2.1 The Pan-Romance Perspective

In Germanic languages, bare nominals are found in object and subject position, as in English (3)–(4) where the indefinite determiner some, more precisely a weak variant of it, often referred to in the literature as s’m (see Milsark 1977; McNally 2020), is possible but not obligatory in either position:

(3)

a.	I	drank	(some)	wine.
b.	I	picked	(some)	violets.

(4)

a.	(Some)	wine	is on the table.
b.	(Some)	violets	are blooming	in my garden.

Romance languages are different from Germanic languages (see Strobel and Glaser, this volume, for a discussion of partitive markers in Germanic). Delfitto and Schroten (1991) observe that while Dutch (like English) has bare nominals in both subject (5a) and object positions (6a), Spanish only has them in object position (6b) vs. (5b), while French disallows them in both positions (5c) and (6c):

Where bare nominals are not allowed, overt determiners must appear. Spanish and French present two different types of overt indefinite determiners: the plural form of the singular indefinite “one” and the so-called “partitive article” formed by a grammaticalized preposition de ‘of’ inflected with a definite article bleached of its definite meaning (Carlier 2007; Ihsane 2008; Carlier and Lamiroy 2014).³ Note that in object position, these determiners are in competition with bare nominals in Spanish but mandatory in French, which is the only Romance language that does not allow for bare nominals:

The subject/object asymmetry found in Spanish is widespread across Romance (Stark 2008a, 2008b, 2016; Carlier and Lamiroy 2018).

Variation is also found in the forms of the determiners: Portuguese, Spanish, Catalan and Romanian display no determiner with mass nouns, as illustrated in (9a-c) and (9f), and the plural form of the indefinite article “one” with plural count nouns, (10a-c) and (10f) (see Dobrovie-Sorin, this volume, for some discussion of Romanian). French and Italian display the “partitive article” with both mass nouns and plural count nouns. In French, (9d) and (10d), the “partitive article” is the only available form, while in Italian it covaries with bare nominals as illustrated in (9e) and (10e). Finally, in Romanian (9f) and (10f), we find the uninflected indefinite determiner nişte for both mass and plural count nouns, on a par with bare nominals and the plural “one” for count nouns:

These facts are well known in the literature on individual languages. What is less known is the fact that the definite article can (marginally) appear in nominals with indefinite interpretation, in all Romance languages except the most lateral ones, namely Portuguese and Romanian, as represented in (11b-e) with a modified mass noun “bottled water” in a generic sentence expressing a habit:

Since Romance languages express reference to kind with the definite article, it could at first sight be argued that the direct object in (11) refers to kind when it is introduced by the definite article. But this would be wrong for two reasons. First, the definite article is much more restricted in similar contexts with an unmodified plural count noun, as in (12), where only Italian and Catalan still allow for the definite article:

Second, this distinction is not found with the attitude predicates in (13)–(14), which can select a kind-referring object (see Laca 1990; Krifka et al 1995). In this case, all Romance languages require the definite article, except Portuguese which, however, does not rule it out:⁴

The contexts in (9)–(12) present different grammatical features that interact with indefiniteness. In (9)–(10) the mass noun “wine” and the count plural “violets” are the object of an episodic predicate in the past tense. In this context all languages, except French, have bare nouns. With mass nouns only Italian and Romanian display an overt alternative indefinite determiner. With plural count nouns all languages have an alternative. In (9)–(10) the definite article is an alternative only in Italian. In (11) the mass noun “water” is modified by a preposition or an adjective.⁵ This is probably what makes the definite article also possible in Spanish, Catalan and French, as well as in Italian. In (12), where the indefinite object is unmodified and under the scope of negation, the definite article is impossible in all the Romance languages, except Catalan where it is only marginal and in Italian where it is fully acceptable.

This short overview in the pan-Romance perspective suggests that French is the language that mostly requires overt determiners, while Italian is the language with the largest variation of forms, including the definite article generalized in the four indefinite contexts in (9)–(12).⁶ When more than one form is possible, the question arises about the conditions that govern the competition among the forms. We expect to find variation in the semantic interpretation and syntactic distribution across speakers, as well as across dialects and regional varieties of Italian.

2.2 Variation in Italo-Romance Dialects

Cardinaletti and Giusti (2018) analyze three ais maps (Jaberg and Jud 1928–1949; Tisato 2009), displaying indefinite nominal expressions, that is, map 637 ‘[to look for] violets’, map 1037 ‘[if there was] water’, and map 1343 ‘[go to the cellar] to take wine’, finding large variation and optionality among four possibilities, illustrated with Italian in (15)–(18).

In (15) we see a bare nominal, with no overt determiner:

(15)

a.

Ho	bevuto	vino.
have.1sg.prs	drunk	wine

b.

Ho	raccolto	violette.
have.1sg.prs	picked	violets

The same indefinite meaning can be conveyed by a definite article, as in (16), which is in principle ambiguous between definite and indefinite interpretation. For example, (16a) is appropriate in the context “In my whole life, I have drunk wine many times”, in which the predicate wine-drinking does not refer to any definite (quantity of) wine; and (16b) can easily be interpreted as referring to the activity of violet-picking, with no reference to previously mentioned violets:

(16)

a.

Ho	bevuto	il vino.
have.1sg.prs	drunk	the wine

b.

Ho	raccolto	le violette.
have.1sg.prs	picked	the violets

In (17), we find the so-called “partitive article”, formed with di and a definite article, typical of Gallo-Romance varieties.

(17)

a.

