V2 and Subject-Verb Inversion in Ladin and Romansh

Abstract

In this article, we analyze the order verb-(clitic) subject that characterizes Rhaeto-Romance languages in V2 and interrogative contexts. In these varieties, the subject is necessarily realized by lexical or pronominal elements in preverbal position, except in inversion contexts. Some of these varieties have subject clitics (SCls) that display a distribution similar to that of full pronouns. Furthermore, in some, subject-verb inversion involves the enclisis of SCls on the verb. Enclitics are distributionally different from proclitics and are characterized by distinct paradigms, which show specialized properties somewhat comparable to inflections. We will study subject syntax in inversion contexts within the Phase framework, where enclitics will be viewed as the result of syntactic Merge and of amalgamation with the verb stem in INFL. A crucial topic is the relationship between functional morphemes and spelling domains.

1. Introduction

Rhaeto-Romance varieties (Ladin and Romansh) exhibit inversion between the lexical subject or subject clitic (SCl) and verb in V2 and in other modal contexts, as in the interrogative. As illustrated by the examples from the Zernez variety (Engadine), the subject–verb order, (1a), is inverted to verb–subject, (1b), if a lexical expression is in first position. (1b) shows the two alternants, one with the enclitic SCl -i and the other with the full pronoun. Finally, (1c) exemplifies interrogative inversion. We note that, in the glosses, the inflection is indicated by the simple specification of agreement features, while the subject enclitic is indicated as SCl.

(1)	a.	els/elas	dorm-an	kwia
		they.m/f	sleep-3pl	here
		‘They sleep here’

	b.	da’vo dorm-an-i	/da’vo dorm-an	els/elas
		later sleep-3pl-SClpl	/later  sleep-3pl	they.m/f
		‘They sleep later’

	c.	dorm-an-i?
		sleep-3pl-SCl3pl
		‘Do they sleep?’
						Zernez

Independent of V2, interrogative inversion occurs in many Gallo-Italic dialects and is reminiscent of inversion in French and Germanic languages. In varieties with SCls in the postverbal position, enclitic forms do not necessarily correspond to proclitic pronouns, and furthermore, there may be an asymmetric distribution between proclitic and enclitic forms.

In the literature, enclitic forms have been treated as the enclisis of an SCl, assuming that proclisis and enclisis are derived through movement from the same syntactic node in the sentence structure. The mapping proposed by the cartographic model1 is, however, weakened by significant asymmetries between proclitics and enclitics. Indeed, some authors conclude that enclitics are inflections of a specialized paradigm (cf. Zamboni, 1974; Sportiche, 1997; see discussion in Section 6), a conclusion suggested by several factors, such as the lack of correspondence between the proclitic and enclitic SCl paradigms and the incorporation of the enclitic in the morphophonological structure of inflection.

In Section 2, we present the data regarding the Romansh varieties of Trun (Surselva), Mulegns (Surmiran), Donat (Sutselva), Zernez (Engadine), and Müstair (Müstair Valley), and the Ladin varieties of La Pli de Mareo (Marebbe Valley) and San Cassiano (Badia Valley). In Section 3, the paradigm of the interrogative inversion of the northern Venetian dialect of Corte (Livinallongo) is exemplified. The data we provide compare the verbal forms with proclitic subjects and those with enclitic subjects in V2 and interrogative contexts. Schemes summarizing the different types of SCls, enclitics, and inflections are provided in Section 4, which show the differences between proclitic and enclitic SCls, both in distribution and morphology. Section 5 is devoted to the notions of clitic and inflection in light of a brief discussion of the relevant literature. Section 6 presents a characterization of SCls, enclitics, and inflections in terms of their selection constraints, seen as the crucial representation of the lexical properties of these vocabulary items (cf. Halle & Marantz, 1994). In Section 7, we analyze the distribution and insertion of clitics and inflections based on the Strong Minimalism Thesis (Chomsky, 2021, 2024a, 2024b). The difference between proclitic and enclitic subject pronouns stems from their subcategorization and selection constraints in relation to the verb. Analysis of the data will lead us to conclude that enclitics differ from genuine inflections. The latter are merged and amalgamated into the verb in INFL, while enclitics are associated with the vP phase and amalgamated with the verb realizing INFL. Preverbal SCls are optionally merged with the verb in the CP domain. The contrast between the declarative and V2 order is related to a different phasal order in the expression of thematic and clausal properties, assuming Chomsky (2024a, 2024b).

2. V2 in Rhaeto-Romance2

The Romance varieties spoken in the Grisons and the Ladin areas of Trentino typically require an expressed subject, whether a lexical item or a pronoun, including the SCls. They exhibit V2 distribution (Haiman & Benincà, 1992). The verb is in second position, or, more precisely, the verb is preceded by the subject or by an adverb or an argument, DP or PP. In these cases, the subject is placed to the right of the verb, both in main and subordinate clauses. The Romance varieties of the Surselva lack SCls, in contrast to the Engadine and Ladin varieties, which exhibit proclitic SCls, albeit with differences and a partially reduced distribution in comparison with the typical paradigms of Northern Italian dialects (Manzini & Savoia, 2002; Manzini & Savoia, 2005, §3.13.2). Furthermore, in Ladin and Engadine Romansh (with differences), in V2 and interrogative contexts, the verb includes an enclitic SCl, compatible with the presence of a postverbal lexical subject.

Let us first consider the Sursilvan variety of Trun. (2a) illustrates the subject–verb pronominal or lexical contexts; (2b) is the V2 inversion, and (2c) is the interrogative inversion3.

(2)

a.

b.

c.

jau

maʎ-ël

suenter

suenter

maʎ-ël

jau

I

eat-1sg

later

later

eat-1sg

I

‘I will eat later’

ti

maʎ-as

tʃai maʎ-as

ti?

you

eat-2sg

what eat-2sg

you?

‘You are eating’

‘What do you eat?’

tia   sora

vɪɲ

tier mai

vɪɲ

tia sora?

your sister come.3sg

to me

come.3sg   your sister?

‘Your sister is coming to me’

‘Does your sister come?’

el klam-a-v-a mai

iɲemprim klam-a-va  el  mai

he call-Past-3sg me

before   call-Past-3sg he  me

‘He called me’

‘Before he called me’

elas dorm-ën

suenter dorm-ën elas

dorm-ën elas/els?

they.f sleep-3pl

later sleep-3pl they.f

sleep-3pl they.f/m?

‘They sleep’

‘They will sleep later’

‘Do they sleep?’

Trun

In Sutsilvan, Vallader (Engadine), and Ladin varieties, subject clitics (SCls) occur, which in inversion contexts are realized as enclitics. They are, in turn, in complementary distribution with full pronouns or lexical elements, at least in proclisis, while in enclitic position, they can co-occur with lexical or pronominal postverbal subjects. However, we must make some further distinctions. First, only Ladin has a broader SCl system in proclisis and enclisis; the other varieties generally have only third-person SCls in proclisis, whereas in enclisis, the paradigm also includes other persons. Second, in Low Engadine (Vallader), here Zernez, and neighboring areas such as Val Mala, here Donat (Sutsilvan), and Surmeir, here Mulegns (Surmeiran), the co-occurrence of the SCl in enclisis and the lexical/pronominal subject is marginal. This distribution is illustrated by the data of the third-class verb ‘to sleep’, whose inflection slightly differs from the Class A verbs, where the exponent -a occurs at least in 3sg and the TV is different. Finally, the following examples display the differences between proclitic and enclitic SCls, whereby proclitics and enclitics may show different forms and a different distribution.

