Agreement and Information Structure in Spanish PRO_[PL] with-DP ^†

Abstract

This paper aims to revisit a construction found in some Spanish varieties which refers to a set constituted by a singular referent and an annex introduced by the item con ‘with’: PRO(noun)_[PL] with-DP. This construction triggers plural agreement and can be doubled by a plural pronoun, indicating that the annex is included in the set to which verbal agreement and the plural pronoun refer. For example, Nosotros con Juan viajamos ayer (literally, ‘We with Juan travelled.1PL yesterday’) means ‘Juan and I travelled yesterday’. We explore the Spanish PRO_[PL] with-DP, taking into account its discursive properties together with the syntactic requirements involved in the agreement patterns. In fact, although the two individuals denoted by this construction are involved as equal participants in the event, they have a different discursive status: one of them introduces new information, while the other refers to the immediate communicative situation. If some notions regarding information structure can be coded by binary features such as [+/−anaphor] and [+/−contrast], it is possible to find plurality triggered by the opposite combination of features within the same syntactic object. PRO_[PL] with-DP is a possibility that the lexicons of some languages offer.

1. Introduction

Plural-person information opens an interesting range of options regarding the interpretation of referents at a discursive level. First-person plural, for instance, clearly refers to the speaker, but some common ground is necessary to complete the rest of the reference involved in the pronoun (1PL = 1SG + someone else). Interestingly, languages show different options to explicitly mention the ‘rest of the reference’ when it constitutes discourse-relevant information. The alternative this paper focuses on is found in different languages and, ever since Schwartz (1988), it has been known as the Plural Pronoun Construction (PPC). The main characteristic of the PPC is that the plurality shown by verbal agreement includes a singular referent and the argument introduced by a (not always covert) preposition-like item: a comitative item. For instance, nous ‘we’ and avons (AUX.1PL) in (1a) have a dual number reading, which includes the speaker and the argument introduced by avec ‘with’ (mon frère ‘my brother’). All the examples in (1) have the same dual interpretation of plural morphology. The crosslinguistic differences lie in the possibility of dropping the plural pronoun (e.g., Catalan and Spanish vs. French) as well as in the presence of a covert comitative item (e.g., Icelandic)1.

(1)	a.	Nous	l’		avons fait	avec	mon frère
		we	it		did.1PL	with	my brother
		‘I did it with my brother’				[French] (Rigau 1990, p. 366)
	b.	Amb	la Maria	us	vau prometre		pel febrer
		with	the Maria	CL.1PL	engaged.1PL		on February
		‘I got engaged to Maria on February’				[Catalan] (Rigau 1989, p. 203)
	c.	Con	mi mujer		nos	casamos	en abril
		with	my wife		CL.1PL	marry.1PL	in April
		‘My wife and I got married in April’				[Spanish] (Mare 2015, p. 275)
	d.	Me	mentiin		Annan	kanssa	kaupunkiin
		we	go.1PL		Anna.GEN	with	town.ILL
		‘Anna and I went to the town’			[Finnish] (Holmberg and Kurki 2019, p. 244)
	e.	Við	Ólafur		fórum.
		we	Olaf.NOM		went.1PL
		‘Olaf and I went/left’			[Icelandic] (Sigurðsson and Wood 2020, p. 3)

Following Holmberg and Kurki’s (2019) remarks on the terminology used to describe the data, the constructions represented in (1) will be named PRO(noun)_[PL] with-DP from now on. As has been widely discussed, these data indicate that the comitative item appears to be the element that triggers plural morphology, as if it were a Boolean phrase (see Lakoff and Peters’ seminal 1969 work; Kayne 1994; Stassen 2000, among others). However, constructions in (1) differ not only from general coordination, but also from the Comitative Coordination (CC) found, for example, in Russian in Maša s Dašej verjat v boga, ‘Maša and Dašej believe in God’ (lit., Maša with Dašej believe in God, Feldman 2002, p. 43). First, in CC, the pronoun is singular when pronounced (DP_[-PL] with DP … verb_[+PL]) and, in many languages, it only involves two non-pronominal DPs (see Feldman 2002, pp. 42–43). The second difference is that neither of the DPs related by the comitative item in the CC can be moved out of the phrase (*DP … verb_[+PL] with DP/ *With DP … DP_[-PL] … verb_[+PL]). The relevance of pointing out these distinctions is that, in any case, the examples in (1) can be analyzed as instances of appositive constructions in which the with-DP phrase is an apposition of the plural pronoun (Mare 2012, for Spanish; Sigurðsson and Wood 2020, for Icelandic).