Ho	bevuto	del vino.
have.1sg.prs	drunk	pa.m.sg wine

b.

Ho	raccolto	delle violette.
have.1sg.prs	picked	pa.f.pl violets

In (18), also parallel to Gallo-Romance varieties outside Italy, we find the bare preposition di. Note that while (15)–(17) are attested in Standard Italian, (18) should be taken as a meta-representation of dialectal data. In fact, only Piedmontese varieties have bare di (Berruto 1974; Cerruti and Regis 2020, and the references therein):

(18)

a.

Ho	bevuto	di vino.
have.1sg.prs	drunk	of wine

b.

Ho	raccolto	di violette.
have.1sg.prs	picked	of violets

The four ways to express indefiniteness, illustrated in (15)–(18), correspond to plural and mass indefiniteness, confirming the general tendency displayed in Romance and Germanic for the two types of bare nouns to behave alike, and unlike singular count nouns. In Italian and all Italo-Romance varieties, singular count nouns display an obligatory indefinite article (19a), with no plural (19b) or mass (19c) counterpart. The only possible interpretation of (19c) is of a (countable) type of wine; for this reason, it is not given as ungrammatical, but it is indicated as #, which stands for unacceptable in the intended meaning (mass indefinite):

(19)

a.

	Ho	raccolto	una	violetta.
	have.1sg.prs	picked	a/one	violet

b.

*	Ho	raccolto	une	violette.
	have.1sg.prs	picked	one.f.pl	violets

c.

#	Ho	bevuto	un	vino.
	have.1sg.prs	drunk	a	wine

Following Cardinaletti and Giusti’s (2018) terminology, I will call the missing article in (15) zero, the definite article with indefinite interpretation in (16) art, the “partitive article” di+art in (17), and the indefinite determiner homophonous to the preposition di in (18) bare di.

Cardinaletti and Giusti’s (2018) detailed analysis of the ais maps 637, 1037, and 1343 reports the following areal distribution of indefinite determiners. The extreme northern and southern varieties favour zero; a large part of Italy displays a strong preference for art; few varieties of the North-West favour bare di; a compact area starting from Liguria and northern Tuscany, including the whole Emilia and Romagna and reaching the northern Marches favours di+art. This distribution can be captured by two independent isoglosses: a North–South isogloss favouring art, surrounded by an area (northern Piedmont, northern Lombardy, northern Veneto, southern Apulia, southern Calabria and the whole of Sicily and Sardinia) favouring zero; a West–East isogloss (from central Piedmont to the central-northern Adriatic regions ending in Ancona) favouring di, surrounded by an area where di is not used. Where the two isoglosses overlap (in southern Lombardy, southern Veneto, the whole of Emilia Romagna and the northern Marches), di+art is the favourite form. According to Cardinaletti and Giusti (2018), this is evidence for the independent categorial status of di and art, the former being a determiner in Specdp and the latter being the overt realization of functional nominal features (number and gender, see Picallo 1991) in D. I will not dwell on the formal analysis of the four forms, referring the interested reader to that work.

Interestingly, very few varieties display just one form. Most varieties have more than one. When several forms are available, it appears that one has core indefinite meaning, whereas the others specialise. With “core indefinite meaning” Cardinaletti and Giusti (2018) intend the most basic notion of indefinites, namely those nominals that do not presuppose the existence of the referent or any other special meaning. This is the case of the ais map 1037 ‘if there was water’, where the indefinite nominal is in the scope of a hypothetical operator. In the ais map 637 ‘to pick violets’ the particular verb-object collocation enforces the special meaning of “small quantity”, because violets are generally gathered in small bunches. The ais map 1343 ‘go to the cellar to take wine’ enforces a “salient” meaning, again due to the verb-object collocation, because a cellar is generally the place where wine is stored. This will be considered in more detail in Section 3.6.

The three ais maps all display weak indefinites with narrow scope in postverbal positions, that is, the referent of the indefinite object is not (necessarily) presupposed to exist (see Brasoveanu and Farkas 2016; McNally 2020). Furthermore, we know from the literature that there are other forms to express indefiniteness in Italo-Romance dialects competing with the four indefinite determiners found with mass and plural count nouns, illustrated in (15)–(18). They appear sporadically on the three ais maps studied by Cardinaletti and Giusti (2018). Dialects have many ways to express small quantity, mostly with quantity nouns selecting the preposition di ‘of’ and no article. This ranges from general ‘a little’, for example, un po’ di vino ‘a bit of wine’ (see ais map 1343, 523 Firenze), un poko de viole mamole ‘a bit of violets’ (see ais map 637, Crespadoro (VI)), to measure nouns selecting a type of substance, for example, un sorso d’ acqua ‘a gulp of water’ (see ais map 1037, left column) or istu pezzu de legname ‘this piece of wood’, (see ais map 538, 624 Rieti); or collective nouns, for example, ne mattso da viola ‘a bunch of violets’, (see ais map 637, 707 Lucera (FG)) or the grammaticalized cardinal ‘two’, for example, du viole bambele ‘two violets’ (see ais map 637, 590 Porto Santo Stefano (GR)).

Another form, reported in the dialectal literature, but not occurring in any relevant ais map is certo ‘certain’. It appears in some southern Italian dialects, as a genuine indefinite determiner. We find an example with a mass noun fieno ‘hay’ in (20a) and a plural noun kundi ‘stories’ in (20b):

(20)

a.