The data for Romansh are in (3)–(6). (a) illustrates the sub-system of proclitic SCls in declarative clauses, where they are in complementary distribution with full pronouns and lexical subjects. Specifically, they alternate stressed forms such as els/elas ‘they.m./f’ with the proclitic forms i ‘they’ for the 3pl, and in some varieties, for instance that of Müstair in (6), also el/ela ‘he/she’ vs. al/la for the 3sg. Insofar as full pronouns and SCl cannot co-occur, their distribution ends up coinciding in preverbal contexts, as suggested by the examples in (a). (b) and (c) illustrate the V2 contexts and interrogatives, where the subject is realized either via SCls in enclisis or postverbal full pronouns/lexical elements. Obviously, the same V2 and interrogatives have the same syntactic alternants.

In V2 and interrogative contexts, pronominal or lexical subjects are in a postverbal position. In these contexts, a richer paradigm of enclitic SCls is attested. The latter can occur alone or co-occur with a lexical subject, thus contrasting with the preverbal position. This possibility is variably allowed in all varieties studied here. In some Romansh varieties, the enclitic in the 1pl and 2pl is absent. The point is that these forms can also occur without a postverbal full pronoun, as in (4) suainter dur’m-e-s ‘later you sleep’ for Donat. This possibility suggests a particular status for the 1pl and 2pl inflection, which we will return to in Section 5.

(3)

a.

b.

c.

ia dorm

siva dorm-a

nua

dorm-a?

I sleep.1sg

later sleep-SCl1sg

where sleep-SCl1sg

te

dorm-ës

dorm-ës(-t)?

you sleep-2sg

sleep-2sg-SCl2sg?

el/ela dorm-a

dorm-al/-la?

he/she sleep-3sg

sleep-3sg.m/f

noks dur’m-i-ŋ

siva durm-i-n-za

dur’m-i-n-za?

we sleep-TV-1pl

later sleep-TV-1pl-SCl1pl

sleep-TV-1pl-SCl1pl?

voks dor’m-i-ts

dur’m-i-ts?

you sleep-TV-2pl

sleep-TV-SCl2pl?

els/elas/i dorm-an

dorm-iʎ?

they.m/f./SCl sleep-3pl

sleep-SCl3pl?

Mulegns

(4)

a.

b.

c.

jau dorm

ots dorm jau

dorm jau?

I sleep.1sg, etc.

today sleep I

sleep I?

tai dorm-as

damaŋ dorm-as(-te)

dorm-as (tai)?

you sleep-2sg

tomorrow sleep-2sg-SCl2sg

sleep-2sg you?

el/ela dorm-a

suainter dorm-al/dorm-la

dorm-al/la?

(s)he sleep-3sg

later sleep-SCl3sg.m/f

sleep-SCl3sg.m/f?

nus dur’m-ai-n

ots durm-ai-n-za

dur’m-ai-n-za?

we sleep-1pl

today sleep-TV-1pl-SCl1pl

sleep-TV-1pl-SCl1pl?

vus dur’m-e-s

suainter dur’m-e-s

dur’m-e-s?

you sleep-2pl

later sleep-TV-SCl2pl

sleep-TV-SCl2pl?

elts/elas/i dorm-ën

damaŋ dorm-ën-i

dorm-ën -i/els/elas?

they/SCL sleep-3pl

tomorrow sleep-3pl-SCl

sleep-3pl SCl3pl/they?

Donat

(5)

a.

b.

c.

eu/jou dorm

kwia dorm-i (jou)

I sleep

here sleep-SCl1sg (I)

ty dorma-ʃ(t)

da’vo mandʒa-ʃ-t

dorma-ʃ-t?

you sleep-2sg

later eat-2sg-SCl2sg

sleep-2sg- SCl2sg?

el/ela dorm-a

da’vo dorm-al/la

dorm-ël/dorm-la?

he/she sleep-3sg

later sleep-SCl3sg.m/f

sleep- SCl3sg.m/f?

SCl plow-a

plouv-i?

it rain-3sg

rain-SCl3sg?

nus dor’m-i-n(-a)

da’vo dorm-i-n-a

a tʃeza dor’m-i-n-a?

we sleep-TV-1pl

later sleep-TV-1pl-SCl1pl

sleep-TV-1pl-SCl1pl?

vus dor’m-i-s

da’vo mandʒ-a-is

dor’m-i-s?

you sleep-TV-2pl

later eat-TV-2pl

sleep-TV-2pl?

els/elas/i dorm-an

wosa dorm-an-i

dorman-i/els/elas?

They.m/f/SCl sleep-3pl

now sleep-3pl-SCl3pl

sleep-SCl3pl/they?

Zernez

(6)	a.			b.				c.
	jau dorm			domauŋ dorm-i (jau)				ndʒua dorm-i (jau)?
	I sleep.1sg			tomorrow sleep-SCl1sg I				where sleep-SCl1sg I?
	ty dorm-aʃ			a’la dorm-aʃ-t bain				ndʒua dorm-aʃ(-t)?
	you sleep-2sg			there sleep-2sg-SCl2sg well				where sleep-2sg-SCl2sg?
	el/ela dorm-a			la dorm-a el/ela
	he/she sleep-3sg			there sleep-3sg he/she
	al/ala dorm-a			kwia nu dorm-al/la				dorm-al/la?
	SCl.m/f sleep-3sg			here Neg sleep-SCl3sg.m/f				sleep- SCl3sg.m/f?
	nu dor’m-ai-ŋ			do: durm-ai-n-i nu				dur’m-ai-n-i?
	we sleep-TV-1pl			later sleep-TV-1pl-SCl1pl				sleep-TV-1pl-SCl1pl?
	vu dur’m-a-wat			la durm-a-wat     vu				ndʒua dur’m-a-w-at?
	you sleep-TV-2pl			there sleep-TV-2pl  you				where sleep-TV-2pl?
	es/elas dorm-an			la dorm-an els/elas
	they.m/f sleep-3pl			there sleep-3pl they.m/f
	i dorm-an			do: dorm-n-i la				dorm-n-i?
	SClpl sleep-3pl			later sleep-3pl-SClpl there				sleep-3pl-SCl3pl
								Müstair

Proclitics, like full pronouns, are in complementary distribution with lexical subjects in (7a), (8a), and (9a). Enclitics can instead combine with the lexical subject, as in (7b), (8b), and (9b) for V2 and (7c), (8c), and (9c) for V2 interrogatives.

(7)

a.

iʎ pup viɲ

b.

siva viɲ-al

iʎ pup

the boy come.3sg

later come.3sg-SCl3g the boy

‘The boy comes’

‘The boy comes later’

c.

dorm-al

iʎ pup?

sleep-SCl3sg

the boy?

‘Does the boy sleep?’

Mulegns

(8)

a.

l ufant  es  ɲ-i:

b.

da’vo veɲ(-ël)

l ufant

the boy is   come-PP

later come.3sg SCl3sg

the boy

‘The boy has come’

‘The boy come later’

c.

dorm-la

mia sour?

sleep-SCl3sg.f

my sister’

‘Does my sister sleep?’

Zernez

(9)	a.	mes fra:r dorm-a			a’la		b.	kwia dorm-a/al			mes fra:r
		my brother sleep-3sg there						here sleep-3sg/SCl3sg			my brother
		‘My brother sleeps there’						‘My brother sleeps here’
							c.	dorm-n-i	az ufaunts?
								sleep-3pl-SCl3sg	the boys?
								‘Do the boys sleep?’
											Müstair

Ladin varieties spoken in La Pli de Mareo and S. Cassiano Badia provide a complete paradigm of SCls in alternation with full pronouns or lexical subjects. As we have underlined, in proclisis, the SCl cannot combine with an expressed subject, as in (10a) and (11a). The examples in (10b,c) and (11b,c) show full pronouns in subject position and the enclitic forms in interrogatives. Compared to Romansh varieties, in the Ladin questions in (10c) and (11c), the enclisis of the SCl is systematic, while in V2, we essentially have three possibilities: verb + enclitic, verb + enclitic + lexical subject, and verb + lexical subject.