This paper aims to revisit the PRO(noun)_[PL] with-DP in Spanish varieties taking into account its discursive properties together with the syntactic requirements involved in the observed agreement patterns. Since most of the previous analyses focus exclusively on the most frequent pattern (the 1PL exclusive reading), our goal is to develop a proposal that also captures the patterns that do not follow a person hierarchy but different discursive properties. Accordingly, we propose that this agreement is the result of a complex DP (as argued by Mare 2015) that refers to individuals with different informational statuses. We argue that there is a tight relation between the difference regarding information status and plural pronouns. In brief, there would be two mechanisms that trigger plural number: one of them operates with features related to person information, for instance, the combination of [−Participant]/[+Author] (Halle 1997), while the other involves oppositions in terms of anaphoricity. In a late insertion model like the one adopted in these pages, the externalization of the element that introduces the DP interpreted as ‘the rest of the reference’, i.e., the comitative item, follows from the syntactic structure transferred and the lexical items compatible with this structure. This means that if a variety does not have a lexical item for this structure, the derivation is ruled out.

The paper is organized as follows. Section 2 provides a general description of PRO_[PL] with-DP in Spanish varieties and compares it with general comitative constructions (GCCs). Section 3 discusses some previous proposals regarding the PRO_[PL] with-DP structure and presents an analysis for Spanish PRO_[PL] with-DP. Section 4 focuses on information structure and the way in which this strategy works depending on the accessibility of the referents. Finally, we systematize the main conclusions drawn from this study.

2. On Spanish Comitative Constructions

Comitative constructions share some properties with plural DPs (DP_[PL]) and coordination (DP_[&]) (see Conti Jiménez 2005, p. 297). This means that, if one of these entities refers to a human individual, for instance, then the other members involved will therefore be interpreted as having human properties, such as volition. The second similarity is closely related to the first: it is concerned with the fact that all the individuals denoted by these constructions relate to the event in a similar way in terms of their theta roles. Accordingly, there are some clear contexts in which these constructions are in complementary distribution, as shown below by the symmetric predicate convivir ‘live together’.

(2)	a.	Mis	amigos		conviven.		[DPPL]
		My	friends		live.together.3PL
		‘My friends live together.’
	b.	Juan	y	Pedro	conviven.		[DP coordination]
		Juan	and	Pedro	live.together.3PL
		‘Juan and Peter live together.’
	c.	Con	Pedro		convivimos.		[PRO[PL] with-DP]
		With	Pedro		live.together.1PL
		‘Peter and I live together.’
	d.	Juan	convive		con	Pedro.	[GCC]
		Juan	lives.together		with	Pedro
		‘Juan lives together with Pedro.’

In this section, we will discuss some relevant data that will hopefully provide us with a thorough description of the PRO_[PL] with-DP construction in Spanish varieties. We will apply a number of syntactic and semantic diagnostics to distinguish between PRO_[PL] with-DP (2c) and general comitative constructions (GCCs) like (2d). In Spanish, as well as in many other languages, both constructions present the same item: con ‘with’. As will be shown, this comparison feeds Mare’s (2012) hypothesis that PRO_[PL] with-DP behaves as DP_[PL], while GCCs present restrictions regarding the projections involved.

Most of the examples in this paper belong to varieties from Argentina and Chile. This is important to mention, because although the PRO_[PL] with-DP construction is widespread, varieties differ in terms of the predicates with which it is combined and the person features involved. Furthermore, there are important differences in productivity, which, as we will show in Section 4, are related to communicative factors, rather than to grammatical properties.

2.1. Comparing PRO_[PL] with-DP with GCCs

As has been pointed out in the literature (Mare 2012, 2013; Mare and Pato 2017), although PRO_[PL] with-DP and GCCs can refer to the same number of participants involved in the event, there are some interesting differences between them. When it comes to verbal morphology, a GCC does not affect verbal agreement, regardless of its obligatory nature in terms of argument structure (3a). By contrast, in PRO_PL] with-DP, the verb presents plural number information, despite the fact that the reference of the argument not introduced by con ‘with’ is a single entity ([-PL]). As can be observed, both the examples in (3) and (4) can be translated in the same way in the varieties under discussion, that is, the speaker with someone else (dual interpretation).