S’ era	corcato	mmiezo	a ccerto	fieno.	Neaples
be.3sg.pst	lying	on	to certain	hay

‘He was lying on hay.’ (Rohlfs 1968, 118)

b.

certi	kundi	Avezzano (L’ aquila)
certain	stories

‘some stories’ (Giammarco 1979, 141)

Certo with determiner function is present in all Italo-Romance varieties, as the mass singular / count plural counterpart of the marked indefinite determiner un certo “a certain”, which is parallel to its English counterpart (cf. Farkas 2002; Schwarz 2011).

In (21) certo occurs with a mass noun (21a) and a plural count noun (21b), but not with a singular count noun (21c), on a par with di and art above. Furthermore, with plural count nouns, it ambiguously specifies either the referent or the type of referent, as indicated by the two translations in (21b); with mass nouns it only refers to the type, (21a); with singular count nouns it only refers to a specific referent (21c):

(21)

a.

Mangio	solo	certa	pasta.
eat.1sg.prs	only	certain	pasta

‘I only eat a certain type of pasta.’

b.

Conosco	certe	persone.
know.1sg.prs	certain	persons

‘I know some people of a specific type.’ / ‘I know some specific people’

c.

Conosco	*(una)	certa	persona.
know.1sg.prs	a	certain	person

‘I know a specific person.’

Such a profusion of forms in the local dialects at the first half of last century witnessed by ais and the dialectal literature, raises questions regarding both the development of these forms in the modern dialects in contact with Standard Italian and the presence and status of these forms in regional (informal) Italian in contact with the local dialects. This paper is a first step of a large research project aiming to explore variation and optionality in the expression of indefiniteness in modern local dialects as well as in modern informal Italian.

3.1 Grammatical Function

The predicates in (22a) and (23a) select a kind-referring subject and are incompatible with an object-referring nominal. A subject with a zero determiner, as in (24a) and (25a), is excluded for the independent reason, seen for Spanish in (5b) and (7a) above, that in Romance languages, bare nouns in subject position are ungrammatical (cf. Dobrovie-Sorin, this volume; and Martin, Carvalho and Alexiadou, this volume, for Brazilian Portuguese examples where the subject is reinterpreted as an event type). But the fact that (24b) and (25b) with an overt “partitive article” are also ungrammatical shows that the definite article in (22a) and (23a) does not express indefiniteness but reference to kind. In (7) above, indefinite subjects in Spanish and French must display an overt determiner (unos in Spanish and the “partitive article” in French). This is also the case in Italian, which displays di+art, as shown by the episodic sentences in (26):

(24)

a.

*	Acqua	abbonda	in questa regione.
	water	abounds	in this region

b.

*	Dell’acqua	abbonda	in questa regione.
	pa.f.sg water	abounds	in this region

‘Water abounds in this region.’

(25)

a.

*	Zanzare	sono	molto diffuse	in questa regione.
	mosquitos	are	very widespread	in this region.

b.

*	Delle zanzare	sono	molto diffuse	in questa regione.
	pa.f.pl mosquitos	are	very widespread	in this region.

‘Mosquitoes are wide-spread in this region.’

(26)

a.

Dell’acqua	sta scorrendo	nella canaletta.
pa.f.sg water	is running	in.the pipe

‘Water is running in the pipe.’

b.

Delle zanzare	stanno ronzando	nella mia camera.
pa.f.pl mosquitos	are buzzing	in.the my room

‘Mosquitoes are buzzing in my room.’

Bare nominals can occur in subject position provided they are modified by postnominal or prenominal adjectives (or prepositional adjuncts). The contrasts in (27)–(28) confirm that zero can never appear in the subject of predicates selecting for kind, while it can appear in indefinite subjects, obeying some restrictions:

(27)

a.

*	Acqua fresca e pulita	abbonda	in questa regione.
	water fresh and clean	abounds	in this region

‘Fresh and clean water abounds in this region.’

b.

Acqua fresca e pulita	scorre	giù dalla montagna.
water fresh and clean	runs	down from the mountain

‘Fresh and clear water runs down from the mountain.’

(28) a. *Pericolosissime zanzare tigre sono diffuse in questa regione.

‘Very dangerous tiger mosquitoes are wide-spread in this region.’

b.

Pericolosissime zanzare tigre	ronzavano	nella mia stanza.
Very dangerous mosquitoes tiger	were-buzzing	in.the my room

‘Very dangerous tiger mosquitoes were buzzing in my bedroom.’

In episodic sentences with positive or negative polarity (29)–(30), the object can be indefinite, as shown by the fact that it can be introduced by zero or di+art. This context also allows for art, compare (29) and (22b), (30) and (23b):

(29)

a.

(Non)	ho	versato	acqua	nel bicchiere.
neg	have.1sg	pour	water	in the glass

b.

(Non)	ho	versato	dell’acqua	nel bicchiere.
neg	have.1sg	be.1sg	pa.f.sg water	in the glass

‘I poured / didn’t pour water in my glass.’

(30)

a.

In questa stanza	(non)	ci	sono	zanzare.
in this room	neg	there	are	mosquitoes

b.

In questa stanza	(non)	ci	sono	delle zanzare.
in this room	neg	there	are	pa.f.pl mosquitoes

‘In this room, there are (no) mosquitoes.’

Note that a definite referential interpretation of the objects in (22b) and (23b) is also possible, as they can be synonymous to the sentences in (31), where the object is modified by a relative clause in the indicative, which enforces definite interpretation, parallel to what we find in (22c) and (23c):

(31) a. Ho versato nel bicchiere l’acqua che era nella tua tazza.

‘I poured in my glass the water that was in your cup.’

b. In questa stanza ci sono le zanzare che mi hanno punto.

