occurrence of a clitic with a postverbal [+hum] non-right-dislocated subject in presentative sentences, as shown in (36), is not so clear-cut (see also Casalicchio and Cordin 2020, p. 112); nevertheless, it is far from being abandoned by the Trentino speakers:

(36)	a.	L’è	bela	to sorela.
		she.cl-is	nice	your sister
		‘Your sister is nice.’
	b.	I	è	za	grandi	i	pòpi.
		they.cl	are	already	grown-up	the children
		‘The children are already grown up.’

Without going into the details of Kruijt et al.’s (n.d.) analysis16, what matters to us is the different situation represented by data collected in Veneto. As the next paragraph will demonstrate, the dialects spoken in Veneto are coherently characterized by the absence of subject agreement (no clitic) with postverbal subjects (strong confirmation of Brandi and Cordin 1981, 1989 generalization, cf. (ii) above) and by the gradual disappearance of the clitic expletive. This graduated loss seems to follow the flow of the river Adige, starting from the southwest (Verona), where it is still attested, eastward (Vicenza, Padova, Treviso, Venezia), where it is marginally residual or even completely absent, which is perfectly in line with the situation already presented and discussed with regard to the Venetian dialects by Benincà (1994).

4. The Typology of Lexical Expletives in the Dialects Spoken in Veneto

The translation of sentence n. 136 (è nevicato ‘it snowed’) in the VinKo database clearly shows a gradual disappearance of the clitic expletive with meteorological verbs in the dialects spoken in Veneto compared with the situation attested in Trentino, as discussed in the previous paragraph.

The data collected in the province of Verona, the southwestern province on the border with Trentino, show that at least a fourth of the speakers maintain the clitic.17 More precisely, the 41 records collected in the province of Verona display a lexical expletive in 12 cases—usually the third singular masculine l’ (cf. 37a–b), and a18 in two cases (cf. 38)19—and with no clitic in 29 records (39)20:

(a)

Examples with the clitic and both auxiliary verbs a ‘has’ and è ‘is’:

(37)	a.	L’a	nevegà.	(Verona, S0136_vec_U0420)
		it.cl-has	snowed
		‘It snowed.’
	b.	L’è	nevegà.	(Verona, S0136_vec_U0439)
		it.cl-is	snowed

(38)	A	ga	nevegà.	(Bonavigo, S0136_vec_U0450)
	prt	has	snowed

(b)

Examples without the clitic:

(39)	a.	A	nevegà.	(Verona, S0136_vec_U0403)
		has	snowed
		‘It snowed.’
	b.	È	nevegà.	(Sona, S0136_vec_U0402)
		is	snowed
	c.	Ghe	nevegà.	(Bonavigo, S0136_vec_U0408)
		has	snowed
	d.	Ga	nevegà.	(Legnago, S0136_vec_U0300)
		has	snowed

In the province of Vicenza, only two out of ten informants (Schio, Sarego) have the clitic expletive, as exemplified in (40), while the vast majority of speakers (four-fifths of the total) do not express it (cf. 41, resembling 39 in the province of Verona):21

(40)	(E)l ga	nevegà.	(Schio, S0136_vec_U0320)
	it.cl-has	snowed
	‘It snowed.’

(41)	Ga	nevegà.	(Recoaro, S0136_vec_U0306)
	has	snowed
	‘It snowed.’

In the province of Padua, none of the recordings feature the clitic. All speakers use the auxiliary ga:

(42)	Ga	nevegà.	(Cittadella, S0136_vec_U0326)
	has	snowed
	‘It snowed.’

The same holds for the province of Treviso (cf. 43), with one exception (cf. 44):

(43)	Ga	nevegà.	(Castelfranco, S0136_vec_U0302)
	has	snowed
	‘It snowed.’

(44)	L’è	nevegà	(ma no tant).	(Codognè, S0136_vec_U0647)
	it.cl-is	snowed	(but not too much)
	‘It snowed (but not too much).’

Finally, the same holds for the province of Venice (cf. 45), with the exception of San Donà del Piave (cf. 46):

(45)	Ga	nevegà	(Venezia, S0136_vec_U0454)
	has	snowed
	‘It snowed.’

(46)	L’è	nevegà	(San Donà, S0136_vec_U0651)
	it.cl-is	snowed
	‘It snowed.’

The presentation and discussion of the data in this section do not pretend to exploit the complexity of the situation in terms of the occurrence of the subject clitics in the dialects spoken in Veneto (as, for instance, the role of auxiliary selection or extra-linguistic factors, such as either the age of the informant or the difference between data collected in city areas versus rural ones). Nevertheless, two generalizations can be drawn:

(a)

The maintenance of lexical expletives in the dialects spoken in Trentino is definitely more robust in comparison with the Venetian region with respect to both impersonal subject clitics and the possible co-occurrence with a postverbal subject;

(b)

The maintenance of lexical expletives in Veneto exhibits a clear decrease from the southwestern area (the province of Verona) toward the southeastern provinces of Vicenza, Padova, and Venice.22

Second Summary

The comparison between the data collected in Trentino and discussed in Section 3 and the Venetan data just discussed in the preceding section (cf. Section 4) is synthesized in the following

Table 2. Rate of expletive realization.