(3)	a.	Estoy	conversando	con	Angélica.	[GCC]
		be.1SG	speaking	with	Angélica
		‘I’m speaking with Angélica.’
	b.	Estoy	escribiendo un trabajo	con	Angélica.	[GCC]
		be.1SG	writing a paper	with	Angélica
		‘I’m writing a paper with Angélica.’

 

(4)	a.	Con	Angélica	estamos	conversando.	[PRO[PL] with-DP]
		with	Angélica	be.1PL	speaking
		‘I’m speaking with Angélica.’
	b.	Con	Angélica	estamos	escribiendo un trabajo.	[PRO[PL] with-DP]
		with	Angélica	be.1PL	writing a paper
		‘I’m writing a paper with Angélica.’

Beyond verbal agreement, the examples above display contrasts in constituent order. For example, Con Angélica estoy conversando ‘I’m speaking with Angélica’ and Con Angélica estoy escribiendo un trabajo ‘I’m writing a paper with Angélica’ are discourse-marked in comparison to the order in (3). In fact, one of the differences between the two comitatives regards information structure: while in the GCC, the unmarked constituent order is Verb-(Direct Object)- with-DP, in the PRO_[PL] with-DP, the unmarked order presents the with-DP phrase on the left of the sentence. As will be shown in the following paragraphs, there are other diagnostics that allow us to go deeper in our understanding of the contrasts related to discursive aspects.

Mare (2012) presents specific diagnostics to distinguish PRO_[PL] with-DP from GCCs. The first one is the combination of these two comitative constructions with other phrases headed by the item con ‘with’. Pascual Pou (1999) observes that comitative and instrument phrases, which are both headed by con in Spanish, occupy the same position in the structure and, consequently, cannot appear together in the same sentence. We hypothesize that this position is related to the head Voice, i.e., the head that introduces agents in the syntactic structure (see Section 3). Following Pascual Pou’s observations, Mare (2012) analyzes the combination of both comitative constructions not only with con-phrases introducing an instrument but also with con-phrases introducing company. As can be observed in the examples below, the differences are straightforward when maintaining the unmarked order for GCCs and PRO_[PL] with-DP2.

(5)	Comitative + Instrument
	a.	??Trabajé	con Andrea	con la computadora.	[GCC]
		work.1SG	with Andrea	with the computer
		Lit., ‘I’m working with Angélica with the computer.’
	b.	Con Andrea	trabajamos	con la computadora.	[PRO[PL] with-DP]
		with Andrea	work.1PL	with the computer
		‘Andrea and I are working with the computer.’

 

(6)	Comitative + Company
	a.	??Bailé toda la noche	con Andrea	con Severino.	[GCC]
		danced.1SG all the night	with Andrea	with Severino
		Lit., ‘I danced the whole night with Andrea with Severino.’
	b.	Con Andrea	bailamos toda la noche	con Severino.	[PRO[PL] with-DP]
		with Andrea	danced.1PL all the night	with Severino
		‘Andrea and I have danced with Severino the whole night.’

Changes in the order of the constituents can affect information structure in the PRO_[PL] with-DP, whereas in the case of GCCs, these changes do not improve the results. Briefly, this contrast seems to feed the hypothesis that PRO_[PL] with-DP and GCCs are associated with different parts of the syntactic structure.

A second distinction regards contrastive focus, i.e., a structure that opens the evocation of alternatives and resolves it through the pronounced option. When this evocation is not resolved, a wh-item is found. Interestingly, when the with-DP constituent is focalized, the dual interpretation of PRO_[PL] with-DP is lost and the recovered referent is not a single entity but a plural one: we and Severino in (7a); you_[PL] and someone else in (7b), capital letters representing the focalized item in (7a).

(7)	a.	#Con SEVERINO	bailamos	toda la noche.
		with SEVERINO	danced.1PL	all the night
		‘We danced the whole night with Severino.’
	b.	#¿Con quién	bailaron	toda la noche?
		with whom	danced.2PL	all the night
		‘Who did you dance the whole night with?’

Something similar happens with cleft sentences. When the VP is focalized, the dual interpretation of PRO_[PL] with-DP is also lost.

(8)	#Fue	bailar	con Severino	lo	que	hicimos	toda la noche.
	was	dance.INF	with Severino	the	what	did.1PL	all the night
	‘It was dancing with Severino what we did the whole night.’

This diagnostic is relevant, because it feeds the hypothesis that in the constructions under study the with-DP phrase is part of a complex subject DP. Consequently, it cannot be focalized with the VP, as expected from any subject DP (9).