‘In this room there are the mosquitoes that bit me.’

Kind-referring objects can be found in the object position of attitude verbs, such as “love” or “hate” (cf. Laca 1990 for English and Spanish; Anscombre 2001 for French) but cannot be the object of consumption verbs such as “eat” or “drink”. This is shown by the contrast between les bananes in (32a) and des bananes in (32b):

(32)

a.

Les singes	aiment	les bananes.
the monkeys	love	the bananas

‘Monkeys love bananas.’

b.

Les singes	mangent	des bananes.
the monkeys	love	pa.pl bananas

‘Monkeys eat bananas.’

In Italian, the contrast is only partially replicated with a major difference, namely, that the article is mandatory in the object of attitude predicates (33), whose object can refer to kind, and optional in the complement of consumption verbs (34), whose object cannot refer to kind as confirmed by the contrast with French (32b), and the pan-Romance overview in (11)–(14) above:

(33)

a.

Detesto	*(il) café.
hate.1sg.prs	the coffee

‘I hate coffee.’

b.

Detesto	*(le) ciliegie.
hate.1sg.prs	the cherries

‘I hate cherries.’

(34)

a.

Bevo	(il) café.
drink.1sg.prs	the coffee

‘I drink coffee.’

b.

Mangio	(le) ciliegie.
eat.1sg.prs	the cherries

‘I eat cherries.’

To conclude, the object position is the most reliable grammatical function to study the variation among bare nominals and overt indefinite determiners. This is because bare nominals cannot appear in subject position unless further modified by an adjective (or other adjuncts). In object position the different determiners give different flavours to the notion of indefiniteness to be better defined in the rest of this section. The kind-interpretation of art can be safely excluded avoiding the predicates that select kind-referring objects, such as attitude verbs. Note that the definite referential interpretation is always possible and must be excluded with appropriate means to be discussed in the following sections.

3.2 Scope

As noted by Chierchia (1997), in Standard Italian, bare nominals in direct object position only have narrow scope, while nominals with the “partitive article” may be ambiguous in the plural between narrow and wide scope, like the ones with a singular indefinite article (see also Zamparelli 2008).

(35)

a.

Non	ho	invitato	ragazzi.	*ⱻ ¬ / ¬ⱻ
neg	have.1sg.prs	invited	boys

‘I didn’t invite boys.’

b.

Non	ho	invitato	un ragazzo.	ⱻ ¬ / ¬ⱻ
neg	have.1sg.prs	invited	a boy

‘I didn’t invite a boy.’

c.

Non	ho	invitato	dei ragazzi.	ⱻ ¬ / ¬ⱻ
neg	have.1sg.prs	invited	pa.m.pl boys

‘I didn’t invite boys.’

This is supported by the fact that, while (35a) is only compatible with the continuation in (36a), (35b-c) are compatible with both continuations in (37) and (38) respectively:

(36)

Non	ho	invitato	ragazzi
neg	have.1sg.prs	invited	boys

a.

ma	solo	ragazze.	¬ⱻ
but	only	girls

b.

#	perchè	erano	antipatici.	ⱻ ¬
	because	be.3pl.impf	unpleasant

‘I didn’t invite boys but only girls / # I didn’t invite boys because they were unpleasant.’

(37)

Non	ho	invitato	un ragazzo
neg	have.1sg.prs	invited	a boy

a.

ma	una	ragazza.	¬ⱻ
but	a	girls

b.

perchè	era	antipatico.	ⱻ ¬
because	be.3sg.impf	unpleasant

‘I didn’t invite a boy but a girl. / # I didn’t invite a boy because he was unpleasant.’

(38)

Non	ho	invitato	dei ragazzi
neg	have.1sg.prs	invited	pa.m.pl boys

a.

ma	solo	ragazze.	¬ⱻ
but	only	girls

b.

perchè	erano	antipatici.	ⱻ ¬
because	be.3pl.impf	unpleasant

‘I didn’t invite boys but only girls / I didn’t invite some boys because they were unpleasant.’

Cardinaletti and Giusti (2016) point out that in the dialect of Ancona, the “partitive article” can only have wide scope. This is consistent with the observation that the “partitive article” is not witnessed in point 539 (Ancona) in the ais maps 637, 1037 and 1343, since the noun phrases illustrated in these maps only have narrow scope:

(39)

Nun	ho	‘nvitato	dei fioli7
neg	have.1sg.prs	invited	pa.m.pl boys

a.

#	ma	solo	le fiole.	* ¬ⱻ
	but	only	the girls

‘I didn’t invite boys but only girls.’

b.

perché	erane	‘ntipatici.		ⱻ ¬
because	be.3pl.impf	unpleasant

‘I didn’t invite some boys because they were unpleasant.’

Cardinaletti and Giusti (2016) further claim that mass nouns can only have narrow scope, at least in Italian, as shown in (40):

(40)

Non	ho	bevuto	del vino,
neg	have.1sg.prs	invited	pa.m.sg boys

a.

ho	bevuto	solo	acqua.	¬ⱻ
have.1sg.prs	dunk	only	only

‘I didn’t drink wine. I only drank water.’

b.

#perché	era	acido.	*ⱻ ¬
because	be.3sg.pst	sour

‘I didn’t drink wine because it was sour.’

The two claims, that di+art with mass nouns can only have narrow scope⁸ and that in the dialect of Ancona it can only have wide scope, correctly predict that mass nouns cannot be introduced by the “partitive article” in this dialect. The two claims also predict that in those Italian varieties that allow the “partitive article” with mass and plural nouns, the “partitive article” is ambiguous between wide and narrow scope.