(10)	a.	SCls			b.	Full pronouns	c.	Enclitics
		i dorm-i				ju dorm-i		o’la dorm-i pa (ju)?
		1sg sleep-1sg				I sleep-1sg		where sleep-1sg Prt (I)?
		‘I sleep’, etc.
		te  dorm-es				tø dorm-ës		o’la dorm-es-te pa (tø)?
		2sg sleep-2sg				2sg sleep-2sg		where sleep-2sg-SCl2sg Prt
									(you)?
		al/ara dorm				el/era dorm		dorm-el/-era?
		he/she sleep.3sg				he/she sleep.3sg		sleep-SCl3sg.m/f?
		i dor’m-ju-ŋ eŋ’lo				nos dor’m-ju-ŋ a’tlo		o’la dor’m-ju-n-ze pa?
		1pl sleep-TV-1pl there				we sleep-TV-1pl there		where sleep-TV-1pl-SCl1pl
									Prt?
		i  dor’m-i-s eŋ’lo				os dor’m-i-s		dor’m-i-z-e os atlo?
		SCl sleep-TV-2pl there				you sleep-TV-2pl		sleep-TV-2pl-SCl2pl you there?
		ai/ares  dorm				ei/eres dorm masa		dorm-ai/-eres pa (ei/eres)?
		3pl.m/f  sleep.3pl				they.m/f sleep much		sleep-SCl3plm/f Prt
								(they.f/m)
								La Pli de Mareo

(11)	a.	SCls		b.	Full pronouns		c.	Enclitics
		i    dorm			jø dorm			u’la dorm-i pa?
		SC1sg  sleep.1sg			I sleep.1sg			where sleep-SCl1sg Prt?
		të dorm-ës			tø   dorm-ës			u’la dorm-ës-t pa?
		SCl2sg sleep-2sg			you sleep-2sg			where sleep-2sg-SCl2sg Prt?
		al/la   dorm			al/ala dorm			u’la dorm-l/ëla?
		he/she  sleep.3sg			he/she sleep.3sg			where sleep-SCl3sg.m/f?
		nos  dur’m-ju-ŋ						u’la dur’m-ju-n-ze pa nos?
		we sleep-TV-1pl				where sleep-TV-1pl-SCl1pl Prt we?
		os dor’m-i-s						u’la dur’m-i-s pa?
		you sleep-TV-2pl						where sleep-TV-2pl Prt?
		ai/alës dorm			ai/alës dorm			dorm-i/lës pa?
		they.m/f sleep.3pl			they.m/f sleep			sleep-SCl3pl.m/f Prt?
								San Cassiano Badia

In V2 contexts, enclitic SCls are compatible with expressed postverbal subjects, as illustrated by the examples in (12) and (13). Thus, in contrast to the canonical declarative structure, in (12a) and (13a), in the other contexts the enclitic combines with the expressed subject, as in (12b) and (13b). (12c) illustrates the enclitic forms of 3ps, which differ slightly from the proclitic in (12a), partially overlapping with full pronouns.

(12)	a.	Maria/Žaŋ/el/era   dorm				b.	de’do dorm-el/-era (Žaŋ/Maria)
		Mary/John/he/she  sleep.3sg					later sleep-SCl3sg.m/f (John/Mary)
		‘Mary/John/he/she sleeps’					‘Later John/Mary sleeps’
		al/ara   dorm					atlo dorm-ju-n-ze (nos)
		SCl.m/f  sleep.3sg					there sleep-TV-1pl-SCl1pl (we)
							‘We sleep there’
		mio fre	n	e nia ɲ-y			to fre e-l pa ɲ-y?
		my brother Neg is Neg come-PP					your brother is-SCl3sg.m Prt come-PP?
		‘My brother has not come’					‘Did your brother come?’
		ei/eres	veŋ
		they.m/f	come.3pl
		ai/ares	veŋ			c.	veɲ-el/era?  /veɲ-ai/eres?
		SClmpl/fpl  come.3pl					come-3sgm/f? come-SCl3plm/-3plf?
		‘They come’					‘Does (s)he come?/Do they come?’
							La Pli de Mareo

(13)	a.	nos dorm-ju-n ai’lo	b.	kilo dorm-ju-n-ze nos
		we sleep-TV-1pl there		there sleep-TV-1pl-SCl1pl we
		‘We sleep there’		‘We sleep there’
		ai më kard-a		de’do më kard-i ai
		they.m me call-3		later me call-SCl3pl they
		‘They call me’		‘They call me later’
			c.	tʃi vaig-l pa to fre?
				what see-SCl3sg Prt your brother?
				‘What do your fre see?’
				S. Cassiano Badia

The fact that enclitics can combine with expressed subjects suggests that these elements have a different syntactic status. The idea we will discuss is that enclitics are part of verbal inflection, thus differentiating themselves from proclitic elements, which are selected in the INFL space.

3. SCls-Verb Inversion in Northern Italian Dialects

Before addressing the analysis of V2, subject-verb inversion, and the status of SCls/enclitics in Rhaeto-Romance varieties, let us briefly consider the distribution of SCls in northern Italian varieties. In these varieties, SCls are systematically combined with the verb, regardless of whether the lexical subject is realized or not. Inversion occurs in interrogatives, while V2 is absent. This typology is already attested by dialects spoken in areas adjacent to the Ladin systems we considered in Section 2, such as the Fodom dialects, exemplified here by the data from Corte (Livinallongo) in (14a,b).

(14)	a.	dorm-e	b.	dorm-jo
		sleep-1sg		sleep-SCl1sg?
		‘I sleep’, etc.		‘Do I sleep ?’, etc.
		te   dorm-e		dorme-to?
		2sg sleep-2sg		sleep-SCl2sg?
		al/a dorm		dorm-elo/ela?
		3sg.m/3sg.f sleep		sleep-SCl3sg.m/f?
		dor’m-jo-ŋ		dor’m-jo-n-zo?
		sleep-TV-1pl		sleep-TV-1pl- SClpl?
		dor’m-j-ei		dor’m-jei-zo?
		sleep-TV-2pl		sleep-TV-2pl-SClpl?
		i/le   dorm		dorm-eli/ele?
		3pl.m/f sleep-3pl		sleep-SCl3pl.m/f?
					Corte

In these dialects, as is usual in North-Italian dialects, the lexical subject combines with the SCl, as in (15a), and a phrase in the first position does not trigger the V2 context, as in (15b).

(15)	a.	la tozata la		dorm
		The girl SCl.fsg		sleep.3sg
		‘The girl is seeping’
	b.	el liber	el	l	a	lie-t
		the book	SCl	OCl	have.3sg	read-PP
		‘He read the book’				Corte

For completeness, let us remember that Haiman and Benincà (1992), Benincà (1994, 2013), and other authors maintain that interrogative inversion is the residue of a more general V2 rule applied in medieval varieties.