(9)	*Fue	bailar	Juan	lo	que	hizo	toda la noche.
	was	dance.INF	Juan	the	what	did.3SG	all the night

Another diagnostic that sheds some light on the behavior of PRO_[PL] with-DP is found in control constructions. As is well known, these structures involve non-finite contexts in which the potential subject is unpronounced and controlled by one of the arguments in the main clause: I_i want to Ø_i see you tomorrow. This null subject has been represented by PRO since Chomsky (1981), and its properties and distribution have led to the assumption that it is ungoverned according to the binding conditions of being both [+pronominal] and [+anaphoric]. The goal of this diagnostic is to define whether it is possible to obtain the dual interpretation of the 1PL in this context. In order to force the dual interpretation, we use a 1PL (10a) and a 1SG (10b) clitic as the controller and a pronominal construction for the subordinate verb. The result is that the dual reading is lost in both cases3.

(10)	a.	#Juan	nos	recomendó	comunicar	-nos	con Severino.
		Juan	CL.1PL	recommended	communicate.INF	CL.1PL	with Severino
		‘Juan recommended me to communicate with Peter.’
	b.	#Juan	me	recomendó	comunicar	-nos	con Severino.
		Juan	CL.1SG	recommended	communicate.INF	CL.1PL	with Severino
		‘Juan recommended me to communicate with Peter.’

In sum, the dual number interpretation that characterizes PRO_[PL] with-DP in Spanish seems to be closely intertwined with both verb–subject agreement and a specific discursive organization.

2.2. Holistic/Distributive Interpretation and Syntactic Functions

To continue with the comparison, let us revise some semantic characteristics of the structures under scrutiny. A relevant distinction refers to the combination of comitatives with different kinds of predicates. As Rigau (1989) explicitly remarks, across languages, comitative constructions contribute to a holistic interpretation of the predicate. This means that the argument introduced by the comitative does not trigger a plural event interpretation or, in other words, a distributive interpretation. For instance, in (11), the event of working in Neuquén is only interpreted as a singular event in (11a) as well as in (11b). Conversely, in (11c), it has a distributive reading (Tom works in Neuquén and Jemmy works in Neuquén).

(11)	a.	Tom works in Neuquén.
	b.	Tom works in Neuquén with Jemmy.
	c.	Tom and Jemmy work in Neuquén.

As Rigau observes, comitative constructions are incompatible with inherently distributive predicates, such as know, understand, be a fan of, etc. Interestingly, GCCs (12) and PRO_[PL] with-DP (13) present differences in their behavior in relation to this kind of predicates, at least in most of the Spanish varieties in which PRO_[PL] with-DP is attested4.

(12)	a.	*Soy	fanática de River con mi hermano.
		be.1SG	a River fan with my brother
		Lit., ‘I’m a River fan with my brother.’
	b.	*Se	tocar la guitarra con mi hermano.
		know.1SG	play the guitar with my brother
		Lit., ‘I know how to play the guitar with my brother.’

 

(13)	a.	Con mi hermano	somos	fanáticos de River.
		with my brother	be.1PL	a River fan
		‘My brother and I are River fans.’
	b.	Con mi hermano	sabemos	tocar la guitarra.
		with my brother	know.1SG	play the guitar
		‘My brother and I know how to play the guitar.’

This contrast is crucial to understand that each type of comitative phrase is related to the other referent in a structurally different way. A GCC merges in a position in which it does not affect the semantic interpretation of the event. In contrast, PRO_[PL] with-DP seems to be part of an argument that must be interpreted as plural and, consequently, it also triggers the distributive reading. Of course, as Conti Jiménez (2005) observes, there seems to be a natural holistic interpretation with not-necessarily distributive predicates such as viajar ‘travel’ (14a). However, this interpretation appears to be part of an inference that also applies to coordination (14b) and plural DPs (14c), and not the result of syntactic compositionality.

(14)	a.	Con Juan	viajamos	en el verano.	Él fue al mar y yo a la cordillera.
		with Juan	traveled.1PL	in the summer	He went to the sea and I went to the mountains
		‘Juan and I traveled during the summer. He went to the sea and I went to the mountains.’
	b.	Juan y yo	viajamos	en el verano.	Él fue al mar y yo a la cordillera.
		Juan and I	traveled.1PL	in the summer	He went to the sea and I went to the mountains
		‘Juan and I traveled during the summer. He went to the sea and I went to the mountains.’
	c.	Los docentes	viajamos	en el verano.	Él fue al mar y yo a la cordillera.
		the teachers	traveled.1PL	in the summer	He went to the sea and I went to the mountains
		‘We teachers traveled during the summer. He went to the sea and I went to the mountains.’

