The Internal Structure of Causal Subordinators and the Attachment Site of Causal Clauses in the History of Italian

Abstract

This paper investigates the syntax and diachrony of Italian causal clauses introduced by perché, siccome, and poiché. Although often treated as near-synonyms in Contemporary Italian, these subordinators differ systematically in their syntactic distribution, interpretive properties, and diachronic development. We show that perché introduces central adverbial clauses, merged within the vP/TP domain, whereas siccome and poiché introduce peripheral adverbial clauses, merged in the left periphery. This structural split correlates with a cluster of diagnostics: only perché-clauses can occur within the scope of matrix focus, negation, or epistemic operators, and only they can function as fragment answers. Conversely, siccome- and poiché-clauses consistently outscope matrix operators and encode non-at-issue content. A diachronic study reveals that the internal and external syntax of causal clauses introduced by each subordinator has remained stable from Old Italian to the present. However, siccome- and poiché-clauses display different semantics, as they derive from non-causal constructions (they originate from comparative and temporal clauses, respectively). We argue that the contrasting behaviors follow from the structural composition of the subordinators.

1. Introduction

This paper investigates the syntactic behavior of adverbial causal clauses in the history of Italian by comparing three causal constructions introduced by perché, siccome, and poiché (‘because/since’). While these items are traditionally treated as near-synonymous causal introducers or complementizers, we show that their syntactic distribution differs in systematic and theoretically revealing ways. These differences can provide new evidence for how changes in the internal analyzability of complex markers shape their morphosyntactic behavior over time. Our central claim is that adverbial causal clauses headed by these markers are merged at different heights within the superordinate clause, and that these differences follow from the distinct internal morphosyntactic structure of the markers themselves (which, in turn, can be linked to their etymological origin). In other words, the syntax of causal complementizer phrases (causal CPs) is tightly connected to the diachronic and synchronic composition of the functional elements that introduce them.

The three causal markers perché, siccome, and poiché can be pre-theoretically described as “complex subordinate markers” that emerged diachronically from the fusion of at least two independent morphemes. Perché derives from the combination of the preposition per ‘for’ and the wh-pronoun che (ultimately derived from Latin quis, quid ‘who, what’), while siccome and poiché originate from an adverbial element—sì ‘so’ and poi ‘then’, respectively—followed by come ‘how/as/like’ and the aforementioned wh-pronoun che. Despite their divergent histories, all three markers currently introduce subordinate clauses expressing the cause of the relevant event, proposition, or speech act represented in the superordinate clause (Frenguelli, 2002; Dardano, 2010; for other languages see Sweetser, 1990; Charnavel, 2019; Frey, 2023). They thus operate on the three cognitive levels originally proposed in Sweetser (1990): they relate two propositions at the content or propositional level, providing an explanation for why the speaker believes a given proposition to be true in the real world (the epistemic or evidential level), or for motivating the performance of a speech act (the speech act level). However, as we show below, these causal CPs do not form a syntactically uniform class.

A closer look at Italian causal clauses reveals non-trivial asymmetries among perché-, siccome-, and poiché-clauses. The asymmetries we discuss here concern distribution, information-structural behavior, and syntactic diagnostics for clause typing. First, while siccome- and poiché-clauses may appear either before or after the superordinate clause, perché-clauses are restricted to postposing:

(1)	a.	[Siccome/Poiché/*Perché mi hai invitato], sono venuto a cena da te.
		siccome/poiché/perché me=have.2sg invited am come to dinner at you
	b.	Sono venuto a cena da te, [siccome/poiché/perché mi hai invitato].
		‘I came to have dinner at your place since/because you invited me.’

Second, only perché-clauses can function as fragment answers to ‘why/for what reason’ questions, indicating that their structural position and clause type make them eligible for ellipsis and extraction-related phenomena:

(2)	A: Per	quale motivo	sei	arrivato		tardi?
	for	what reason	are.2sg	arrived		late
	B: Perché/*siccome/*poiché		ero	in	palestra.
	perché/siccome/poiché		was.1sg	in	gym
	A: ‘For what reason did you arrive late?’
	B: ‘Because I was at the gym.’

Third, perché-clauses—but crucially not siccome- or poiché-clauses—may undergo focalization (3a), fall under the scope of focus-sensitive operators such as solamente ‘only’ (4a), or participate in cleft constructions (5a). The ungrammaticality of the corresponding sentences with siccome or poiché (3–5b) highlights deep differences in the syntactic integration of these clauses and in the feature composition of their complementizers.

(3)	a.	Sono	venuto	a cena	da te,
		am	come	at dinner	at you
	PERCHÉ		MI HAI INVITATO		E NON PERCHÉ NE AVESSI VOGLIA.
	perché		me=has invited		and not perché		of.it= had.1sg desire
b.	*Sono venuto a cena da te,
	SICCOME/POICHÉ MI HAI INVITATO E NON SICCOME/POICHÉ NE
	AVESSI VOGLIA.
	‘I came to have dinner at your place BECAUSE YOU INVITED ME, NOT
	BECAUSE I WANTED’
(4)	a.	Sono venuto		a cena	da te	solamente perché hai insistito tanto.
		am come		to dinner	at you only perché have.2sg insisted much
	b.	*Sono venuto a cena da te solamente siccome/poiché hai insistito tanto.
		‘I came to have dinner at your place only because you insisted much.’
(5)	a.	È perché hai insistito tanto che sono venuto a cena da te.
	b.	*È siccome/poiché hai insistito tanto che sono venuto a cena da te.
		‘It is because you insisted much that I came to have dinner at your place.’

In this contribution, we argue that the empirical contrasts in (1–5) can be systematically derived from the internal structure of the subordinate markers that introduce these causal clauses. Differences in morphosyntactic composition give rise to distinct functional heads, feature specifications, and structural positions within the higher clause, ultimately shaping the syntactic and interpretive behavior of each causal CP. By tracing the link between diachronic development, morphosyntax, and syntactic distribution, this study intends to contribute to a finer-grained understanding of clause-typing, the syntax of causal clauses, and the evolution of analyzability of causal complementizers in the Romance languages.

In Section 2, we describe the differences between the causal clauses introduced by the three markers in Contemporary Italian and propose that their contrasting behavior can be accounted for by assuming that they are merged in different structural positions within the main clause. In Section 3, we show that in the history of Italian, the main difference between perché on the one hand and siccome/poiché on the other is that the latter emerge as markers of different clause types and the causal meaning is a secondary development; however, their syntactic distribution remains stable. Finally, in Section 4, we examine the differences between the causal subordinators and argue that these differences reflect their internal structural composition.

2. Causal Clauses in Contemporary Italian

We focus here on causal constructions introduced by perché, siccome, and poiché (‘because/since’) in Contemporary Italian. All three markers introduce a subordinate clause that can encode event, epistemic, and speech act causality (Frenguelli, 2002; Dardano, 2010; for other languages, see Sweetser, 1990; R. K. Larson, 2004; R. Larson, 2022; Frey, 2012, 2023; Charnavel, 2017, 2019; Jędrzejowski & Fleczoreck, 2023).