3.3 Noun Classes

The different properties of mass and count nouns lead us to the third feature, that is, a finer-grained distinction in the classification of nouns, in particular with regard to their being mass or count.

It is well-known that abstract nouns, such as “courage” and “talent”, behave differently from both singular mass and plural count nouns (cf. Tovena 2001). We will not consider them here. We introduce here a less known distinction between singular mass nouns such as pasta and plural nouns that can or must be conceptualized as mass, such as spaghetti and spinaci ‘spinach’ respectively.

The grammaticalized cardinal due ‘two’ (devoid of cardinal interpretation) can appear with genuine plural count nouns such as amici in (41a) as well as with plural mass nouns such as spaghetti (41b) and spinaci (41c):

(41)

a.

Ho	invitato	due	amici.
have.1sg.prs	invited	two	friends

‘I invited two friends.’ / ‘I invited (a small number of) friends.’

b.

Ho	preparato	due	spaghetti.
have.1sg.prs	prepared	two	spaghetti

‘I cooked (some) spaghetti.’ / ‘I cooked two strings of spaghetti.’

c.

Ho	raccolto	due	spinaci.
have.1sg.prs	picked	two	spinach.m.pl

‘I picked (a small quantity of) spinach. / *I picked two pieces of spinach.’

Other low cardinals can be grammaticalized to the point that they occur in idiomatic expressions, for example, tre soldi (lit. three coins, ‘little money’), quattro gatti (lit. four cats, ‘few people’). Higher cardinals do not have this possibility and can only be interpreted quantificationally.

The quantitative interpretation of cardinals allows us to detect a difference between what we may define as “semantically mass” plural nouns like spinaci in (41c), which can hardly be enumerated, from nouns like spaghetti (41b), which are ambiguous between the “semantic mass” and the regular plural count interpretation (cf. Gerards and Stark, this volume, on mass determiners and nominal plural marking). In fact, with a high cardinal such as ‘twelve’, which only has the quantitative interpretation, unambiguously semantically mass nouns are not allowed, cf. *dodici spinaci ‘twelve spinaches’, while ambiguous nouns only have plural interpretation, cf. dodici spaghetti ‘twelve [strings of] spaghetti’.

Grammaticalized low cardinals are used as indefinite determiners in all Italian dialects and regional varieties and provide a good alternative to the definite article, which, as pointed out in Section 3.1 above, is three times ambiguous. This is particularly relevant in the study of those varieties that do not allow for the “partitive article” and / or for bare nominals. Grammatically plural but unambiguously semantically mass nouns allow us to distinguish between the determiner and quantitative functions of low cardinals.

3.4 Aspect (Telic / Atelic)

Since Verkuyl (1972, 1993), the interpretation of object-referring nominals is strictly related to the aspect of the sentence. The canonical test that allows us to distinguish between telic and atelic (or resultative) aspect is the compatibility with adverbials such as ‘in an hour’ and ‘for hours’ respectively. De Swart (2006) reports that Bosveld-de Smet (1998) uses this test as a diagnostic to show that des-NPs in French correlate with atelic interpretation (42), unlike singular un ‘a’ and plural les ‘the’, which correlate with telic interpretation (43) (cf. Ihsane, this volume, for a discussion of des-NPs in telic sentences):

(42)

a.

Marie	a cueilli	des fraises	pendant des heures.
Marie	has picked	pa.pl strawberries	for pa.pl hours

‘Mary picked strawberries for hours.’

b.

*	Marie	a cueilli	des fraises	en une heure.
	arie	has picked	pa.pl strawberries	in an hour

(43)

a.

*	Marie	a cueilli	une fraise / les fraises	pendant des heures.
	Marie	has picked	a strawberry / the strawberries	for pa.pl hours

b.

Marie	a cueilli	une fraise / les fraises	en une heure.
Marie	has picked	a strawberry / the strawberries	in an hour

‘Marie picked the strawberries in an hour.’

Italian, once again, is different from French in the distribution of the “partitive article” and the definite article, although it behaves like French with respect to the distribution of the indefinite singular article.⁹

In Italian, di+art is almost unacceptable with atelic aspect (44a) and fully possible with telic aspect (44b).¹⁰ In (45) art is possible with both aspects, while zero correlates with atelic aspect:

(44)

a.

Maria	ha raccolto	(??delle) fragole	per un’ ora.
Maria	has picked	pa.f.pl strawberries	for an hour

‘Mary picked strawberries for an hour.’

b.

Maria	ha raccolto	delle fragole	in un’ ora.
Maria	has picked	pa.f.pl strawberries	in an hour

‘Maria picked strawberries in an hour.’

(45)

a.

Maria	ha raccolto	(le) fragole / *una fragola	per un’ ora.
Maria	has picked	the strawberries / a strawberry	for an hour

‘Maria picked strawberries for an hour.’

b.

Maria	ha raccolto	*(le) fragole / una fragola	in un’ ora.
Maria	has picked	the strawberries / a strawberry	in an hour

‘Maria picked the strawberries / a strawberry in an hour.’

The definite article is not ambiguous in (45). With atelic aspect (45a), le fragole ‘the strawberries’ is synonymous to the bare nominal fragole ‘strawberries’; with telic aspect (45b), le fragole is a referential definite plural noun phrase.