Another source of empirical data that supports this distinction is provided by quirky subjects, which—as is well known—are arguments of distributive reading predicates in Spanish. Again, in contrast with GCCs, PRO_[PL] with-DP can freely occur in these constructions, at least in Argentinean and Chilean Spanish varieties5.

(15)	a.	??Con Ana	me	interesan	esos libros.	[GCC]
		with Ana	CL.1SG	interest.3PL	these books
		Lit., ‘I’m interested in these books with Ana.’
	b.	Con Ana	nos	interesan	esos libros.	[PRO[PL] with-DP]
		with Ana	CL.1PL	interest.3PL	these books
		‘Ana and I are interested in these books.’

This behavior is remarkable because quirky subjects present the dative case in Spanish and PRO_[PL] with-DP does not fit well with syntactic functions related to accusative and dative case. This is an important difference between Spanish varieties and Slavic languages which allow PRO_[PL] with-DP in all syntactic functions (Ionin and Matushansky 2002). Additionally, it is the main difference between PRO_[PL] with-DP and DP_[PL] or DP_[&]. Grammaticality judgments on these examples in Spanish are not homogenous: in some cases, neither PRO_[PL] with-DP (16) nor GCCs (17) are accepted, while in some others, the problem is that the plural marker is not interpreted as dual (18).

(16)	a.	??Con Pedro	nos	saludó	Juan.	[*nos = Pedro and me]
		with Pedro	CL.1PL	greeted.3SG	Juan
		‘Juan greeted Pedro and me.’
	b.	Con Pedro	nos	regaló un libro	Juan.	[*nos = Pedro and me]
		with Pedro	CL.1PL	gave.3SG a book	Juan
		‘Juan gave Pedro and me a book.’

 

(17)	a.	Juan	me	saludó	con Pedro.	[*Pedro and me]
		Juan	CL.1SG	greeted.3SG	with Pedro
		‘Juan greeted Pedro and me.’
	b.	Juan	me	regaló un libro	con Pedro.	[*to Pedro and me]
		with Pedro	CL.1SG	gave.3SG a book	Juan
		‘Juan gave Pedro and me a book.’

 

(18)	a.	#Los militares	nos	interrogaron	con mi hermano	durante horas.
		the military	CL.1PL	interrogated.3SG	with mi brother	for hours
		‘The military interrogated my brother and I for hours.’
	b.	#Los militares	nos	arrojaron	con mi hermano	a una celda.
		the military	CL.1PL	threw.3SG	with mi brother	into a cell
		‘The military threw my brother and me into a cell.’
	c.	#Nos	insultaron	con mi hermano	en las redes sociales.
		CL.1PL	insulted.3PL	with my brother	in the social media
		‘People insulted my brother and me in the social media.’

In spite of this restriction, when the passive construction is possible, PRO_[PL] with-DP is allowed as subject (19), which suggests that the presence of PRO_[PL] with-DP is strongly conditioned by verbal agreement, regardless of its thematic role interpretation.

(19)	a.	Con mi hermano	fuimos	interrogados	durante horas.
		with my brother	were.1PL	interrogated.PL	for hours
		‘My brother and I were interrogated for hours.’
	b.	Con mi hermano	fuimos	arrojados	a una celda.
		with my brother	were.1PL	thrown.PL	into a cell
		‘My brother and I were thrown into a cell.’
	c.	Con mi hermano	fuimos	insultados	en las redes sociales.
		with my brother	were.1PL	insulted	in the social media
		‘My brother and I were insulted in the social media.’

2.3. Summary

All in all, there are clear differences between PRO_[PL] with-DP and GCCs, as summarized in

Table 1. Diagnostics for PRO_[PL] with-DP syntactic and semantic behavior.

Diagnostics	PRO[PL] with DP	GComP	DP[PL]	DP[&]
Verbal agreement	Yes	No	Yes	Yes
Combination with other with-DP phases	Yes	No	Yes	Yes
Holistic interpretation	Yes	Yes	Yes	Yes
Distributed interpretation	Yes	No	Yes	Yes
Quirky subjects	Yes	No	Yes	Yes
Accusative object	??	??	Yes	Yes
Dative object	??	??	Yes	Yes

, which also accounts for the behavior of Spanish DP_[PL] and DP_[&] in the contexts revised.