(6)	a.	Le strade	sono ghiacciate	perché ha fatto molto freddo.
		the roads	are frozen	perché has done much cold
	b.	Poiché ha fatto molto freddo,			le strade	sono ghiacciate.
		poiché has done much cold			the roads	are frozen
	c.	Siccome ha fatto molto freddo,			le strade	sono ghiacciate.
		siccome has done much cold			the roads	are frozen
		‘Because it has been very cold, the roads are frozen.’

(7)	a.	Lucia deve essere a casa,	perché ci sono le luci accese.
		Lucia must.3sg be.inf at home	perché there=are the lights on
	b.	Poiché ci sono le luci accese,	Lucia deve essere a casa.
		poiché there=are the lights on	Lucia must.3sg be.inf at home
	c.	Siccome ci sono le luci accese,	Lucia deve essere a casa.
		siccome there=are the lights on	Lucia must.3sg be.inf at home
		‘Since the lights are on, Lucia must be home.’

(8)	a.	Andiamo a prendere	questo caffè,		perché insisti tanto.
		go.1pl to take.inf	this coffee		perché insist.2sg much
	b.	Poiché insisti tanto,		andiamo a prendere		questo caffè.
		poiché insist.2sg much		go.1pl to take.inf		this coffee
	c.	Siccome insisti tanto,		andiamo a prendere		questo caffè.
		siccome insist.2sg much		go.1pl to take.inf		this coffee
		‘Since you insist that much, we are going to have this coffee.’

R. K. Larson (2004) argues that when ‘because’ operates at the content level, as in (6), it introduces quantification over events and therefore interacts with adverbs of quantification, whether lexicalized or not. The matrix clause provides the restriction of the main event quantifier, while the causal clause contributes the scope of event quantification. Such causal clauses have been labeled eventuality-related causal clauses. In contrast, in (7), the content of the subordinate clause is not related to the propositional content of the matrix clause. Instead, the speaker takes the presence of the lights as sufficient evidence to infer that Lucia is at home. Since (7) involves epistemic reasoning, these clauses are epistemic causal CPs. The clause in (8) is a speech act causal clause, as it provides the cause for the speech act associated with the matrix clause.

Although all three markers can operate at the three cognitive levels, they do not do so uniformly. While judgments concerning (7) are relatively stable, the sentence in (8a) is more problematic: various speakers do not accept perché-CPs as speech act causal clauses. This variation already points to subtle differences among the three markers.

In what follows, we restrict our investigation to eventuality-related causal CPs introduced by the three subordinators and show that the syntactic distribution of perché/siccome/poiché causal CPs differs. We argue that these adverbial causal clauses merge at different heights within the host clause depending on the lexical properties of the subordinator. More precisely, we identify two merge positions: (a) one in the matrix vP layer for perché-CPs, and (b) one in the left periphery for siccome/poiché-CPs.

As noted in the Introduction, the three types of causal CPs differ with respect to focalization (examples 3–5). Only perché-CPs can be focalized, can fall within the scope of focus particles, and can be clefted; siccome- and poiché-CPs cannot. In addition, only perché-CPs can serve as fragment answers to ‘why/for what reason’ questions (example 2; Bianchi, 2000, p. 90).

Moreover, only perché-CPs can appear within the scope of negative operators. Interestingly, when the subject is a pre-verbal negative quantifier, two readings are available; the relevant one is reading (B), in which negation has scope over the perché-CP.

(9)	Nessuno	è venuto	perché faceva troppo caldo.
	nobody	is come	perché did too.much hot
	Reading A: ‘The reason why nobody came is that it was too hot.’
	Reading B: ‘It is not true that anyone came because it was too hot.’

Conversely, with siccome/poiché-CPs, only reading (A) is available, with the adverbial clause taking scope over the matrix negation. Interestingly, a similar pattern emerges for perché-CPs when the quantified subject appears in post-verbal position.

(10)	a.	Nessuno è venuto siccome/poiché faceva troppo caldo.
	b.	Non è venuto nessuno perché faceva troppo caldo.
	Only possible reading B: ‘The reason why nobody came is that it was too hot.’

Another difference concerns the possibility of occurring within the scope of the matrix epistemic adverb ‘probably’. Once again, while perché-CPs can fall within the scope of the adverb—yielding the reading ‘probably’ > perché—siccome/poiché-CPs outscope the modal adverb.

(11)	a.	Probabilmente Gianni è uscito in fretta e furia perché era in ritardo.
		probably Gianni is gone.out in hurry and fury perché was in lateness
		‘Gianni left in a hurry probably because he was late.’
	b.	Probabilmente Gianni è uscito in fretta e furia siccome/poiché
		probably Gianni	is gone.out in hurry and fury siccome/poiché
		era in ritardo.
		was in lateness
		‘Probably Gianni left in a hurry because he was late.’

These contrasts suggest that the three causal clauses are not alike. There is a clear split between perché-CPs, on the one hand, and siccome/poiché-CPs on the other. While the former can be used as fragment answers and can fall within the scope of focus, negation, and modal matrix operators, the latter cannot. We account for these differences in terms of merge height. Since perché-CPs fall within the scope of matrix operators—and scope relations are standardly captured in terms of c-command (Büring, 2005)—the examples above indicate that the matrix operators c-command perché-CPs but not siccome/poiché-CPs. We therefore can conclude that perché-CPs are merged in the specifier of a functional projection within the vP layer, crucially above the functional projection hosting post-verbal subjects (see ex. 10b), as independently argued in Sanfelici and Rodeghiero (2024). Conversely, siccome/poiché-CPs are merged above TP and above the projection hosting epistemic modals. A plausible position is the specifier of SceneSetting projection, which hosts non-focalized phrases (Benincà & Poletto, 2004), and, indeed, neither the clause nor its content can be focalized. The examples above show that the content of these clauses is not at issue: it cannot serve as a fragment answer. Since non-at-issue content cannot be fully focalized (Simons et al., 2016), the impossibility of clefting and of occurring within the scope of focus naturally follows. The position proposed for the two clauses naturally accounts for the lack of variable binding in the following examples.

(12)	a.	Poiché/Siccome suai mamma è arrivata in ritardo, [ogni bambino]i era triste.
		poiché/siccome	his mum	is arrived	in lateness	every child	was sad
	b.	[Ogni bambino]i era triste, poiché/siccome suai mamma è arrivata in ritardo.
		‘Every child was sad since their mother was late’

A quantifier subject can bind a pronoun positioned in another clause if and only if the quantified DP c-commands the pronoun. Thus, the data in (12) show that (at the relevant level) poiché/siccome-CPs are not in the domain of the quantified subject, plausibly merged at the TP level.

Hence, despite often being treated as near-synonyms and despite encoding an eventuality-related causal relation with the host clause, causal clauses introduced by perché, siccome, and poiché are fundamentally different, being merged in two distinct structural positions (13).