3.5 Clause Type

According to Krifka et al. (1995), generic sentences do not necessarily have arguments that refer to kind, as observed in (11) and (12) above, and, vice versa, kind-referring nominals may be the arguments of non-generic sentences: potatoes in (46a) is the kind referring subject of an episodic sentence; a potato in (46b) is the indefinite singular nominal subject of a generic sentence:

(46) a. Potatoes were introduced into Ireland by the end of the 17th century.

b. A potato contains vitamin C, amino acids, protein and thiamine.

We have already observed in Section 2.1 that in Romance languages the definite article introduces kind referring nominals, and only in Italian it can generally introduce indefinite nominals (we call this indefinite determiner art). Cardinaletti and Giusti (2020) present the results of a questionnaire on the expression of indefiniteness in colloquial Italian and find that art is much more frequently used in generic negative sentences in the present, such as ‘I don’t eat meat’ and ‘I don’t drink wine’, than in episodic positive sentences in the past, such as ‘We ate meat’ and ‘We drank wine’. Furthermore, in generic sentences di+art is never present, unlike what is found in episodic sentences, where it is in competition with zero and art. We will get back to this in Section 4.2.

Polarity is another sentential feature interacting with indefiniteness. Sentence negation allows us to check for the scope of the indefinite object, as observed in (39)–(40) above (cf. Dobrovie-Sorin, this volume, for a discussion of French nominals with a “partitive article” in the scope of negation). In Italian, we have observed that while bare nominals can only have narrow scope, the “partitive article” is ambiguous between narrow and wide scope. This not only holds of positive and negative declaratives (47a-b), but also of questions, as in (47c):

(47) a. Ho mangiato (dei) biscotti.

‘I ate (some) biscuits.’

b. Non ho mangiato (dei) biscotti.

‘I didn’t eat (any) biscuits. / I didn’t eat some biscuits.’

c. Hai mangiato (dei) biscotti?

‘Did you eat (any) biscuits? / Did you eat some biscuits?’

In a variety which only allows for wide scope of di+art, such as Anconetano, (47c) is expected to be felicitous only in the interpretation in which the existence of biscuits that I did not eat is presupposed. This is because Anconetano only allows for the wide scope interpretation of di+art. This prediction is borne out, at least in my native speaker capacity.

Mood and modality generally interact with the presupposition of existence of the referent of the indefinite complement. A predicate in a conditional clause does not state the existence of its internal argument. The Italian sentences in (48) are equally felicitous whether I have in mind some specific biscuits or friends (strong interpretation) or not (weak interpretation):

(48) a. Mangerei (dei) biscotti.

‘I would eat (some) biscuits.’

b. Arriverebbero (delle) amiche.

‘There would arrive (some) friends.’

This also holds of relative clauses in the subjunctive mood, which strongly favour the weak interpretation of the object of predicates such as “look for” or “wish”:

(49) a. Cerco (dei) biscotti che non facciano ingrassare.

‘I am looking for (some) biscuits that do not make you fat.’

b. Desidero (delle) amiche che mi vogliano bene.

‘I wish [to have] (some) friends who love me.’

In Anconetano, the contexts enforcing narrow scope, such as those in (49) are predicted to be ungrammatical. This is borne out in my native speaker judgement. Fieldwork is needed to confirm this judgement.

3.6 Specialization of Meaning

Cardinaletti and Giusti (2018) argue that the different distribution of zero, art, bare di, and di+art in given areas of the Italian territory can be due to different specializations of meaning associated to weak indefinites. The examples corresponding to the ais map 1037 ‘[if there was] water’ have the largest occurrence of bare nominals, because a mass noun in postverbal subject position of an existential predicate in a conditional sentence does not trigger any special meaning. Zero thus represents the core form to express uncontroversial indefinites. Art occurs more often in the ais map 1343 ‘[go to the cellar] to take wine’, because wine is the typical substance stored in a cellar and this suggests that art specializes for saliency. Di+art occurs more often in the ais map 637 ‘[to look for] violets’, because violets are usually picked in small quantities, a notion encoded by the “partitive article” in those Italian dialects that display it. This proposal was based on our personal intuitions and needs to be confirmed by the metalinguistic observations of a larger number of native speakers. This is one of the aims of the questionnaire presented in Section 4.

It is well-known that different indefinite determiners convey different types of indefiniteness. The semantic literature is abundant in distinguishing “identifiability” of an indefinite referent in epistemic Logic (Horn 2000; Jayez and Tovena 2002). It is therefore expected that coexisting determiners specialize for different interpretations. What is difficult to establish is what exactly these specializations are and how they correlate with the other features interacting with indefiniteness. One case in point in our repertory of indefinite determiners is certo “certain”. Certo is present in all Italian varieties. It combines with the indefinite article (una certa persona, ‘a certain person’) with count singular nouns, but behaves as a determiner, therefore competes with di+art and art, with plural count nouns ((*delle/*le) certe persone ‘(*some) certain people’) and mass nouns ((*della/*la) certa carne ‘certain meat’). Given that certo is reported to have core indefinite meaning in some southern Italian dialects, as in (20) in Section 2.2, the issue arises as to whether it can be a core indefinite in certain varieties.

3.7 Collocation (Frequency of Possible Predicate-Object Combinations)

The notion of saliency does not only concern our encyclopedic knowledge and shared assumptions; it also regards the frequency with which the predicate and direct object (or other arguments) combine in the same collocation. For example, “eat meat” or “drink water/wine” are supposedly more frequent and certainly more general than “eat potatoes” or “drink spirits”. The issue of the frequency of the verb-noun collocations is an important matter that requires a separate dedicated search in large corpora of Italian. Such corpora are unfortunately not available for Italian dialects and comparison across regional varieties of Italian and local Italo-Romace varieties would be impossible in this respect.