4. Paradigms: SCls, Enclitics, and Inflections

Let us consider the data presented, starting from the verbal morphology in declarative sentences. In the internal structure of the declarative verb forms in (3a)–(6a) and (10a), (11a), and (14a), different patterns emerge, as illustrated for the present indicative in (16), where the inflectional exponents combine with the root. A pattern with the simple root in the third person appears in three of the seven paradigms; ∅ also characterizes the first singular in five cases. The first and second plural include the thematic vowel TV; in the second plural, TV occurs alone in (16g), according to a pattern widely attested in northern Italian dialects (Baldi & Savoia, 2022), or it is followed by the personal inflection in other cases. In the Ladin varieties in (16), the third person exponents are ∅ in all conjugations, as in the case of the verb exemplified in (10) and (11), except in the first conjugation, where we find -a.

(16)	Declarative inflection of the present indicative
		1sg	2sg	3sg	1pl	2pl	3pl
a.	Müstair	∅      BRING NEAR!!	-aʃ	-a	-TV-ŋ	-TV-wat	-an
b.	Zernez	∅	-aʃ(t)	-a	-TV-n(a)	-TV-s	-ën
c.	Donat	∅	-as	-a	-TV-n	-TV-s	-ën
d.	Mulegns	∅	-es	-a	-TV-ɲ	-TV-ts	-aŋ
e.	La Pli de Mareo	-i	-es	∅/-a	-TV-ŋ	-TV-s	∅/-a4
f.	S. Cassiano Badia	∅	-ës	∅/-a	-TV-uŋ	-TV-s	∅/-a
g.	Corte	-e	-e	∅/-a	-TV-oŋ	-TV	∅/-a

In the inverted forms, the enclitic combines with the stem, as in the third singular person and the third plural person; the first person and third singular exponents can also combine with the root, replacing the inflection, as highlighted in (17). Generally, in the first and second plural persons, the specialized enclitic combines with the inflectional exponent. In what follows, ‘enclitics’ refers to the SCl forms in postverbal position.

(17)	Enclitics
		1sg	2sg	3sg	1pl	2pl	3pl
a.	Müstair	-i	-aʃ-t	-al/-la	-TV-n-i	-TV-wat	-i
b.	Zernez	-i	ëʃ-t	-al/-la	-TV-n-a	-TV-s	-i
c.	Donat		-as	-al/-la	-TV-n-za	-TV-s	-i
d.	Mulegns	-a	-t	-al/la	-TV-n-dza	-TV-ts	-iʎ
e.	La Pli de Mareo	-i	-te	-el/-er-a	-TV-n-ze	-TV-ze	-ai/-er-es
f.	S.Cassiano B.	-i	-s-t	-ël/-ëla	-TV-n-ze	-TV-s	-i/-lës
g.	Corte	-jo	-to	-elo/-ela	-TV-n-zo	-TV-zo	-eli/-ele

The comparison between (16) and (17) shows that enclitic paradigms systematically introduce a distinct characterization of the subject in correspondence with each person. This discrepancy suggests a different interpretive nature of the two types of realization of agreement. A similar difference separates enclitics in (17) from proclitic SCls in (18), which, in turn, appear to be less specialized for persons, save in the third person, where distinct forms generally occur.

(18)	Subject clitics (proclitics)
		1sg	2sg	3sg	1pl	2pl	3pl
a.	Müstair			al/la			i
b.	Zernez						i
c.	Donat						i
d.	Mulegns			l			i/ʎ
e.	La Pli del Mareo	i	te	al/ara	i	i	ai/ares
f.	S. Cassiano B.	i	të	al/ala	∅	∅	ai/alës
g.	Corte	∅	te	al/la	∅	∅	i/le

Examining a sample of subjective clitic paradigms of northern Italian dialects, Renzi and Vanelli (1983) suggested some (weak) implications, such that if there is only one SCl, it is the SCl of 2ps, and if there are two SCls, they are the 2ps and the 3ps. However, a much larger sample shows that similar implicational constraints can suggest tendencies at most, rather than strict regularities (Manzini & Savoia, 2005). An implicit functional logic would predict that non-distinct SCls or ∅ generally combine with specialized inflections and vice versa. For example, specialized SCls, such as al/ara and ai/ares of La Pli de Mareo in (18), seem to be related to reduced inflection in the third-person singular and plural. However, we find distinct inflections combining with distinct SCls, as in the case of the second person, mostly characterized by a specialized SCl and a specialized inflection. All in all, we do not expect a close correlation between SCls and inflections (including enclitics), insofar as they have a different syntactic status.

In Rhaeto-Romance varieties, where SCls and full pronouns are mutually exclusive, there is not always a different paradigm of proclitics and full pronouns, while enclitics generally appear more differentiated. If we focus on the third person, we notice that in some varieties there is a unique form, as in (19) for the varieties of La Pli and S. Cassiano.

(19)				full forms		proclitic		enclitic
	Mulegns:			el/ela		l/al/la		al/la
				els/elas		iʎ		iʎ
	Müstair:			el/ela		al/la		al/la
				es/elas		i		i
	La Pli de Mareo:			el/era		al/ara		el/era
				ei/eres		ai/ares		ai/eres
	S. Cassiano			al/ala		al/la		ël/ëla
				ai/alës		ai/alës		i/lës

However, the enclisis of a form such as eres ‘they.f’ in La Pli can combine with a postverbal subject, differently from its preverbal occurrence, as in the comparison between (20a) and (20b).

(20)	a.	tyes fies	dorm	a tʃaza
		your daughters	sleep.3pl	at home
		‘Your daughters sleep at home’
	b.	magari	dorm-eres	tyes fies
		maybe	sleep-SCl3pl	your daughters
		‘Maybe, your daughters sleep’
				La Pli de Mareo

We must conclude that the enclisis represents a special type of occurrence, which corresponds to Chomsky’s (2021, 2024a, 2024b) amalgamation of the pronoun with the verbal stem, as do other inflectional exponents.

5. Clitics and Inflections

The difference between clitics and inflections is a point widely discussed in typological, descriptive, and syntactic approaches (Anderson, 1992, 2005; Spencer & Luís, 2012). According to Anderson (2005, p. 79), clitics differ from inflectional affixes, which are morphemes that combine with the root in the formation of a word, in that they are ‘small words’ with a special syntax. Inflectional affixes are constrained by an obligatory morphophonological distribution5. Anderson (1992, p. 201 ff.; 2005) distinguishes simple clitics, i.e., the phonologically determined cliticization, from ‘special clitics’ that ‘display a distinctive syntax’. This is the case of pronominal clitics in Romance and Balkan languages. His idea is that word formation rules operate in the same way as inflection and special clitics by inserting phonological and morphological material in correspondence with the lexical nodes of the structure. In the case of clitics, configurational properties of the phrase, such as the argument structure, are involved.

In the syntactic theoretical framework, the difference between pronominal clitics and inflectional exponents is expressed by the fact that inflectional elements are subcategorized for the verbal root, while clitics occupy an argument site (XP or head positions) in the structure. Distributed morphology preserves a superficial distinction between clitics and inflectional affixes, although the featural properties of clitics make them substantially undistinguishable from other morphological material (Halle & Marantz, 1993, 1994). These authors identify morphology with an autonomous component, in which functional morphemes are not represented in syntax and are inserted via ‘Late insertion’ into a terminal node if its features are a subset of the features of the terminal node (Halle & Marantz, 1994). This allows the featural content of syntactic terminals to be manipulated by adjustment rules such as impoverishment, with the effect of obscuring the relation between syntax and interpretation. In this model, sub-word elements (affixes and clitics), labeled “dissociated morphemes” by Embick and Noyer (2001), have only an indirect and unclear relationship to syntactic information. This solution appears unnatural since there seems to be no principled reason why language should obscure meanings by complicating the relationship between the Conceptual–Intentional (C-I) and Sensorimotor (SM) systems.