As observed, the main contrast between PRO_[PL] with-DP and other plural DPs regards their syntactic distribution. On the other hand, following previous studies in the literature, we mentioned some other contexts in which the dual interpretation of plural verbal agreement is lost (focalization of the with-DP and focalization of the VP). The next section revises the PRO_[PL] with-DP structure.

3. The Internal Structure of PRO_[PL] with-DP

Most of the diagnostics developed in Section 2 feed the hypothesis that PRO_[PL] with-DP behaves as a plural DP and that the with-DP phrase is part of it (Feldman 2002; Vassilieva and Larson 2005; Vassilieva 2005). This fact would explain why the holistic and the distributive interpretation are both possible with PRO_[PL] with-DP, although there is a comitative item that, in the general case, rejects the distributive reading. Moreover, if with-DP is part of a plural DP, agreement patterns follow without any further assumption: the verb agrees with the plural DP, not with a ‘grammaticalized preposition’ or a Boolean phrase that change its status throughout the derivation (Lakoff and Peters 1969; Kayne 1994).

In line with Ionin and Matushansky’s (2002) analysis, we argue that the comitative item inside the Spanish PRO_[PL] with-DP is a preposition, as long as the comitative item in the GCC is also analyzed as a preposition. This means that, beyond the proposed label, they behave in the same way, but each one appears in a different syntactic context. Stolz et al. (2006) note that the label comitative refers to the relationship between two entities belonging to the same entity class and participating in the event simultaneously. In harmony with these authors, Maslova (2007, p. 337) proposes the label ‘participant set’ to refer to the entities related by the comitative marker.

Inspired by these ideas, Mare (2020) argues that the comitative item introduces a DP but fails to assign a thematic role to it. The consequence of this failure is that the DP inherits the thematic interpretation from the projection to which the comitative item gets adjoined. In other words, in English, for instance, the interpretations for the item with and the DP introduced by it are obtained compositionally. If the DP has person features and the syntactic structure represents a transitive creation event (like work), it will be interpreted as part of the set referring to agents (John works with Peter). By contrast, if the DP refers to an inanimate entity and is introduced by with in the same syntactic structure, the interpretation obtained is that of an instrument, which is in some way linked to the agent (John works with a hammer).

This approach is also relevant to understand why GCCs cannot combine easily with other with-DP phrases. As mentioned above, both ‘company’ and ‘instrument’ are interpretations related to the agent and it is the head Voice that introduces the external arguments that are interpreted as agents and causers (Kratzer 1996). This means that with-DP_[COMPANY] and with-DP_[INSTRUMENT] occupy the same position in the structure, i.e., a projection adjoined to Voice. Conversely, PRO_[PL] with-DP is compatible with other with-phrases because the DP introduced by the comitative is inside a DP and does not occupy the same position as other with-phrases.

In the following sections, we focus on the syntactic structure of PRO_[PL] with-DP.

3.1. First-Person Plural as a Cue

Among plural pronouns, 1PL has the particularity of referring to a heterogeneous set of entities according to their participation in the communicative act. We, for instance, refers to the speaker and to another entity (or other entities) associated with it, which may be a part of the speech act (inclusive reading) or not (exclusive reading). As is well known, the general use of 1PL refers to the speaker and the group/entity associated with it. This description is very close to the notion of associative plurality developed by Daniel and Moravcsik (2013), who distinguish between associative and additive elements. In their words:

An example of the additive plural is English boys. It is additive in the sense that it refers to a set where every member is a boy and thus the set is referentially homogeneous: every referent of the plural form is also a referent of the stem. In contrast, the associative plural designates a heterogeneous set.

(Daniel and Moravcsik 2013, p. 1)

1PL is a clear example of an associative plural and, in fact, when Halle (1997, p. 129) decomposes person information in features such as Participant in Speech Event ([PART]) and Author in Speech Event ([AUTH]), he remarks that two apparently incompatible values can be combined in the syntax and that this combination triggers plural number. By way of illustration, [+PART] and [−PART] can coexist inside a DP insofar as this DP refers to a group formed by a hearer and someone else. Of course, the result of that is DP_[+PL]. The difference between a ‘triggered’ plural number and a ‘free’ plural number is hence tightly related to Daniel and Moravcsik’s distinction.

In the case of 1PL, following Halle’s notion of a fourth person, we consider that plural number is triggered by the combination of [+AUTH] with either [+PART] or [−PART]. In both cases, the result is a referentially heterogeneous set, i.e., an associative plural.