(13)	[ForceP … [SceneSettingP [AdvCPs siccome/poichéCPs] [FinP [EpistemicP
	[EpistemicP probabilmente [NegQP…[vP [perchéCPs] [vP ….]]]]]]]]

In Haegeman’s (2003) terms, perché-CPs are therefore central adverbial clauses, that is, adjunct clauses that structure the event with which they are associated and that are merged within the TP domain of their host clauses. Conversely, siccome/poiché-CPs are peripheral adverbial clauses, namely adjunct clauses that introduce background propositions and are merged above the host clause’s TP, thereby providing the discourse framework within which the proposition of the host clause is interpreted.

Notice that (13) refers to what Frey (2016) calls eventuality causal clauses. A slightly different conclusion must be reached in terms of merge positions when these three CPs encode a causal epistemic relation and a causal speech-act-related one. Indeed, none of the three CPs fall into the scope of a quantified subject when they encode epistemic (14) and speech act (15) relations.

(14)	a.	*Ogni bambinoi deve essere malato, perché la suai fronte scotta molto.
	every child		must.3sg be.inf ill perché the their forehead burns much
	b.	*Ogni bambinoi deve essere malato, siccome la suai fronte scotta molto.
	every child		must.3sg be.inf ill siccome the their forehead burns much
	c.	*Ogni bambinoi deve essere malato, poiché la suai fronte scotta molto.
	every child		must.3sg be.inf ill poiché the their forehead burns much
	‘Since their forehead is really burning, every child must be ill.’

(15)	a.	*Ogni paccoi è in ritardo,	perché chiedi		sempre il suoi tracciamento.
		every shipment is in lateness perché ask.2sg			always the its tracking
	b.	*Ogni paccoi è in ritardo,	siccome chiedi		sempre il suoi tracciamento.
		every shipment is in lateness siccome ask.2sg			always the its tracking
	c.	*Ogni paccoi è in ritardo,		poiché chiedi	sempre il suoi tracciamento.
		every shipment is in lateness poiché ask.2sg			always the its tracking
		‘Since you ask about their tracking, every shipment is late.’

Moreover, perché-CPs when encoding these two relations cannot occur in cleft structures nor can they be focalized, thereby paralleling the behavior previously described of siccome- and poiché-CPs (16, cf. 7–8).

(16)	a.	??/*È perché le luci		sono accese, che Maria è a casa.
		is perché	the lights	are on	that Maria is at home
		‘It is because the lights are on that Maria is home.’
	b.	*È perché insisti tanto, che prendiamoci questo caffè!
		is perché insist.2sg much that take.1pl=us this coffee
		‘*It is because you insist so much that we are going to get this coffee!’

Interestingly, epistemic-related causal CPs and speech-act-related ones differ with respect to Principle C. A Principle C violation is detected when the causal clauses encode an epistemic relation, but not when they encode a speech act relation. This contrast suggests that the former but not the latter are syntactically integrated within the host clause.

(17)	a.	??/*Maria	dovrebbe	averloi	lasciato ieri
		Maria	should.3sg	have.inf=him left yesterday
		perché/siccome/poiché Giannii era ridotto molto male.
		perché/siccome/poiché Gianni was reduced very bad
		‘Mary should have left him, since Gianni was in a very bad shape.’
	b.	Digliei-lo,	perché/siccome/poiché Giannii già lo sa.
		tell=him=it	perché/siccome/poiché Gianni already it=knows
		‘Tell him that, since Gianni already knows it.’

Hence, once the semantic relations of causality are considered, the typology of merge positions for the three subordinators should be enriched, yielding a rephrase of (13) as proposed in

Table 1. Types of Causality.

	Subordinators	Merge Positions
Eventuality Causality	perché	vP
	siccome/poiché	SceneSettingP
Epistemic Causality	perché/siccome/poiché	SceneSettingP
Speech Act Causality	perché/siccome/poiché	Non-integrated CPs

.1

We thus conclude that causal CPs in Contemporary Italian are a heterogenous class and that their syntactic behavior correlates with the level at which causality is computed, thereby confirming Frey’s (2016) typology. Only perché-CPs are central adverbial clauses, and they are so only when they encode an eventuality-related cause relation. Siccome-CPs and poiché-CPs are peripheral adverbial clauses when they encode both eventuality and epistemic cause relations. When the three subordinators operate at the speech act level, they are non-integrated adverbial clauses. The question we raise now is what regulates the different merge positions of perché and siccome/poiché when they encode eventuality-related causality. We will show that this distinction is rooted in the diachronic development of the subordinators, which, throughout the history of Italian, displayed the same properties observed in Contemporary Italian when used as causal complementizers.

3. Historical Data and Diachrony of Italian Causal Clauses

Whereas perché has consistently functioned as a subordinator introducing causal clauses since the earliest attestations (Frenguelli, 2002), siccome and poiché underwent a grammaticalization process.

Siccome is a univerbated form composed of two elements: the comparative–similative wh-pronoun come ‘how’ and the demonstrative adverbial pronoun sì (<Lat. sic) ‘so’. As shown in Sanfelici and Rodeghiero (2024), siccome changes significantly over the history of Italian, developing from a similative–comparative marker in Old Italian (18) to a causal subordinator in Contemporary Italian (19).

(18)

[Si-ccome

lo

vermine

consuma

il

legno e

so-as

the

worm

consumes

the

wood and

le

tarme

le

vestimenta]

così

consuma

la

invidia

the

moths

the

clothes

so

consumes

the

envy

il

corpo de-ll’

uomo…

the

body

of-the

man

‘As the worm consumes the wood and the moths consume the clothes,

likewise envy consumes the human body.’ (Anon., Fiore di virtù,

III; first half 14th cent.)

(19)	[Si-ccome	piove],	Gianni non verrà.
	So-as	rains	Gianni neg come.fut.3sg
	‘Since it rains, Gianni will not come.’

This change affected the morpho-phonology of the subordinator, its meaning, and the syntactic distribution of the clause. While in Contemporary Italian the causal meaning is expressed only by the orthographically univerbated form siccome, texts up to the 18th century exhibit both the univerbated form <siccome> and its non-univerbated variant <sì come>. Importantly, both forms introduced comparative–similative and causal CPs with no obvious differences until the end of the 17th century. Beginning in the mid-18th century, however, written sì come is attested only with a comparative meaning and ceases to encode a causal relation, which is expressed exclusively by the univerbated form siccome. Thus, in the 18th century, the division of the semantic space covered by the two forms—comparative vs. causal adverbial relations—is orthographically reflected.

Since siccome is formed from the adverb sì and the wh-item come ‘how’, Sanfelici and Rodeghiero (2024) conclude that sì underwent a shift from a free form to a bound morpheme. The orthography reflects this process (but notice that the doubling of /k/ is independently motivated by the ‘Raddoppiamento Sintattico’ phenomenon, according to which, in Tuscan and in Standard Italian, word initial consonants become long after a stressed syllable).