In syntactic approaches, SCls of northern Italian dialects and French are inserted in their preverbal position, where they realize the φ-features associated with the EPP position of INFL/T (cf. Poletto, 2000; Manzini & Savoia, 2005, 2007). As for the inflectional morphemes, Baker’s (1988) mirror principle explains the formation of the verbal word by raising the verb to the closest suffix: V attaches to T, and then T-V moves to AgrS, which closes the complex word. The full form is the result of head movement in syntax. Manzini and Savoia (2011b) underline that the model allows for the interpolation of inflection and clitics. In fact, in the structural representation of the sentence, the head-raising of the root to C can leave free the positions to his right for clitics in-between C and I, where the inflectional exponent is inserted. As for OCls, some authors argue for the raising from their basic position within VP to functional heads in the domain of INFL (Kayne, 1991; Tortora, 2002). Another approach places them directly in specialized structural positions adjacent to the verb (Manzini & Savoia, 2005; Savoia & Baldi, 2023). The enclisis of OCls in imperatives and in infinitival contexts is generally derived from the high position of the verb in the C-field (Rivero, 1994; Mavrogiorgos, 2010; Manzini & Savoia, 2011b; cf. discussion in Savoia & Baldi, 2023).

In the approach of Harizanov and Gribanova (2019, p. 486), the word formation mechanism is implemented by amalgamation, understood as “results in the creation of a head-adjunction structure”. Amalgamation can be implemented at each point of the structure, i.e., T, v, V, and can be obtained by local dislocation, based on the linear precedence and adjacency of vocabulary items. This notion is referred to in the recent minimalist works of Chomsky, where it is connected to the externalization procedures.

In the Chomskyan model, conceptual and formal reasons exclude head movement, and, specifically, V-to-T raising is unjustified because ‘interpretation is the same whether a verb raises to INFL or stays in-situ’, and it is counter-cyclic (Chomsky, 2021, p. 30). So, Chomsky (2021, p. 36) assumes that the complex [INFL, [v, root]] externalizes the properties of the CP phase. Externalization, however, requires an operation that amalgamates the inflectional elements INFL and v along with the roots, presumably cyclically. The operation amalgamate forms a complex [INFL, [v, root]]. There are further effects with some variation, as is characteristic of morphology

According to Chomsky, both Internal Merge (IM) and External Merge (EM) ‘implement the simplest structure-building operation’, and ‘EM is associated with theta-roles and IM with discourse/information-related functions’ (Chomsky, 2021, pp. 15, 18). If we adopt the Chomskyan formulation, the notions of lowering and raising become meaningless because no head movement is allowed. It is the amalgamation that applies to lexical elements, combining them cyclically and creating complex words subject to Transfer and Spell-Out.

In our analysis, we stick to the Strong Minimalist Thesis (Chomsky, 2021, 2024a, 2024b), where, specifically, word formation excludes head movement. In keeping with Chomsky’s approach, we conceptualize categorizers, such as v and n, as the bundles of φ-features that characterize the functional content of words entering into the agreement operations (Manzini et al., 2020). If the objects constructed in syntax are mapped by Transfer onto interpretable representations, morphology is part of the syntactic computation, and lexical elements, including functional items, are endowed with interpretive content (Manzini & Savoia, 2011b, 2018; Savoia & Baldi, 2023). Agreement is the manifestation of the identity between referential features corresponding to the same arguments. Consequently, there is no autonomous morphological component, and morphology represents one of the realizations of the syntactic computation (Manzini & Savoia, 2018).

6. Merge and Amalgamation in Phases

As we saw in Section 2, SCls correspond to specialized pronouns for the realization of φ-features of the subject. Regarding the morphosyntactic status of clitic pronouns, Cardinaletti and Starke (1999) analyze them as deficient pronouns, lacking some features contributing to the content of full pronouns. The idea is that full pronouns, i.e., the pronouns that have the same distribution of nouns, like el/ela ‘he/she’ in Müstair, are endowed with case and other structural properties in addition to phi-feature, while clitics, for example al/la, would only include the latter.6 We do not believe that subject and object clitics lack case properties, as illustrated by the paradigm we find in different varieties. On the contrary, they form complex case systems distinguishing nominative (SCls) and, in the case of objects, oblique (dative) and accusative third-person forms, and a single objective form in first and second person clitics, as in (21) for La Pli de Mareo:

(21)	Clitics	nominative oblique				accusative		oblique/accusative
		SCls		i		le, la, i, les		mo, te, se
			to him/her/them			him, her, them		me, you, Reflexive

Clitics, including SCls, are subcategorized for the verb, while full pronouns are free and behave like nouns.

Before turning to the phasal analysis of subject-verb inversion—V2 and interrogative—let us consider the morphological properties of the forms involved in the alternations we have examined and their lexical nature. As an example, we propose, in (22), the functional morphemes and, in (23), the subcategorization/selection restrictions (vocabulary items in the sense of Halle & Marantz, 1994) governing the distribution of inflectional exponents and clitics of the La Pli variety in (10). We identify the third-person elements with N and analyze their endings as the standard nominal inflections: -a for the feminine singular, -i for the plural (subset), and -es for the plural in feminine contexts, where feminine refers to a subclass of nouns. The thematic vowel, TV, is the stressed exponent inserted between the stem and the person inflection that occurs in a subset of verb forms in Romance paradigms. According to Manzini and Savoia (2005) and Baldi and Savoia (2022), TV introduces a subject-bound variable “x”, as in (22a). In many varieties, the TV of 1pl has a specialized form, for example, jù- in the Ladin of La Pli de Mareo. The application of the subcategorization restrictions in (23) is governed by the Elsewhere principle.

(22)	Thematic Vowel			(23)
	a.	-jù TV - = x		a.	-ju- ↔ Root] __ 1pl Present
		-ì/à/èi TV- = x			TV ↔ Root] __2pl/Tense /Mood
		person inflections:
	b.	-i = subset		b.	i ↔ Root ]__
		-es = Interlocutor			es ↔ Root] __
		-n = 1pl			-n ↔ TV__
		-s = 2pl			-s ↔ TV__
		SCls
	c.	(-)i = subset		c.	i ↔ __ V
		te = Interlocutor			te ↔ __ V
		a(l)/ar- = N			a(l) ↔ __ V
		-es = subsetf			es ↔ Root N] __
		-a = f			-a ↔ Root N] __
		Enclitics
	d.	-te = Interlocutor		d.	-te ↔ es INFL] __
		-el/-er- = N			el/er- ↔ ([Root V] __)
		-ze = Interlocutor and pl			-ze - ↔ TV-(n 1pl]) __
		-es- = subsetf
							La Pli de Mareo

As (23a–d) indicate, we can treat clitics and inflections as subcategorized by obligatory morphosyntactic contexts. The most obvious difference is that SCls are selected in terms of the verb’s lexical category, while inflections and enclitics require more subtle information, such as the internal structure of the word and its categories.

In (22), some cases of syncretism occur, as is generally the case with the inflection and clitic paradigms in Romance languages. In the variety of La Pli, the exponent (-)i realizes the SCl of 1sg and 1/2pl, the plural of the 3pl, and the inflection of 1sg. Assuming that plurality is the basic interpretation of -i, according to a proposal by Manzini and Savoia (2011a, 2017b), plural morphology can be associated with the part–whole/subset predicate (cf. Chierchia, 1998), i.e., the inclusion relation [⊆], as in (24).

(24)	(-)i = [⊆]

This reading can be applied to other types of referents, such as 1sg, to the extent that the subset relation can introduce referents into the universe of discourse, as demonstrated by our dialects. Therefore, we only need to characterize (-)i in a single way to capture its distribution.