4.1 A Protocol for Indefinite Determiners in Italo-Romance

In Table 8.1 we find a simple protocol that can be applied to any Romance language to establish the distribution of the various indefinite determiners found in that language. The horizontal axis is filled with six determiners in non-dialectal informal Italian among those presented in Section 2.2 (but we could have more). The vertical axis presents the three main noun classes we adopt. Here we could be more detailed, as regards noun subclasses, as we have seen for spaghetti and spinaci in (41) above. For each property we attribute a value: [+] indicates that the form is present, [–] that it is absent, and [#] that it exists but has a different interpretation:¹²

The determiners in Table 8.1 interact with the other features discussed in Section 3. Let us start with the observations we made about the grammatical functions of the nominal phrase. These functions are represented in Table 8.2, where the value [0] indicates that the feature is not relevant. We take bare di to be irrelevant, because it is attributed a [-] for each noun class in Table 8.1:¹³

We can look for possible specializations of meaning as discussed in Sections 2.2, 3.2 and 3.6. The protocol in Table 8.3 displays the value [?] for unchecked items. This means that we do not have an answer to the following research questions expressed in the protocol: How does the grammaticalized determiner due behave with respect to the different specialized meanings of indefiniteness seen in Section 3? Is it restricted to core meaning? Is it used for small quantity (as is probably the case, if I can trust my personal judgement)? Is it compatible with specific interpretation? Does it display scope ambiguities, like di+art? Appropriate fieldwork is needed to answer these questions.

In Sections 3.4 and 3.5, we discussed sentence types and different properties of the predicate. The questionnaire illustrated in Sections 4.2 and 4.3, only includes present vs. past tense associated to generic vs. episodic sentences and telic vs. atelic aspect. For this reason, protocol in Table 8.4 nests Tense (present / past) and Aspect (telic vs atelic) as sub-features of sentence types (generic vs episodic). The values filling the protocols in Tables 8.1–8.4 refer to the discussions in the literature and in some cases to my own judgement:

Generativists are often criticized because they rely on few native speakers’ judgements and do not check these judgements on a larger scale. This would enable us to observe that judgements are not as categorical as presented in the generative literature. The protocol approach is precisely meant to transfer theoretical claims such as the ones reported in Section 3 into a tool for empirical research (the questionnaire) to detect variation and optionality in the distribution of indefinite determiners and provide a widely accessible presentation of the results.

4.2 A Questionnaire Based on the Protocol

This section presents the design of a questionnaire meant to check a subset of all possible combinations of the properties arising with indefiniteness as highlighted in the four protocols discussed in the previous section. The questionnaire was made of 25 items:

• 9 items presenting multiple possibilities for direct objects (singular mass vs. count plural) to express indefiniteness in different contexts (tense / aspect / polarity).

• 2 open substitution tasks, asking the participant to replace a singular mass noun encountered in the previous context with a plural count noun.

• 6 open comments on possible differences in interpretation, in case the participant selected more than one choice, to collect metalinguistic observations.

• 4 items asking the participant to judge the coherence of statements with a follow-up causative clause, to check the occurrence of determiners in narrow vs. wide scope.

• 3 open questions on the linguistic attitude of the participant (confidence in their judgments, their normative vs. descriptive attitude, their personal appreciation of the experience of completing the task).

• The last item requested the consensus to use the data in anonymized form.

The items were administered in a fixed order with no fillers.¹⁴ This made the questionnaire as short and accessible as possible, allowing the participants to “warm up” with the shorter and simpler sentences and to complete the more engaging tasks without getting tired towards the end of the session. All instructions and items where provided in Italian. In Table 8.5, I provide the English glosses and the variation among the five determiners in order to save space but mind that each item of the table was spelled out as 5 full sentences:¹⁵

The combinations of features to be investigated are spelled out in Table 8.6. In the first column, we find the item numbers of Table 8.5. In the last column, we find the type of task that was used to collect the data. Items [13] and [15] were split according to the different value of telicity to be checked. Items [18], [20] and [21] did not have the choice with certo. The [YES], [NO], [YES/NO] values should be read as follows: [YES]: the feature is positively enforced in the stimulus; [NO]: the feature is negatively enforced in the stimulus; [YES/NO]; the feature is not enforced in the stimulus. The value for frequency of the collocation are arbitrary estimations and need to be checked.

Cardinaletti and Giusti’s (2020) quantitative study of items [1–4, 5–7] shows that zero and art are the most common. zero is more often used than art in negative generic sentences in the present [1–4]. Art is more often used than zero in positive episodic sentences in the past [5,7]. Bare di never appears in non-dialectal informal Italian. Di+art and certo do not occur in generic sentences, thereby confirming their specialized meanings.

Table 8.7 reports the results of the whole questionnaire. Recall that it was possible to choose more than one possibility; all cells could in principle be filled with the value 100 %. Any deviation from this percentage means that some participants did not find the determiner appropriate:

In Table 8.7, only zero and art massively appear in all contexts. zero has the highest preference (≥80 %) in items [1, 2, 3, 4, 5, 11, 15a, 18]. Art has the highest preference (≥80 %) for items [2, 9, 13a, 17, 20, 21]. Art prevails over zero in the two telic contexts [13b, 15b], and the contexts with salient objects: [20] (“wine” salient to cellar); [21] (“beefsteaks” salient to the butcher’s); [9] (“violets” salient to mountains). Another saliency context is item [2]: according to the comments collected in item 3, “potatoes” was interpreted by many participants as salient to a hypernym “vegetables” (to be contrasted to “meat” found in the previous item). Finally, item [13a] presumably presents a highly frequent collocation of the predicate “cut” with the mass noun “grass”, which favours art, even if zero is robustly present.