In her approach to pronouns, Mare (2023) proposes a system in which Hum(an), Part and Auth project in the syntactic structure. HumP can merge with both Auth and Part. AuthP can be absent, while PartP can present a positive [+] or a negative [−] value. As was mentioned above, the feature [+PL] on # can be triggered by particular combinations of features. With that in mind, the structures for associative plural pronouns are formed in the syntax as schematized in (20), (21) and (22) (Mare 2023, p. 7)6.

(20) 1PL inclusive	(21) 1PL exclusive	(22) 2PL exclusive

The previous structures shed light on the PRO_[PL] with-DP construction. In some way, it seems to be a strategy that some varieties present in order to make explicit the part of the reference that is not linked to the communicative situation. In fact, in most of the examples, with-DP introduces a [−PART] referent which otherwise is lost. In Section 4, we show how this syntactic property relates to information structure.

3.2. The Syntactic Analysis of PRO_[PL] with-DP

As mentioned in the introduction, the PRO_[PL] with-DP construction is found in several languages. Feldman (2002) analyzes this construction in Russian and observes that speakers regard this construction as more natural when compared to ordinary coordination in which pronouns are involved. She also notes that the order of the elements related by the comitative item is restricted with respect to person hierarchy: the plural pronoun must be higher in the hierarchy (1 > 2 > 3) than the DP introduced by the comitative item. Feldman analyzes Russian instances of PRO_[PL] with-DP as transitive plural pronouns that select the with-DP as the complement. To account for the fact that the denotation of the complement is included in the plural pronoun, Feldman assumes a pragmatic restriction according to which one element of a set comprises another element in its denotation (Feldman 2002, p. 60).

As has been mentioned, there are many proposals for this kind of construction in different languages, but Feldman’s hypothesis highlights the close relation between PRO_[PL] with-DP and the characteristics of plural pronouns outlined in Section 3.1. Kratzer (2009) proposes that associative elements present the feature [GROUP], which is independent of [+PL]. In particular, she notes that

[I]n addition to speakers and addressees, contexts c may determine a function that assigns to selected individuals a not necessarily proper plurality of individuals that consist of the individuals themselves and their associates with respect to c.

(Kratzer 2009, p. 224)

Following Kratzer’s proposal, Mare (2012, 2013) introduced the feature [GROUP] in her analysis of comitative constructions in general as a property of the prepositional element that categorizes the root CON ‘with’, in line with the Categorization Assumption (Embick and Marantz 2008). This approach is interesting because it captures two properties of the constructions under study, i.e., their associative nature and the apparent lack of semantic content of the comitative item mentioned above. However, in a fine-grained analysis of pronouns, such as the one presented in Section 3.1, neither the feature [GROUP] nor a preposition categorizer are necessary to explain the PRO_[PL] with-DP construction.

As discussed earlier, the interpretation of a referent with an associated group is obtained by combining features with different values, as in (20), (21) and (22). This kind of hypermarking triggers plural number. Furthermore, the idea that the comitative item introduces a DP but that the interpretation of this DP depends on its own referential properties and on the projection in which the with-DP merges can be codified without further assumptions. If arguments are introduced by relational projection and con ‘with’ does not add any semantics, the result of merging this projection—named p-CON, from now on—in the scope of PartP will give rise to the proper interpretation. The resulting structure is (23).

(23) PRO[PL] with-DP, 1PL exclusive

The structure in (23) presents the projections corresponding to the 1PL exclusive, which is the more frequent construction. The DP introduced by p-CON is interpreted according to the position p-CON merges with (i.e., PartP). Consequently, this DP makes the reference of [−PART] explicit. One advantage of this analysis is that it can account for the PRO_[PL] with-DP construction in different languages. For instance, in a non-pro-drop language like French, ellipsis is impossible in this case and consequently the pronoun nous ‘we’ is going to externalize or lexicalize the projections of the main structure, while in Spanish varieties, they remain covert. On the other hand, in a language like Icelandic, in which PRO_[PL] with-DP does not present something like a preposition to introduce the annex (við Ólafur, literally, ‘we Olaf’ for ‘Olaf and I’), it could be the case that the relational projection p does not have a root as a complement. Of course, each language has its own characteristics regarding null elements and lexicalization; however, the basic structure of (23) could be extended to analyze data from other languages, beyond Spanish varieties7.