Our analysis partly follows Manzini and Savoia (2005) and explains the difference between SCl in northern Italian dialects and in Rhaeto-Romance varieties as a contrast dependent on the subcategorization in terms of the verb, as in (25a). OCl implies a similar constraint, governed by Elsewhere, in (25b). We note that the two constraints are slightly different: according to (25a), there is a class of pronouns that is selected in the context of the verb; according to (25b), OCl is inserted in a position adjacent to the verb.

(25)	a.	SClφ ↔ __ (OCl) [ Vφ
	b.	OCl → __ [ V

φ-features in INFL are realized both by the inflection, as in Italian, and by specialized pronouns selected by the verb, in (26). The SCl of 3sg la of Corte in (13), la tozata la dorm ‘The girl SCl sleeps’, is selected in the context of the verb. The subject agrees with INFL_φ, and it is inserted freely, as in (26b), where it can be a topic or a focalized element.

(26)	a.	alφ ↔ __ [ Vφ
		b.	la tozata	(C)    < laφ, dormφ >	[v	VR
						Corte

In languages where the co-occurrence of SCls and lexical subjects is excluded, only an expressed subject is required. Thus, in Rhaeto-Romance with SCls, the latter, while able to externalize INFLφ, are not obligatorily selected by the verb, though lexically limited to verbal contexts, as in (27). More precisely, in these languages, the verb is subcategorized for a syntactic element bearing [φ-features] E-merged to it. Therefore, these contexts may be occupied by full pronouns or lexical subjects, which, however, unlike SCls, can also occur in other positions. On the other hand, all northern Italian or Rhaeto-Romance clitics share the same requirement to be selected in the context of the verb; the Elsewhere condition implements (27) at the expense of (25a): The more specific requirement precedes the less specific one.

(27)	Verb ↔ [DP [φ-features]] (OCls) ___ INFLφ

As suggested by an anonymous reviewer, our solution somewhat recalls previous proposals concerning the separation between northern Italian SCls, which combine with lexical subjects, and French SCls, which are excluded by lexical subjects. Brandi and Cordin (1981) and Rizzi (1986) differentiate northern Italian SCls from French SCls based on position. The idea is that SCls that combine with lexical subjects are in INFL, essentially an autonomous realization of agreement properties of the sentence. The pronominal subjects of French are in Spec of INFL, like full pronouns in Germanic languages, whereby the alternation with lexical subjects is derived. Alexiadou and Anagnostopoulou (1998) assume a similar conclusion, whereby in Null-Subject languages, the verb morphology has the same properties as a full pronoun. In the case of SCls that combine with lexical subjects, such as those of northern Italian dialects, they satisfy a “functional node” instead of occurring on the lexical verb, as inflections (Alexiadou & Anagnostopoulou, 1998, p. 517).

Obviously, the exclusion of movement and the probe-goal mechanism leads to a reconceptualization of the EPP principle (Chomsky, 2021, 2024a, 2024b). Our proposal suggests that SCls that combine with lexical subjects are simply a class of pronouns that verbs mandatorily require. Substantially, in these languages, the lexical subject is not necessarily within the INFL phrase. In varieties where SCls are in complementary distribution with lexical subjects, we must conclude that the verb is subcategorized with respect to a D(P) element E-merged, as in typically Non-Null Subject languages.

These languages require the realization of the agreement features by an SCl or a full pronoun or a lexical subject in the INFL domain, as suggested in (28) for the sentence mio fre/el/al dorm ‘My brother/he sleeps’ of La Pli de Mareo.

(28)	Cφ		INFL	[v	VR
		mio freφ/elφ/alφ	[ dormφ
					La Pli de Mareo

Following Chomsky (2024b), in the sentence el/al mo kerd-a ‘He calls me’ in (29a), [[Root] INFL _φ] created by E-Merge realizes INFL by amalgamation. The object clitic is E-merged in the phase CP, in (29b), where it realizes the agreement of v and the sentence id derived. The subject, realized by the full pronoun el or the SCl al, is inserted by the requirement in (27), regardless of the nature of the pronominal element.

(29)	a.	el/al	mo kerd-a
		he/SCl3sg.m	me call-3sg
		‘He calls me’
	b.	C elφ/alφ < mo, [INFL kerdRoot -a1sg] > [vφ			[VP
					La Pli de Mareo

Summarizing the discussion in Section 6, as (23a–d) indicate, proclitics (SCls), enclitics, and inflections can be treated as items subcategorized with respect to obligatory morphosyntactic contexts. A subtle difference is that SCls are selected in terms of the verb’s lexical category, while inflections and only partially enclitics depend on the internal categories of the word. This point will be reformulated in the following Section.

7. V2 and Postverbal Subject in Rhaeto-Romance

In cartographic mapping frameworks, subject-verb inversion is generally linked to the verb’s movement toward the CP area (cf. Rizzi & Roberts, 1989; Cardinaletti & Roberts, 1991; Haiman & Benincà, 1992)7. The movement of the verb is triggered by the feature [+Q] of C or another position within its field, which functions as a probe that the verb, itself endowed with the feature Q, satisfies (Rizzi, 1996, 1997). Benincà (2013) hypothesizes a C field comprising different positions for the verb, interpolated by Topic and Focus, as in (30), where, in V2 contexts, the verb rises to the C⁰ position on the right.

(30)	[CPC0sub./rel. [Topic] [Op Focus/wh] V/C0wh [IP[(Subject)] (Benincà, 2013, p. 73)

Poletto (2000) proposes that the verb moves to a position within the C field and considers the possibility that the inverted SCl are inflections. This latter conclusion, already formulated in historical approaches (cf. Rohlfs, 1968), has been supported by several authors (cf. Zamboni, 1974; Sportiche, 1997).

In Rhaeto-Romance main clauses, subject-verb inversion is usually associated with topical elements on the left of the verb. A preverbal subject excludes the inversion even if it occurs in the topic fields, as in (31a) for La Pli de Mareo, (32) for Zernez, and (33a,b) for Müstair. However, the informant of La Pli does not exclude inversion when the subject precedes a topic, such as magari ‘maybe’ in (31b).

(31)	a.	to møt Topic [magari Topic]	veɲ
		your boy       maybe	come.3ps
		‘Maybe your son comes’
	b.	to møt Topic [magari Topic]	veɲ-el
		your boy    maybe	come-SCl3sg
		‘Maybe your son comes’			La Pli de Mareo

(32)	[ nus Topic]	dor’m-i-na	kwia
	we	sleep-TV-1pl	here
					Zernez

(33)	a.	[dumauŋ Topic] [kwel om Topic]	nu   veŋ
		tomorrow     that man	Neg  come.3ps
		That man will not come tomorrow’
	b.	[kwiʃta dona Topic][dumauŋ Topic/new information?]		nu veŋ
		This woman      tomorrow		Neg come.3ps
		‘This woman does not come tomorrow’
					Müstair

Nevertheless, this pattern also characterizes focalized subjects, as in (34a,b)–(35a), including indefinite elements, as in (34d), (36b), and (37b). wh- subjects show different solutions. In Ladin, in (34c) and (35c) subject-verb inversion occurs, whereas in Romansh, in (36a) and (37a) inversion is excluded8.