Regarding the semantics of the subordinator, siccome/sì come displayed almost exclusively a comparative–similative meaning in Old Italian, and this remained the most frequently attested value until the 17th century. From the late 17th century onward, however, occurrences with a causal interpretation increase, while those with a comparative reading decrease. This trend reaches a turning point in the mid-18th century. From that point on, instances of siccome/sì come with a causal meaning outnumber those with a comparative reading, and by the 20th century the causal value becomes the only one attested.

With respect to the syntactic properties and distribution of the clauses, Sanfelici and Rodeghiero (2024) observe that comparative and causal siccome/sì come-CPs have always differed. Two properties are crucial for our argument: (i) the position of the subordinate clause relative to the host clause, and (ii) the (im)possibility of being in the scope of focal operators and negation. Throughout the history of Italian, causal CPs most often—or almost exclusively—precede the host clause, whereas comparative CPs usually follow it. This positional asymmetry becomes particularly clear from the 16th century onward. From this period, occurrences in which causal siccome/sì come-CPs appear to the left of the host clause increase steadily, eventually constituting the overwhelming majority—and ultimately the entirety—of the corpus. Conversely, comparative siccome/sì come-CPs appear to the right of the host clause in roughly two-thirds of cases, and when they occur to the left, they are typically accompanied by a resumptive/correlative adverb such as così ‘so’, as in (18).

In addition, causal and comparative clauses have consistently differed in terms of (non-)at-issueness: the former are not at issue, while the latter may be at issue. As shown in the previous section, in Contemporary Italian, causal clauses introduced by siccome cannot (i) serve as fragment answers to ‘why/for what reason’ questions, or (ii) fall within the scope of focus particles or negation (Renzi et al., 2001; Frenguelli, 2002; Dardano, 2010). In these respects, they differ from causal clauses introduced by perché ‘because’. Likewise, siccome-clauses also differ from comparative clauses introduced by (così) come ‘how’, which can serve as fragment answers to ‘how’ questions (20a) and can fall within the scope of focus particles and negation (20b–c) (Renzi et al., 2001).

(20)	a.	A: Come hai cucinato il pollo?
		how have.2sg cooked the chicken
		B: (Così) come mi aveva consigliato mia mamma.
		so	as	me=had.3sg suggested my mum
		A: ‘How did you cook the chicken?’
		B: ‘As my mum suggested to me.’
	b.	Ho cucinato il pollo non (così) come mi aveva insegnato mia mamma (ma
		come diceva la ricetta).
		‘I cooked the chicken not as my mum told me (but as the recipe suggested).’
	c.	Ho cucinato il pollo solo (così) come mi aveva insegnato mia mamma.
		‘I cooked the chicken only as my mum told me.’

Sanfelici and Rodeghiero (2024) report zero occurrences of causal siccome/sì come preceded by negation or focus particles—specifically, by the items non ‘not’ and solo/soltanto/solamente ‘only’—across the history of written Italian. Conversely, the only instances in which the element appears within a focalization structure involve siccome/sì come functioning as a comparative subordinator, as in (21).

(21)

Li

oficiali

de-lle castella

de-l

detto

Comune

the

officers

of-the castles

of-the

said

Commune

non

possano

per

alcuno

modo

spendere

o

neg

can.sbjv.3pl

for

any

way

spend.inf

or

fare

spendere

o

diliberare

che

si

make.inf

spend.inf

or

approve.inf

that

refl=

spenda

per

inanzi,

o

che

si

dea

spend.sbjv.3sg

for

onwards

or

that

refl=give.sbjv.3sg

o

che

si

paghi,

alcuna

cosa

de-lla pecunia

or

that

refl=pay.sbjv.3sg

any

thing

of-the money

de-l

detto

Comune

se

non

[sì

come

si

of-th

said

Commune

if

neg

so

as

refl=

diliberrae

una

volta

o

più]…

approve.fut.3sg

one

time

or

more

‘The officers of the castles of the mentioned Commune cannot in any way spend or cause to be spent or approve that anything of the money of the mentioned Commune be spent henceforth, or given or paid, except as it shall be approved once or more […].’

(Ordinamenti provvisioni e riformagioni del Comune di Firenze; 1355–1357)

In conclusion, the syntactic distribution and properties of siccome/sì come-CPs, when encoding a causal adverbial relation, have remained substantially stable throughout the history of Italian. From the earliest attestations onward, the content of siccome/sì come-CPs was not at issue, suggesting that these clauses were merged in the left periphery of the host clause, i.e., in Spec,SceneSettingP. In addition, siccome/sì come-CPs predominantly—or even exclusively—preceded the host clause. Interestingly, this distributional preference persists in Contemporary Italian: for many speakers, the preverbal position is strongly preferred, and for some speakers it is the only acceptable position for causal siccome-CPs.

The same conclusions apply to the other subordinator that underwent a grammaticalization process, namely poiché ‘after that/since’. Poiché is composed of two elements: the adverbial form poi ‘after, then’ and the subordinator che ‘that’. In the earlier stages of Italian, poiché primarily introduced temporal adverbial clauses, which modified the temporal coordinates of the situation denoted by the main clause—i.e., they contributed to identifying the time of the eventuality in the main clause (22a). Over the course of the language’s history, however, causal poiché-CPs increased in frequency. In Contemporary Italian, it functions exclusively as a causal subordinator, and its original temporal meaning is no longer available (Patota, 2005) (22b):

(22)	a.	Rompi	le	loro	navi	[poi	che	l’
		break.imp.2sg	the	their	vessels	after	that	them=
		avrai	somerse].
		have.fut.2sg	sunk
		‘Break their vessels, after you sank them.’
		(Andrea Lancia, Compilazione della Eneide di Virgilio, 1316)

	b.	[Poi-ché	hai	una	sorella],	non	sei	figlio	unico.
		after-that	have.2sg	a	sister	neg	are	child	only
		‘Since you have a sister, you are not an only child.’

Hence, the subordinator underwent a shift from a temporal to a causal conjunction. As we saw for siccome/sì come, this change affected the morpho-phonology of the subordinator, its meaning, and the syntactic distribution of the clause. Following Sanfelici and Rodeghiero (2024), to investigate the diachronic development of the Italian subordinator poiché, we collected data from MIDIA (Morfologia dell’Italiano in DIacronia), a corpus of approximately 800 Italian texts of various genres, covering a period from the early 13th century to the mid-20th century.2 Data on Contemporary Italian were based on native-speaker introspective judgments and were further verified through the CORIS corpus3 and its later updates. While in Contemporary Italian the causal meaning is expressed only by the univerbated form poiché, both the univerbated poiché and its non-univerbated variant poi che are attested in texts from the 13th to the 17th century. Up to this period, both forms could introduce temporal and causal CPs, with no obvious differences detectable. From the 16th century onward, however, the use of the non-univerbated form decreases, while the univerbated poiché becomes increasingly frequent, eventually becoming the almost exclusive form in 20th-century texts. Crucially, strict adjacency between poi and che is obligatory in the non-univerbated form: no intervening material is ever attested. Unlike in the history of siccome, the development of poiché/poi che does not involve a morphological specialization of forms; rather, the non-univerbated variant simply disappears.