Di+art is virtually absent (≤3 %) in generic negative sentences in the present [1–4]. It is more frequently chosen (between 45 % to 68 %) in items [5, 7, 9, 12, 20, 21], which favour small quantity interpretation: wine is drunk in small quantity [5, 7, 20], violets can be collected in small bunches [9], steaks for a dinner are certainly bought in the appropriate (small) quantity [21]. It is mildly chosen (27–33 %) in narrow scope contexts [11] and [18], in atelic contexts [13a, 15a], and with small objects, such as blackberries [15a–b] independently of telicity. It is possible but less popular (15 %–18 %) with mass nouns with a telic predicate [13b] or with presupposition of existence (wide scope) [17]. It is interesting to contrast item [12], which enforces presupposition of existence of plural count nouns and has di+art chosen by 49 % of the speakers and item [11], which disfavours presupposition of existence of plural count nouns, and is chosen only by 30 % of the speakers. This confirms the ambiguity of di+art between wide and narrow scope, and the preference of some speakers to only use it with wide scope. Note that the results of the questionnaire partially contradict my judgement of (44a), which ruled out the “partitive article” in narrow scope contexts. My judgement was in fact representative of my central Italian competence in contact with Anconetano (my native dialect).

The distribution of certo was only tested in items [1–17]. According to the comments collected in items [6, 8, 10, 14, 16], certo is interpreted as either specifying the intension (a specific type of N) or the extension, that is as denoting a specific referent. This is confirmed by the observation that certo is the most chosen determiner in item [12], which not only presupposes the existence of the referent but also provides a context to pick specific (types of) individuals (with the characteristic of unpleasantness), as stated in the continuation of the sentence (“… because they were unpleasant”). More research is needed to establish whether it is appropriate to distinguish the specification of the referent as opposed to the specification of the type.

4.3 A Pilot Adaptation of the Questionnaire to Italo-Romance Dialects

The protocol methodology is designed to capture cross-linguistic variation. It would be desirable to translate the questionnaire into individual dialect points or areas across the Italian territory. The values in Table 8.8 are the results of pilot studies conducted by students of the MA programme in Language Sciences at Ca’ Foscari University of Venice, attending my Comparative Syntax class in the academic years 2017–2018 and 2018–2019. The dialects investigated partially cover the Italian territory: Campomolino (TV), Furlan (2018); southern Friulian (Castions di Strada, Pocenia and Gonars, UD), Perinot (2018); Piacenza, Molinari (2018); Altamura (BA), Vicenti (2019); the Neapolitan area (Casalnuovo, Casoria, Soccavo, Bagnoli, Pozzuoli, Santa Lucia, San Ferdinando, Vasto Napoli, Somma Vesuviana, Frattamaggiore), Procentese (2019); Galati (RC), Maesano (2019); Lecce, Antonaci (2018). The Ancona dialect is represented by my own judgements and needs proper fieldwork but is inserted here to represent a central variety. Totally missing areas are the North-West, Sicily and Sardinia. Despite these areal gaps, the organization of the protocol allows us to confirm some of Cardinaletti and Giusti’s (2018) generalizations on Italian dialects, based on the ais maps, and Cardinaletti and Giusti’s (2020) generalizations on informal Italian, based on the first 6 items of the questionnaire. The straight +/- value indicates that the form is present or absent. The value in parenthesis indicates that it is present only in some contexts.

The “partitive article” di+art is only present in Gallo-Romance varieties, mostly in Emilia (Piacenza), where it can have core indefinite meaning (as indicated by a straight [+]). In the rest of the North (Friulian and Campomolino), di+art is possible, as in Italian, only in episodic contexts, as indicated by the parenthesized [(+)] value. In the South di+art is totally absent, as indicated by a straight [-] value.

Bare di is possible in the dialect of Piacenza in the scope of negation, as represented by [neg > +]. The dialect of Piacenza can thus be considered as the area in which bare di covaries with di+art to express core indefiniteness. Cerruti and Regis (2020) report the possibility of bare di in object position in modern Piedmontese (cf. Stark and Gerards, this volume, on Francoprovençal, which has a bare de in many contexts). In future research, an adaptation of our questionnaire to Piedmontese will allow us to ascertain whether it coexists with other forms, and if so, how the coexisting forms distribute in the different contexts.

Note that in no dialect, not even the southern ones, have we detected the use of certo with core indefinite interpretation, which is reported by Rohlfs (1968, 118) and Giammarco (1979, 141). The sparseness of our data may well be the reason of this gap. Research on other areas of the South is needed to check the survival of this form, which is in strong competition with itself in the specialized meaning of specific interpretation.

Zero is not only absent in the central dialect of Ancona, but, surprisingly, even in the southern dialect of Altamura (central Apulia). Note that in Piacenza and Neaples, it is possible but not favoured as indicated by the value in parentheses (+). Zero is the unmarked determiner at the extreme South (Lecce (southern Apulia) and Galati (Southern Calabria)) and in the North (Venetan dialect of Campomolino and Friulian). This confirms the hypothesis of the North-South isogloss of art in indefinite nominals surrounded by the area in which zero has core indefinite interpretation and art is limited to special readings (namely the ones involving saliency or in generic sentences).