(34)	a.	[ to møt Focus ]	veɲ		endomaŋ	(no le mio)!
		your boy	come.3ps		tomorrow	not Art mine
		‘Your son will come tomorrow, not mine!’
	b.	[ el Focus ]	veɲ	endomaŋ	(no era)!
		he	come.3ps	tomorrow	not she
		‘He will come tomorrow, not her!’
	c.	[ke  Focus ]	veɲ-el		pa?
		who	come-SCl3sg		Prt
		‘Who is coming?
	d.	[degyɲ/valgyɲ  Focus ]	te        kerd-a
		nobody/someone	you      call-3sg
		‘Nobody/someone calls you’
						La Pli de Mareo

(35)	a.	al vaɲ	ala	nia	al
		SCl come.3sg	she	Neg	he
		‘She’s coming, not him!’
	c.	[ke  Focus ]	e-l	pa	ɲ-y?
		who	is-SCl3sg	Prt	come.PP
		‘Who has come?’
					S. Cassiano Badia

(36)	a.	[tʃi  Focus ]	veɲ?
		Who	come.3ps
		‘Who is coming?’
	b.	[nindʒyŋ Focus]	nu	maindʒ-a
		nobody	Neg	eat-3ps
		‘Nobody eats’
					Müstair

(37)	a.	[tʃi Focus]	klom-a mai? /veɲ?
		who	call-3sg me?/come.3sg
		‘Who calls me?/ Who comes?’
	b.	[neɟin Focus]	ai veɲ-i-u
		nobody	is come.PP-sg.m
		‘Nobody has come’
					Trun

In the varieties here examined, a specialized syntax occurs in which the third-person subject is postverbal, the verbal inflection is in the third-person singular, and the participle does not agree in number and generally realizes the masculine singular, as in (38). This type of syntax characterizes postverbal subjects in most northern Italian varieties, as discussed in Manzini and Savoia (2002, 2007). In some varieties, we find an SCl generally coinciding with a 3sg form, as in (38i) for Corte and (38ii) for La Pli de Mareo (Manzini & Savoia, 2005, pf. 2.9). In Romansh varieties, postverbal subjects are introduced by the SCl specialized for indefinite and impersonal structures, such as i(d) and i(ʎ), on (36iii–v).

(38)	i.	l	e	veɲ-u	i	tozatʃ	vs.	i tozatʃ    i	e	veɲ-u-s
		SCl3sg.m	be.3	come-PP	the boys			the  boys SCl3plm	be.3ps	come-PP-pl
		‘The boys have come’
										Corte
	ii.	al	e	ɲ-y	tyes  mitaŋ-s		vs.	ki mituŋs   e	ɲ-y-s
		SCl3sgm	be.3	come-PP	your	girls		these boys  be.3	come-PP-pl
		‘Your daughters have come’						‘These boys have come’
										La Pli de Mareo
	iii.	i	veŋ	kindalts			vs.	as kindalts	veɲ-aŋ	do:
		SCl3ps	come.3ps	boys				the boys	come-3pl	later
		‘Boys have come’						‘The boys come later’
										Müstair
	iv.	iʎ	e	vaɲ-e-u		umfans	vs.	iʎz umfans	ɪn	va’ɲ-i-s
		SCl.3ps	be.3s	come-PP-sg.m		boys		the boys	the boys	come-PP-pl
		‘Boys have come’						‘The boys have come’
										Donat
	v.	iʎ	ai	veɲ-i-u		(ils) afon-s	vs.	ils afons	ain	veɲ-i
		SCl.3ps	be.3sg come-TV-PP			the boys		the boys	be.3pl come-PP
		‘The boys have come’						‘The boys have come’
										Trun

Manzini and Savoia (2007, p. 64) relate these forms to the operator variable structure in (39), proposed by Chomsky (1981, p. 196) for Focus, whereby ‘the relevant subject clitic and/or verb inflection simply introduce the D sentential argument as a variable […] identified by the postverbal subject through what we can construe as a predication relation’.

(39)	there is an x such that x comes and x_the boys

We may expect that there is a link between the postverbal subject in these structures and V2 order, suggesting a connection with the clausal semantics level.

Morphosyntax in the Phases

V2 structures raise at least two problems: the status of the phrase in first position and enclisis of the postverbal subject. Excluding raising to Spec-XP, Chomsky (2024b, 7 ff.) proposes that I(nternal)-M(erge) and Copy deal with theta-marked elements occurring in a non-thematic position, associating them with the edge of the phase, where the two copies are in a c-command-configuration. The element at the phase edge is I-Merged and boxed, i.e., it is accessible for interpretation at the Conceptual–Intentional system, but it is segregated with respect to the derivation:

[…] the element E is IM-ed to the phase edge as being put in a box, separate from the ongoing derivation D. E must however be accessible to D at later phase levels for interpretation at the interfaces […] it is immune to theta-marking

(Chomsky, 2024b, p. 8)

Hence, clausal categories such as ‘force- and information-related (interrogative, topic, and focus)’ are associated with IM. IM does not distinguish between A- and A’-movement,, and takes place at vP- or CP edge-phase, also deriving (quantified/topicalized) first position subjects; IM-ed boxed DPs ‘will be accessed for instructions at later phases’ (Chomsky, 2024b, pp. 5, 9).

Taking all this into account, we deal with the element in the first position as IM-ed, i.e., segregated from ongoing derivation and boxed at the edge of C, where it is accessible at later phase levels for interpretation. Let us first consider the interrogative with the subject wh-, as in ke veɲ-el pa? ‘Who is coming?’ of La Pli di Mareo in (34c). The wh- is in a boxed position where it is separated from the derivation and interpreted at the C-I interface. The reference to the subject is realized in addition to the inflection by the enclitics through amalgamation. More precisely, the verb lexicalizes INFL and through amalgamation is united to the enclitic in [_INFL veɲ-el], as in (40), where the copy of ke in vP remains unexpressed.

(40)	DP	[C       [  INFL  [vP	[   DP
	ke	<veɲ,     el >	(ke)
					La Pli de Mareo

At the interpretive level, the agreement features of the verb must be coreferential with ‘who’, except to give rise to a deviant result. This analysis can be applied to the lexical subjects in the first position in the questions, as in (41i,ii), where it is in a boxed position.

(41)	i.	[ tya so]	dorm-era?
		your sister	sleep-SCl3sgf
		‘Your sister, does she sleep?’
				La Pli de Mareo
	ii.	[la pwob-a ]	dorm-la?
		the girl	sleep-SCl3sgf
		‘Does the girl sleep?’
				Müstair

In V2 contexts, the specialized inflection (enclitic) of the verb and a postverbal subject occur in the presence of a lexical element in first position. V2 contexts represent a particular case of the type of syntax in which informational properties are implemented.9

In the case of (42a), atlo dorm-ju-n-ze (nos) ‘We sleep there’, the amalgamation of the root and the inflection is cyclically applied in INFL, through (42b) and (42c), where it yields the complex form. The verb and the enclitic, in vP, are amalgamated into the complex form in (42d), while the lexical subject, nos ‘we’ in (42d), is in VP, its thematic domain. We note that the enclitic in vP is not prevented by PIC from amalgamating with the verb: the enclitic is associated with the phase-head, and amalgamation works on the basis of linear adjacency. We can conclude that the phrase atlo is boxed in the edge of C, where it realizes informational content.

(42)	a.	atlo dorm-ju-n-ze			nos
		there sleep-TV-1pl-SClpl			we
		‘We sleep there/it is us who sleep there’
	b.	< dormRoot, juTV > → [[dorm-] ju-TV]
	c.	< dorm-ju-TV-, nφ > → [[dorm-juTV-]nφ]
	d.		(C)  INFLφ	[vP     Encl	[VP   DP
		atlo	[ [dorm- Root-ju-n INFL]		zeφ]	nosφ
					La Pli de Mareo

What we see is that when there is no expressed lexical subject in the preverbal position, a specialized morphosyntax is realized.