Regarding the semantics of the subordinator, poiché/poi che predominantly carries a temporal meaning (57%) in 14th-century texts, while the causal interpretation is less frequent, accounting for 30% of occurrences. However, as we progress through the centuries, this situation reverses. By the 15th century, the causal interpretation reaches 51% of occurrences. In the 17th century, the causal interpretation increases sharply to 91%, whereas both the temporal and ambiguous interpretations decline dramatically. This distribution remains substantially stable from the 17th to the 19th century. We also observed that the few temporal uses of poiché in 19th-century texts occur mainly in poetry and theater. In Contemporary Italian, poiché exclusively introduces causal adverbial clauses.

With respect to the syntactic properties and distribution of the clauses, Sanfelici and Rodeghiero (2024) show that temporal and causal poiché/poi che-CPs have always differed. As with siccome, the argumentation is based on specific properties: (a) the position of the subordinate clause relative to the host clause, and (b) the (im)possibility of occurring within the scope of focal, modal, and negative operators.

Both causal and temporal CPs may occur to the left or to the right of the host clause across all centuries. Temporal clauses occur more frequently in a pre-matrix position, and this trend remains stable over time, aside from the general decline in temporal CPs. Causal CPs, however, undergo a positional shift: while from 1250 to 1500 causal poiché-CPs occur both pre- and post-matrix at similar rates, the post-matrix position becomes predominant from the 17th century onward. Thus, from ca. 1600, the temporal and causal interpretations are typically reflected in their merge position, at least in terms of frequency.

Furthermore, while no occurrences of causal poiché/poi che-CPs within the scope of negation or focalizing adverbs are attested, only temporal poiché/poi che-CPs appear preceded by negation (23) or by focalizing adverbs such as eziandio ‘even’ and maximamente ‘especially’ (24):

(23)

Che per l’ avenire

niuno, […] possa essere messo

ad alcun

that for the future

noone can.sbjv.3sg be.inf put

to any

partito o in

alcun

squictino, che per qualunche modo si farà

party or in

any

ballot that for any manner refl= make.fut.3sg

per la dett’arte

o

nella

casa della

dett’arte

for the said art

or

in-the

house of-the

said art

se non

[poi che saranno

compiuti

e

passati cinque

if neg

after that be.fut.3pl

completed

and

passed five

anni

dal

dì

che

così sarà

ricevuto

years

from-the

day

that

so be.fut.3sg

received

‘In the future, no one […] shall be proposed for any position or included in any ballot, in any manner conducted by the said art or within its house, until five years have fully passed from the day of their registration.’

(Anon., Statuti dell’Arte dei medici e speziali di Firenze, 1532)

(24)	E eziandio	[poiché		fu	fatto	uomo] in		cotali	cose
	and even	since		was	made	man in		such	things
	continuamente		studiò			e	brigò
	continually		studied			and	worked
	‘And even since he was made man, he continually studied and worked
	in such matters.’
	(Bartolomeo da San Concordio, Sallustio volgarizzato, ca. 1300)

Likewise, only temporal poiché/poi che-CPs could be embedded within a conditional clause, modifying its eventuality (25).

(25)

[Se

non

ristituiscano

esso

avere

[poi

che

saranno

if

neg

return.sbjv.3pl

it

possession

after

that

be.fut.3pl

conventi]],

quello cotale

avere

sieno

tenuti

di

agreed

that such

possession

be.sbjv.3pl

required

of

restituire

con

tre

den.

per

libra

ogni

mese

return.inf

with

three

den(arii)

for

pound

each

month

‘If they do not return it after they will have agreed, it shall be required to return

it with three denarii per pound each month.’

(Anon., Lo statuto dell’arte della mercanzia senese, 1342–1343)

Conversely, no occurrence of causal poiché/poi che-CPs was found embedded within a conditional clause. When a conditional clause was present, causal poiché/poi che-CPs either preceded it or the conditional clause was embedded within the causal clause.

(26)	Poiché,	[se la	forza delle percosse	fosse		infinita],
	since	if the	force of-the blows	be.sbjv.pst.3sg		infinite
	doverebbe ogni percossa benche piccola, fare				effetto infinito.
	must.cond.3sg each blow even.if small make.inf				effect infinite
	‘Since, if the force of blows were infinite, then every blow, however
	small, ought to produce an infinite effect.’
	(Evangelista Torricelli, Lezioni accademiche, 1517)

Hence, Sanfelici and Rodeghiero (2024) concluded that only temporal poiché/poi che-CPs can occur within the scope of focalizing, negative, and modal operators, whereas causal poiché/poi che-CPs invariably outscope them. As observed for siccome-CPs, the syntactic distribution and properties of poiché/poi che-CPs encoding a causal adverbial relation have remained substantially invariant throughout the history of Italian. From the earliest attestations, the content of causal poiché/poi che-CPs is not at issue, suggesting that these clauses are merged in the left periphery of the host clause, namely in Spec,SceneSettingP, with remnant movement of the lower structure to a higher position.

A final property shared by causal CPs introduced by siccome and poiché concerns the licensing of higher adverbs and sentence connectives. While the discussion of poiché/poi che-CPs is based on the findings of Sanfelici and Rodeghiero (2024), data on siccome/sì come were collected following the same methodology as in Sanfelici and Rodeghiero (2024). The results show that, although low adverbs can be fronted into the left periphery in both types of subordinate clauses, higher adverbials—such as epistemic, evidential, and sentence-evaluative adverbs—appear exclusively in causal CPs. No occurrences of high adverbs are found in comparative–similative or temporal CPs.

(27)	e [siccome sicuramente	senza di detta Stella			non posson
	and since surely	without of said star			neg can.3pl
	giungere al porto],	tengono	pertanto in essa sempre fissi
	reach.inf at-the harbor	keep.3pl	thus	in it always fixed
	gl’occhi.
	the eyes
	‘Since they cannot surely reach the harbor without the
	above-mentioned Star, thus they always keep their eyes fixed on it.’
	(Anon., Statuti droghieri Roma, 1761)

(28)	[Poiché forse vero è	che	le Muse	non sono nemiche
	since perhaps true is	that	the Muses	neg are enemies
	degli esuli]…
	of the exiles
	‘Since perhaps it is true that the Muses are not enemies of exiles […].’
	(Ugo Foscolo, Discorso sul testo della Commedia di Dante, 1825)

Only causal poiché- and siccome-CPs occurring to the left of the host clause license sentence connectives that outscope the subordinate clause itself and are, conversely, interpreted within the host CP. No such connectives are attested in their comparative–similative or temporal counterparts.

(29)	tutto osservando con attentissima attenzione, e l’osservato esponendo con lodevol candore, parendomi questo uno di que’ punti mastri che vogliono essere a sommo studio trattati e discussi pel finimento di questa causa.