		TRENTO	VERONA	VICENZA	PADOVA	VENEZIA
(47)	Impersonal subject clitics	85.7% (12 su 14)	29.9% (12 su 41)	20% (2 su 10)	none	none
	Expletives with a postverbal subject [+/-hum]	(a) 18.1% (b) 94.7% (a. 8 su 44 [-hum]; b. 18 su 19 [+hum])	none	none	none	none

(exceptional cases excluded):23

As already noted in the preceding sections (3) and (4), the percentages clearly show that the overt realization of expletives decreases “along the Adige River”. In this perspective, the province of Verona emblematically represents the turning point coherently with the course of the river. As we will see in the next concluding section, along the same axis, verb–subject inversion (i.e., enclisis to the finite verb) in the interrogative context is gradually replaced by other structures (clefts, mainly).

5. Linguistic Borders and the Potential Role of Language Contact

The special position of the dialects spoken in Trentino with respect to the maintenance of subject clitics is well documented in terms of both impersonal constructions with weather-related verbs and the co-occurrence with postverbal subjects (of a certain kind, that is, [+human]), as already shown by the data presented and discussed in Section 3. Further expected confirmation comes from the position of the clitic in direct questions (cf. 48 and 49) in these same contexts (i.e., with weather verbs and unaccusative verbs):24

(48)	a.	A=lo	nevegà?
		has=it.cl	snowed
		‘Did it snow?’
	b.	È=lo	nevegà?
		is=it.cl	snowed

(49)	È=loi	rivà	to	nonoi?
	is=he.cl	arrived	your	grandpa
	‘Did your grandpa arrive?’

The different position of the subject clitic in [+WH] versus [-WH] root clauses (i.e., enclisis versus proclisis) is an uncontroversial argument for the assumption of finite verb movement to C in the interrogative context25 and can be extended to the full paradigm of subject clitics with highly complex and intriguing results concerning the possible gaps attested by the different dialectal paradigms.

In terms of the dialects spoken in Trentino, the following patterns can be considered canonical patterns:26

(50)	a.	Cosa	fa=lo?
		what	does=he.cl
		‘What is he doing?’
	b.	Cosa	a=lo	combinà?
		what	has=he.cl	done
		‘What did he do?’
	c.	‘Ndo	va=lo?
		where	goes=he.cl
		‘Where is he going?’

The maintenance of subject enclisis in direct interrogative clauses (cf. 50) co-exists with alternative constructions that rely on different strategies:

(i) Polarity questions may eventually rely on a different stress pattern without word order variations, meaning no subject clitic inversion (raising intonation, similar to standard Italian, cf. 51a-c, in specific pragmatic contexts as with an expression of surprise):

(51)	a.	L’a	magnà	su	tut?
		he.cl-has	eaten	prt	all
		‘He ate it all, didn’t he?’
	b.	L’a	nevegà	anc’ancò?
		it.cl-has	snowed	also-today
		‘It snowed today too, didn’t it?’
	c.	L’i è	rivà	anca	to	nonoi?
		he.cl-is	arrived	also	your	grandpa
		‘Your grandfather arrived too, didn’t he?’

(ii) wh-questions can be realized with a cleft construction, where the subject clitic of the ‘that’ clause regularly occurs on the left of the finite verbal form (cf. 52 in comparison with 50a):27

(52)	a.	Cos’è=(l/lo)	che’l	fa?28			(Casalicchio and Cordin 2020, p. 120)
		what-is=it.cl	that-he.cl	does
		‘What is he doing?’
	b.	End’è=l	che	l’a	metü	la	bórsa?
		Where-is=it.cl	that	have.2ps	put	the	bag
		‘Where did you put the bag?’					(adapted from Casalicchio and Cordin 2020, p. 320)

We do not want to enter the complexity of facts, either extra-linguistic (i.e., the age of the speaker)29 or purely morphological (i.e., person and number variations), which favors one construction over the other(s). Most relevant to our line of argumentation is that the maintenance of subject enclisis to the finite verb in root wh-clauses seems to follow the same path already noted for the occurrence of lexical expletives with meteorological verbs.

Comparing the data discussed for Trentino with those of Veneto, the following three generalizations can be proposed:

• Whenever the expletive subject does not occur in the declarative sentence, it does not appear in the interrogative context either in enclisis (cf. 53a-b) or in the much more common cleft construction (cf. 54):

  (53)	a.	Ga	nevegà?			(Cittadella, Padova)
  		has	snowed
  		‘Did it snow?’
  	b.	Dove	ga	nevegà	uncò?
  		where	has	snowed	today
  		‘Where did it snow today?’

  (54)	Dove	xe	che	ga	nevegà	uncò?	(Cittadella, Padova)
  	where	is	that	has	snowed	today
  	‘Where did it snow today?’