In (43a), the element dë’do ‘later’ is dislocated in an informational position, as in (43c). The verb requires a more specialized inflection, provided by the enclitic exponent, while Maria ‘Mary’ is in vP. In these varieties, the pronouns with a single form, such as era ‘she’ (see the discussion concerning (19)), are available for amalgamation to the verbal root, as suggested in (43b), where era is in VP like other enclitic exponents.

(43)	a.	de’do	dorm-era	Maria	(cf. (12a))
		later	sleep-SCl3sg	Mary
		‘Mary sleeps later’
		b.	< [dorm INFL], era → [INFL [dorm] era]
	c.	DP	(C)  [  INFL	[vP Encl	[VP   DP
		de’do	< dorm,       eraφ >			Mariaφ
					La Pli de Mareo

Enclitics, as inflections, are compatible with lexical subjects, insofar as they do not interfere with the occurrence of the lexical subject, as in (43a). This accounts for the different behavior of the Sursilvan varieties, such as that of Trun, in (1), where the pronominal subject is not amalgamated and simply realizes the subject in VP, as in (44).

(44)	suenter	dorm-ën elas		‘Later they(f) sleep’
	DP	([C) [INFL	[vP	[VP DP
	suenter	dorm-ënφ		elas
					Trun

In (27), we have suggested that the preverbal subject of a declarative is able to license the φ-features of INFL, as in (45i,ii). More precisely, the EA is IM-ed in CP, and the phase head INFL is realized by the verb, which takes on inflectional features and agrees with the DP, yielding a correct derivation, as in (45iii).

(45)	i.	tya so		mandʒ-a
		your sister		eat-3sg
		‘Your sister is eating’
							La Pli de Mareo
	ii.	la pwoba		dorm-a
		the girl		sleep-3sg
		‘The girl sleeps’
	iii.	DP	(C)	INFLφ	[vP	[DP	[VP
		la pwobaφ		dorm-aφ		(la pwoba)
							Müstair

We have noted in (31b) that subject clitic inversion can also occur in contexts where the subject is, in turn, followed by another informational epistemic element, as in to møt magari veɲ-el! ‘Maybe, your son comes!’ This possibility suggests that the segregated IM-ed position can regard the subject; in these cases, we can find both the inverted structure, as in (46a), and the other structure, without inversion, as in (46b) (from (31a)).

(46)	a.	to møtφ	[magari]	C	…	veɲINFLφ… el	[VP
	b.	to møtφ	[magari]	veŋINFLφ [vP			[VP
							La Pli de Mareo

This solution recalls the structure where, in the presence of the subject wh-, interrogative inversion is not realized, as in Sursilvan varieties (Trun) (cf. (37a)), as in (47).

(47)	DP		[C       [  INFL		[vP	[ DP
		tʃiφ		veɲφ		(tʃi)				Trun

We must conclude that agreement between the subject and the verb is sufficient to give rise to a correct interpretation, regardless of the position of the subject.

This analysis excludes head movement, as already discussed, specifically that of the verb to a high position in C (cf. Chomsky, 2021, p. 36). The inverted order is due to the Internal Merge of a phrase in a boxed/segregated position, a solution which accounts for V2 in embedded clauses, as in (48).

(48)		V2 in embedded clauses
	i.	ʎ am on		detʃ	tʃe iʎs umfants viɲ-ën dumaŋ /tʃe dumaŋ			viɲ-iʎ	iʎs umfants
		SCl me	have.3pl	told	that the boys come-3pl tomorrow /that tomorrow			come-SCl3ps the boys
		‘They told me that the boys will come tomorrow/ that tomorrow the boys will come.
									Müstair
	ii.	es   m  aun      dit		tʃa	las pwobas dorm-an /tʃa in tʃombra  dorm-n-i				(as pwops)
		they me  have.3pl told		that the girls		sleep-3pl /that in room		sleep-3pl-SCl3pl the boys
		‘They told me that the girls sleep / that the boys sleep in the room’
									Müstair
	iii.	i m a		dit	ke to fre	veŋ	eŋdomaŋ /ke eŋdomaŋ  veɲ-el		to fre
		SCl me have.3pl told		that your brother come.3sg tomorrow /that tomorrow come-SCl3sgm your brother
		‘They told me that your brother is coming tomorrow/that tomorrow your brother is coming’
									La Pli de Mareo

Subject-verb inversion also appears in the consequent clause of hypothetical constructions, in (49a) for La Pli de Mareo and (50a) for Müstair. (49b) and (50b) show that the insertion of the lexical subject in the consequent clause is combined with the usual weak inflection on the verb.

(49)	a.		[ ʃ   al        veŋ]	(ʃe) te     kerd-ai
			if  SCl.3sgm  come.3ps	(if) you  call-SCl3pl
	b.			(ʃe) ei    te    kerd-a
				(if) they  you  call-3pl
			‘If he comes, they call you’
					La Pli de Mareo

(50)	a.		[ ʃa   el   riv-a  ]	(ʃi) ta klom-al
			if  he   arrive-3sg	(if) you call-SCl3sg
				(ʃi) el ta klom-a
				(if) he you call-3sg
			‘If he arrives, he calls you’
					Müstair

In both varieties, the hypothetical introducer, ʃe/ʃa ‘if’, can be doubled in front of the consequent clause. These constructions seem to question the hypotheses based on the movement of the verb in relation to illocutionary force properties.

8. Some Conclusions

We can schematize the data concerning V2 as in (51):

(51)	a.	A lexical DP/a pronoun/a SCl in preverbal position licenses the φ-features of
		INFL.
	b.	In the absence of the preverbal subject, the specialized enclitic paradigm
		is realized.

Based on Chomsky (2024b, p. 5), we consider that the structure Subject/SCl-verb realizes in the CP ‘secondary semantic properties’. In comparison with Subject/SCl-verb syntax, the enclitic structure includes a type of pronoun that is not only more numerous and systematic than SCls, but also more differentiated, as highlighted in (16) and (17). This supports the conclusion that enclitics are not some sort of (phonological) alternant to SCls. We have hypothesized that they are associated with vP and joined to the verb by amalgamation, also differing from canonical inflection inserted in INF. In other words, a reduced/weak kind of inflection is realized on the verb when a DP occurs as the subject in the CP Phase, as in (52a). In contexts of inversion, that is when the subject is in the VP or is not realized, a more complete and richer paradigm emerges, as in (52b).

(52)	a.	[CP	DPφ	[INFLφ	||	[vP
	b.	[CP	…	[INFLφ	||	[vP     encliticφ	[VP    (DPφ)

If the inflected verb and the subject belong to the same phase, the standard paradigm is inserted, as in (53a). If the inflected verb in CP and the subject belong to different phases, or the subject is not realized, the richer inflection is inserted, as in (53b):

(53)	Externalization
	a.	Weak agreement on the verb if the subject is within the phase CP.
	b.	Elsewhere, strong agreement (inversion) + subject.

Needless to say, inversion is not necessarily associated with an expressed postverbal subject. When the lexical subject occurs in vP phase, nothing prevents it from agreeing with the verb in INFL.

While generative literature generally treats subject-verb inversion as the result of the verb movement to the C field in order to check the uninterpretable features that trigger head-raising, we have followed the formulation of Strong Minimalist Thesis proposed by Chomsky (2020, 2021, ff.). A crucial point of our analysis of V2 in Rhaeto-Romance is that inversion phenomena are interpreted in terms of two types of subject–verb agreement depending on the phase domain. Specifically, agreement in the CP phase is associated with more limited inflection than when the subject is in the vP phase or is absent.