	Siccome adunque	l’Autore ne ha dato un semplice lampo
	siccome thus	the author of.it=has given a simple glimpse

con quel suo nobile esperimento, crederò di far cosa non ispiacente a’ lettori, se porrò loro sott’occhio un piccol fascio di osservazioni e sperienze riguardanti cotal materia da me intraprese in ambo i regni vegetabile ed animale.

‘Observing everything with great attention, and expressing what has been observed with praiseworthy candor, this seems to me to be one of those key points that deserve to be treated and discussed with the utmost care for the completion of this matter. Since the Author has thus given a simple glimpse of it with his noble experiment, I believe I will do something not displeasing to the readers, if I place before them a small bundle of observations and experiments concerning this matter undertaken by me in both the vegetal and animal kingdoms.

(Spallanzani Lazzaro, Saggio di osservazioni microscopiche concernenti il sistema della generazione, 1765)

(30)

Quindi è che se le parti de’’ loro scheletri compariscono bene spesso nella propria naturale posizione, talvolta ancora sono fuori di luogo e lontane per qualche tratto l’una dall’ altra. Questo incidente medesimo, giacchè vuolsi tener conto di tutto, comprova che essi non devono aver fatto lungo tragitto, altramente, disunite e smembrate che fossero state le loro ossa, si sarebbero in cotal guisa disperse che non si verrebbe più a capo di raccapezzarle.

	Poiché dunque gli elefanti e i rinoceronti soggiornavano un tempo sotto
	poiché then the elephants and the rhinoceroses lived.3pl one time under
	le latitudini dell’Europa,
	the latitudes of-the Europe

dovremo noi inferirne che questi paesi fossero allora tanto caldi quanto lo sono adesso quelli del tropico?

‘Therefore, if the parts of their skeletons often appear in their own natural position, sometimes they are also out of place and distant from each other by some distance. This very incident, since everything must be considered, proves that they cannot have traveled a long way, because, had their bones been disjointed and scattered, they would have been dispersed in such a way that it would have been impossible to gather them. Since, then, elephants and rhinoceroses once lived in the latitudes of Europe, should we infer that these regions were as warm as those of the tropics are now?’

(Giovan Battista Brocchi, Conchiologia fossile subappennina con osservazioni geologiche sugli Appennini e sul suolo adiacenti, 1814)

Notice that in the very same contexts, in Contemporary Italian, sentence connectives can be licensed in poiché/siccome causal CPs and can have scope over the entire host clause, as clearly illustrated in (31).

(31)	Sono tornata tardi tutte le sere;
	am returned late all the evenings
	siccome/poiché però mia mamma mi ha sgridato, stasera tornerò presto.
	siccome/poiché however my mum me=has scolded tonight return.fut.3sg early
	‘I’ve been going back home late every night. But since my mom scolded
	me, I’ll be home early tonight.’

In conclusion, causal siccome- and poiché-CPs have displayed remarkably consistent and stable internal and external syntactic properties from their earliest attestations to the present day. The univerbation processes undergone by both subordinators appear to be concomitant with their semantic developments, but ultimately do not affect the syntax of causal clauses in any substantial way.

The same conclusion can be extended to the last causal subordinator under discussion, namely perché/per che. Following the methodology adopted for siccome and poiché, we searched the MIDIA corpus for all morphologically related forms of perché listed in Frenguelli (2002). In the earliest stages of Italian, perché exhibited a rich set of variants. Alongside phonologically conditioned forms—such as perchéd, found before vowel-initial words—and the very rare perchéne, attested three times in the OVI and once in MIDIA (always introducing a nominalized interrogative clause), historical documents also record the non-univerbated form per che. Further variants include forms with an intervening demonstrative between the preposition and the subordinator, namely (im)perciò che and (im)per-ò che ‘for it that’, as well as the apocopated form ché, typically restricted to the right periphery of the host clause.

(32)	a.	la Fede Cristiana aperse loro la via e lasciolle venire,
		the faith Christian opened.3sg them the way and left=them come.inf
		perché s’accorse			che veniano molto sfrenatamente
		perché refl=realized.3sg			that came.3pl very unrestrainedly
		e con gran furore e con molte parole
		and with great fury and with many words
		‘The Christian faith opened them the way and let them enter, because it
		realized that they were coming unrestrainedly, with great fury and many
		words.’
		(Bono Giamboni, Libro de’ vizi e delle virtudi, XLII, ca. 1292)
	b.	Ma per ciò che le cose			che sono senza modo non possono
		but for it that the things			that are without measure neg can.3pl
		lungamente durare, io […] estimo […]
		longly	last.inf	I consider.1sg
		‘But since those things which are without measure cannot last long, I
		consider […].’
		(Giovanni Boccaccio, Decameron, Introduzione, 1349–1353)
	c.	Ancora si conviene guardare			da vento et da troppo lume,
		still refl= befits look.inf			from wind and from too.much light
		imperciò che inducono infertadi d’occhi et alcuna volta tolgono il vedere
		in.for.it that cause.3pl illnesses of eyes and some time take.3pl the sight
		‘Moreover, it is worth being wary of wind and too much light, because they
		cause illness to the eyes and sometimes blindness.’
		(Restoro d’Arezzo, La composizione del mondo, ca. 1282)
	d.	E però che questa ultima parte è lieve a intendere,
		and for.it that this last part is easy to understand.inf
		non mi travaglio di più divisioni
		neg me= deal of more subdivisions
		‘And since this last section is easy to understand, I do not operate more
		subdivisions.’
		(Dante, Vita nova, 32, 1293–1294)
	e.	Ohimè trista, ch’io sono tutta trambasciata
		alas unfortunate that I am all pained
		‘Alas unfortunate, because I am all pained.’
		(Franco Sacchetti, Trecentonovelle, XXVIII, 1399)

While perché is generally the most frequent form introducing causal clauses in the MIDIA corpus across all stages of Italian, the distribution of its historical variants varies significantly across authors, text types, and periods. For instance, ché is the most common causal subordinator in the Novellino and in Dante’s Convivio, and it is predominantly found in poetry. Although its frequency has decreased, ché remains marginally possible in Contemporary Italian. The subordinator (im)però che is very frequent in the earliest stages of Italian up to the 16th century and, in certain texts—such as the Vita nova—it constitutes the primary causal subordinator. The form (im)per ciò che is mostly restricted to prose, particularly scientific prose, and is the most frequently used causal subordinator in works by Restoro d’Arezzo, Zucchero Bencivenni, and in Boccaccio’s Decameron. Both (im)però che and (im)per ciò che decline steadily from the 16th century onward, dropping below 4% in 19th- and 20th-century texts and disappearing entirely from Contemporary Italian causal use. Finally, the non-univerbated form per che is frequently attested in early texts up to the late 15th century; by the 16th century, its frequency drops below 10%, and it vanishes completely in texts from the fourth period (1692–1840).