• Enclisis appears regularly in polarity questions (cf. 55):

  (55)	a.	Va=lo	via?		(Verona)
  		goes-he.cl	away
  		‘Is he leaving?’
  	a.’	Va=eo	via?		(Cittadella, Padova)
  		goes-he.cl	away
  	b.	È=lo	na	via?	(Verona)
  		is-he.cl	gone	away
  		‘Did he leave?’
  	b.’	Z=eo	onda(to)	via?	(Cittadella, Padova)
  		is-he.cl	gone	away
  	c.	A=lo	magnà	tuto?	(Verona)
  		has=he.cl	eaten	all
  		‘Did he eat it all?’
  	c.’	Ga=eo	magnà	tutto?	(Cittadella, Padova)
  		has=he.cl	eaten	all

The rising intonation without word order variation (cf. 56) is possible in specific pragmatic contexts (e.g., an expression of surprise):30

(56)	a.	El	va via?	Sul serio?			(Cittadella)
		he.cl	goes away.	Seriously
		‘He is leaving. Seriously?’
	b.	L’è	ndà	via?	Sul serio?		(Cittadella)
		he.cl-is	gone	away.	Seriously
		‘He left. Seriously?’
	c.	L’à	magnà	tuto	anca	uncò?	(Cittadella)
		he.cl-has	eaten	all	also	today
		‘Did he eat it all today as well?’

III.

In WH-questions, enclisis is maintained (cf. 57), but there is a strong preference for the cleft construction whose attestation definitely increases from southwest (Verona) moving eastward (Padua) and represents the only option in Venice:

(57)

a.

Ndo

va=lo?

(Verona)

where

goes=he.cl

‘Where is he going?’

a.’

Dove

xè che

el

va?

/

Dove

va=eo?

(Cittadella, Padova)

where

is

that

he.cl

goes

/

Where

goes=he.cl

b.

Cos’a=lo

magnà?

(Verona)

what-has he.cl

eaten

‘What did he eat?’

b.´

Cosa

zè

che

el

ga

magnà?

/

Cosa

ga=eo

magnà?

(Cittadella, Padova)

what

is

that

he

has

eaten

/

what

has=he.cl

eaten

The geographical distribution of the data discussed to this point can be visualized in the following simplified but essential map, where the arrows correspond to the river Adige:

(58) Geographical distribution:

Why do the dialects spoken in Trentino maintain both lexical expletives and enclisis to the finite verb (see the yellow arrow in 58) moderating the success of the cleft construction in Romance dialectal syntax (see the blue arrow in 58)?

Our answer relies on a theory of language contact that attributes a fundamental role to the linear convergences deriving from different structural systems in terms of (a) regulating the speed of the syntactic change and (b) determining the direction of the diachronic development among the array of possible paths that are compatible (i.e., coherent) with a specific language system.31

As a matter of fact, the political border between Trentino (once called Welschtirol, referring to Italian Tyrol) and South Tyrol coincides with the German–Italian linguistic border with a well-studied situation of plurilingualism (at both the dialectal and the standard linguistic level).32

As for Trentino (the southern province of the autonomous region Trentino–Alto Adige/South Tyrol), the relevance of German (as a second language in school education) and the protection of linguistic minorities (both Ladin and Germanic varieties) is well known from both political and legal perspectives.

What we want to suggest is that the influence of German could have played a role with respect to both the relatively stronger preservation of expletive subjects and the maintenance of German-like patterns, such as pronominal subject inversion (enclisis of the pronominal subject to the finite verb) in the wh root clause damming the preference for the Romance cleft construction. This convergence applies at the superficial (i.e., linear) level, coherently with the Romance structure of the dialects spoken in Trentino, which remain:

(i)

Null-subject languages (following Brandi and Cordin 1981, 1989, we analyze the clitic as the manifestation of subject agreement);

(ii)

’residual‘ V2 languages (i.e., V2 languages), such as Italian and English, where finite verb movement to C is limited to well-defined contexts.

The notion of “residual V2,” as introduced by Rizzi (1990, 1996), suggests the idea that both Old English and Old Romance were characterized by a more ’consistent‘ V2. As Poletto (2019) already pointed out, V2 is not a unique phenomenon but implies a complex, fine-grained typology: different types of V2 allow (and, in fact, did allow) different ways of losing processes. The role of contact with varieties of the German type, where V2 does not depend on the informational structure (i.e., it is not compatible with a ’criterial‘ approach, and it is not compatible with proclisis and, ultimately, with a positive value of the NSP) is limited to both the speed of the change and the maintenance of residual aspects of the phenomenon.33

Conversely, the conservative aspects of Cimbrian syntax (cf. Section 2), which behaves like German regarding both the negative value of the NSP and the maintenance of structural V2 (i.e., generalized finite verb movement to the lower projection of the C-domain in both declarative and interrogative clauses), could have found in the morphosyntax of the surrounding Italian dialects—subject clitics and residual V2 in the interrogative context—a robust dam that has slowed the speed of the potential loss of Germanic V2 and the resetting of the NSP.