No syntactic differences emerged among the causal clauses introduced by the various items with respect to their positioning. Across all periods, causal clauses predominantly follow the host clause. Up to the 16th century, they may occasionally precede the host clause, but without any identifiable focal effects. In Contemporary Italian, causal perché-CPs can follow the host clause only, unless they are focalized. Additionally, these causal clauses may occur within the scope of matrix negation or focal operators, and they can function as fragment answers to why-questions. Relevant examples were found for perché, per che, perciò che, and però che, while no occurrences were retrieved for ché or for the composite form imperciò che in these configurations.

(33)	a.	Questo perché?	Non per altro se non perciò che in niuna cosa separata
		this why	neg for other if neg for.it that in no thing separated
		si può trovare quella perfettione che nelle molte insieme si ritrova
		refl=can find that perfection which in-the many together refl=finds
		‘Why this? For nothing else but because in no separated thing one can
		find that perfection which is found when things are together.’
		(Muzio Girolamo, Lettere, ca. 1551)
	b.	Per qual cagione ricercandosi la tragica commiserazione vi si
		for which reason searching=refl the tragic commiseration there=refl=
		meschia anco l’attrocità? Non per altro se non perché un istesso effetto
		mixes also the atrocity neg for other if neg perché a same effect
		a cose diverse paragonato, et atroce e miserabile chiamar si puote.
		to things different compared and dreadful and miserable name refl=can
		(A. Carriero, Breve e ingenioso discorso sopra l’opera di Dante, 1582)
	c.	et in pace hanno obtenuto altro, non			per che questo General
		and in peace have obtained other neg			for that this general
		sia meglio di quello,		ma per che questo non li impedisce […]
		is better than that		but for that this neg them forbid
		(Vittoria Colonna, Lettere, 1540–1546)
	d.	Ma s’io potesse, ben vi pregheria che ’l meo servir voleste ad altra cosa,
		but if I can well you=implore that the my service wanted to other thing
		madonna, sol però che faticosa m’è troppo questa, a far credere altrui.
		my-lady only for.it that hard to.me=is too.much this to let believe to.other
		(Cino da Pistoia, Rime, 1290–1330)

We notice that all the historical examples with focalization or sensitivity to negation involve causal clauses which operate at the content domain; i.e., they are thus eventuality-related causal clauses.

Finally, we found some occurrences of higher adverbials, like epistemic and evidentiality adverbs.

(34)	Ho stampato altre varie bagatelle,		ma non ve ne mando copia
	have printed other various bagatelle		but neg you=of.it=send copy
	perché probabilmente	non v’importeranno.
	because probably	neg you=concern.fut.3pl
	‘I printed various other bagatelle, but I will not send you a copy, because
	probably you are not interested.’
	(Francesco Orioli, Carteggio Francesco Orioli—Filippo Saveri, 1817–1827)

Differently from siccome- and poiché-CPs, we found no examples of perché-CPs containing vocatives or sentence connectives. Overall, the syntactic distribution and properties of causal perché-CPs appear remarkably stable across the centuries. The only detectable diachronic difference concerns their obligatory right-peripheral position in Contemporary Italian (unless focus movement applies), whereas in earlier stages right-positioning, though strongly preferred, was not mandatory. For reasons of space, we set this issue aside and leave a full explanation for future research.

When we compare the syntactic distribution and properties of the three subordinators and their diachronic development, striking similarities emerge both across subordinators and across historical stages.

Although the diachronic trajectories concerning their relative frequencies differ, siccome- and poiché-CPs consistently pattern alike with respect to both internal and external syntax and, crucially, display remarkable stability throughout the history of Italian. Causal clauses introduced by siccome and poiché have always behaved as peripheral adverbial clauses, i.e., adjuncts that introduce background propositions and provide the discourse framework within which the host clause is interpreted. Since we found no instances of causal clauses with these subordinators occurring within the scope of matrix focal, modal, or negative operators—while such configurations are attested for their comparative–similative and temporal uses—we conclude that causal siccome- and poiché-CPs have been merged above the host TP, plausibly in Spec,SceneSettingP, from their earliest attestations.

A specular behavior is exhibited by causal perché-CPs, which can occur within the scope of matrix focal and negative operators, at least when they express an eventuality-related cause. This suggests that eventuality-related perché-CPs are thus at issue. This property is diachronically stable, persisting unchanged in Contemporary Italian. Hence, causal clauses introduced by perché have always been central adverbial clauses, i.e., adjuncts that structure the event with which they are associated and merge within the TP domain of their host clauses.

Table 2. Properties of causal introducers.

	Siccome	Poiché	Perché
Sensitivity to negation, focalizing and modal operators	*	*	ok
At-issue content	*	*	ok
Merge position	SceneSettingP	SceneSettingP	vP
Type of Adverbial CP	Peripheral	Peripheral	Central

offers a summary of our findings.

Notice that Table 2 only considers the behavior of the three causal CPs when operating at the content and epistemic domain. While many occurrences in the corpus are uninformative with respect to the syntactic position, many others are also semantically ambiguous between an epistemic or speech act reading and between an epistemic or eventuality-related interpretation. This ambiguity especially arises when the subject of the host clause is first or second person singular and when the main predicate is an utterance verb, such as dire ‘to say/tell’ or domandare ‘to ask’. Despite these methodological caveats, it is interesting to note that very few instances of perché were coded as speech act causal CPs in comparison with poiché and siccome, and they were all very late, starting being attested from the end of the IXX century. We also acknowledge that in our examples we found no syntactic properties to discriminate between peripheral and non-integrated causal CPs. In addition, no syntactic properties were detected in the corpus which enabled us to individuate a different position for epistemic perché-CPs.

4. Discussion

The diachronic dimension highlights a stable and invariant syntax for the three Italian causal subordinators: the grammaticalized markers siccome and poiché consistently introduce peripheral causal clauses, whereas the non-grammaticalized perché has consistently introduced central causal clauses throughout the history of Italian when expressing an eventuality-related cause. While this conclusion may appear unsurprising for perché, which has functioned as a causal subordinator since its earliest attestations, it raises interesting questions for siccome and poiché. It is not immediately straightforward to reconcile their semantic change and morphological composition with the concomitant invariance of their syntactic behavior. More generally, it invites the question of why perché has always introduced central adverbial clauses (at least when encoding an eventuality-related cause), while siccome and poiché have always introduced peripheral ones (at least when encoding an eventuality-related or epistemic cause). We argue that the different external and internal syntactic properties exhibited by these two groups of subordinators are tightly dependent on the structural composition of the subordinators themselves.

Both siccome and poiché combine with adverbial items—sì and poi, respectively—which are not related to the vP layer. The item sì is a focalizing element that can modify different projections, including high CP ones as in (35) (cf. Poletto, 2014, and subsequent work).

(35)	E	parlando=mi	così,	sì mi cessò	la forte fantasia
	and	talking=me	so	so me=stopped	the strong fantas
	‘While he was talking to me like that, I stopped dreaming.’ (Vita Nova 98)

Both Benincà (1995) and Poletto (2005) have convincingly demonstrated that when occurring in the CP domain, sì is located in Spec,Focus: it is always adjacent to the inflected verb, and it occurs after left-dislocated topics, scene-setting temporal elements and hanging topics; furthermore, it only triggers proclisis in Tobler–Mussafia contexts, and subject inversion is found after sì.

Poi is originally a temporal/aspectual adverb, but already in Old Italian it also functioned as a discourse/topic marker, roughly corresponding to inoltre ‘moreover’ as in (36), and as a modal particle, i.e., “an element with a reduced categorial status with the purely pragmatic function of expressing the speaker’s attitude toward the utterance’s content” (Cognola & Cruschina, 2021, p. 89), as in (37).

(36)	I’	ho		poi	moglie,	e	figliuolo		c’
	I	have.1sg		then	wife	and	soon.dim		there=
	ha	quaranta		anni;	tre	battaglie	di	campo
	has	forty		years	three	battles	of	field
	ho		poi	fatte.
	have.1sg		then	made
	‘I have also a wife, and my son is forty years old. I have also fought three field battles.’ (Anon., Novellino, 1300 ca.)

(37)	Io	non	sono	poi	risoluto	di	partir	di	Roma.
	I	not	am	then	convinced	of	leave.inf	from	Rome
	‘I am not convinced to leave Rome, after all.’ (Annibal Caro, Lettere, 1507)

Interestingly, both lexical items were reinforced by additional functional material when occurring in the vP domain, a development already underway in Old Italian. Sì was replaced by così (<Lat. eccum sic), which became the dominant correlative in Old Italian and the only one in Contemporary Italian. Poletto (2014, pp. 28–29) noticed that while both sì and its reinforced variant così can modify an adjective, a participle or an adverb without any clear difference in meaning, only sì can occur in front of che and come and can occur in the left periphery without any detectable focalization mechanism.

Conversely, così accesses the CP layer only when it is contrastively focalized, like other low adverbs. Sì persisted as CP marker until approximately the 1840s, while it had already restricted its lower uses by around the XVII century, appearing mainly or only with prenominal lexically specific participles or adjectives. Poi, when used to introduce a temporal clause and functioning as a preposition, was replaced by dopo (<Lat. de post). Indeed, poi as a preposition is exceedingly rare in Old Italian, occurring mainly with a few calendric nouns (Ferroni & Sanfelici, 2026). Conversely, when modifying functional projections higher than those in the vP domain, poi has remained a TP/CP item up to Contemporary Italian. Both poi and sì have also been preserved within the causal subordinators until the present.

We therefore conclude that the high merge position of causal siccome and poiché clauses directly follows from the high merge position of the adverbial items composing the subordinators. The adverbial elements remain transparent when combining with che/come, and their original syntactic properties are still detectable in the internal architecture of the causal subordinator: (a) the adverbial merges high, and (b) it is inherently relational. Diachronically, causal siccome originates from a comparative clause, while causal poiché originates from a temporal clause of anteriority. Both items are inherently relational: siccome requires a second term of comparison, and poiché requires a second topic time to be ordered relative to the embedded topic time (Pitsch, 2016). Once reanalyzed as causal markers, these relational properties are preserved at the propositional level: siccome establishes a correlation or comparison between the embedded clause and the host clause, while poiché orders the host clause sequentially with respect to the embedded clause. Put differently, it is the high merge position of the adverbials themselves that determines the high attachment of the associated causal clauses.

This conclusion immediately raises the question of how to account for the high merge of perché-CPs encoding epistemic causality in the SceneSetting field, as proposed in Section 2. Differently from poi and sì, per is a prepositional head that requires a complement. Per cannot be moved without its complement, nor can it be stranded in any stage of Italian. In addition, the lexical item is not inherently relational: no second correlate is inherently required by the preposition. Hence, the high merge position of epistemic perché-CPs should result from a different mechanism from the feature specification of the lexical item forming the subordinator. Although we lack a comprehensive explanation for this, one possibility is that the high merge of epistemic perché-CPs is a derived one, obtained via a non-operator-like A’-movement from its low merge position. Alternatively, one plausible route is to propose that what we labeled epistemic causal perché-CPs are in fact central adverbial clauses but modify the content level of a host clause which contains a silent predicate like “to infer/to say”. If so, what we labeled the host clause should be analyzed as the complement of this silent predicate. Suggestive evidence comes from the different prosody of epistemic and eventuality-related causal CPs. While the latter can be uttered without an intonational pause at the border between the main and the subordinate clause, the former can only be pronounced with a pause in between. Interestingly, a pause is also mandatory in perché-CPs operating at the speech act level, for those speakers who accept them. The presence of this intonational pause may be tentatively interpreted as a phonological reflex of some hidden syntactic material or operations. Leaving the precise analysis of the merge position of epistemic perché-CPs for future research, in the case of perché the internal architecture of the subordinator is also still transparent. Only perché-CPs can be merged at the vP (when encoding an eventuality-related cause relation), since only the PPs headed by per, but not poi and sì adverbials, can be merged inside the vP (on this see Cinque, 1999; Schweikert, 2005; Ferroni & Sanfelici, 2025). The tight link between per-PPs and the vP can be further appreciated when considering perché-CPs operating at the speech act level. They were very infrequent and attested very late in the historical records. In addition, speaker variability was found in the acceptability of speech act perché-CPs. For some speakers, only when the verb of the speech act is lexicalized can perché-CPs be used, e.g., Andiamo a prendere questo caffè! Te lo dico perché insisti tanto ‘Let’s have a coffee! I say this because you are really insistent’, which suggests the need to anchor perché-CPs to the vP domain.

5. Concluding Remarks

This paper investigated the syntax behavior of adverbial causal clauses in the history of Italian by comparing three causal constructions introduced by perché, siccome, and poiché (‘because/since’). By looking at their diachronic and synchronic behavior, we showed that the three types of causal clauses have different internal and external syntaxes. We further demonstrated that these differences have remained stable from Old Italian to the present. When encoding eventuality-related cause relations, only perché-CPs were and still are central adverbial clauses, while poiché- and siccome-CPs were and still are peripheral ones. We captured their contrasting behaviors from the differences in the structural composition of the subordinators. More precisely, we claimed that the merge position of the adverbial causal clause directly follows from the merge position of the adverbial/prepositional items composing the subordinators. While the adverbials poi and sì are merged at the CP level, only PPs headed by per can be merged at the vP.

This account generates interesting predictions concerning possible and impossible diachronic developments. On the one hand, only one direction of change is expected: central adverbial clauses may be reanalyzed as peripheral ones. This type of change is cross-linguistically common (Jędrzejowski & Fleczoreck, 2023), whereas the reverse development is, to our knowledge, virtually unattested. On the other hand, such a reanalysis is constrained by the syntactic nature of the elements that compose the subordinator: a central adverbial clause can become peripheral only if the subordinator contains an adverbial item not restricted to the vP layer. Accordingly, the diachronic stability of perché—and, by extension, of per as a preposition—follows naturally: perché does not contain an adverbial component.4