Last but not least, according to our analysis, the three pronouns represented in (20–22) can be involved in the PRO_[PL] with-DP construction. Nevertheless, there is a difference in productivity between the 1PL exclusive and the 2PL exclusive due to discursive factors. Accordingly, the less frequent PRO_[PL] with-DP construction is the one that includes [+Auth]/[−Part] (inclusive reading). In sum, the syntax of the varieties under analysis allows all the possibilities, and the difference in productivity is attributable to pragmatic factors. In the next section, we complete this panorama by making some remarks on information structure.

3.3. Summary

In this section, we reviewed some of the proposals that, as far as we understand, shed light on the data under study and offer some relevant answers to the behaviors exposed in Section 2. Accordingly, we focused on the analyses that understand the PRO_[PL] with-DP as a plural DP (Feldman 2002; Vassilieva 2005). Moreover, the observations regarding the characteristics of the group denoted by this construction (Maslova 2007; Daniel and Moravcsik 2013) were relevant to a widespread pronominal element: the first-person plural pronoun. This element denotes a heterogeneous group with at least two different referents (Halle 1997). They differ in terms of their participation in the speech act. With these ideas in mind, we design a proposal that takes into account pronoun properties and the distinction between homogeneous plurality and heterogeneous plurality. As a result, we consider syntactic structures for associative plural pronouns (20 to 22) that can be found in any language, but in some of them, these structures allow for the introduction of a DP. The introducer can be an overt p, as in Spanish, or a covert p, as in Icelandic. The advantage of this proposal is that it allows the derivation of all the possible combinations of features that give rise to associative plurals without adding any particular theoretical assumptions. Of course, when discussing variation, it will be crucial to observe in which cases a DP can be introduced in each dialect, under the scope of an associative plural pronoun. In the next section, we revise the notion of heterogeneous plurals codified in syntax, in addition to considering information structure in order to account for other combinations of features.

4. Information Structure and PRO_[PL] with-DP

It is impossible to talk about PRO_[PL] with-DP without revisiting the referential properties of plural pronouns. These elements have curious properties in terms of information structure also. As has been mentioned, 1PL refers to a group which is heterogeneous in terms of participation in the speech act. In the inclusive reading, 1PL refers to the author of the speech act (Auth) and the participant of the speech act (Part), while the exclusive reading involves the speaker and a non-participant. Although this last reading is not the only one available when talking about PRO_[PL] with-DP, it is clearly the most frequent one. Something similar is found in relation to 2PL: it can refer to a homogeneous group of participants or to a group constituted by a participant and a non-participant. The latter option is exemplified in (24), where A is taking about holidays to B, who knows that Juan is A’s partner.

(24)	A:	- Estoy de vacaciones.
		‘I’m on holiday.’
	B:	- ¿Se van	a algún lado con Juan?
		CL go.2PL	to some place with Juan?
		‘Are you travelling somewhere with Juan?’

The corresponding structure for the PRO_[PL] with-DP sequence in B’s interaction is represented in our analysis by adding the p-CON to the pronominal construction in (22), as in (25).

(25) PRO[PL] with-DP, 2PL exclusive

Finally, it is essential to pay attention to 3PL, even though PRO_[PL] with-DP is not found as frequently in the 3PL as in the 1PL and the 2PL. As is well known, 3PL is anaphoric by nature, so the heterogeneity involved in PRO_[PL] with-DP cannot be understood in terms of participation in the speech act but in terms of topic orientation (Leonetti 2022). Let us observe the example in (26).

(26)	- ¿Sabés algo de Paco?
	‘Any news from Paco?’
	- Están	entrenando con Ali para la Regata.
	be.3PL	training with Ali for the canoe race
	‘He is training for the canoe race with Ali.’

Neither of the referents for the 3PL found in the answer—i.e., Paco and Ali—are part of the speech act. Nevertheless, they present differences regarding information structure: the reference to Paco in the answer is retrieved from the question, while Ali is part of the new information.

The relevance of these data is that they change the way in which the relation between referents inside the PRO_[PL] with-DP is analyzed. Following typological studies (Stassen 2000; Stolz et al. 2006, 2013, among many others), it is attractive to look for explanations in terms of person hierarchy, and the most frequent data feed this kind of approach: 1PL agreement is found when Auth is the unpronounced referent and the p-CON introduces the “annex”. This approach correlates heterogeneity in the kind of participation in a speech act with person hierarchy: the syntactic structure of PRO_[PL] with-DP implies a structural hierarchy in which the most marked participant is higher and the less marked participant is introduced by the comitative. Nevertheless, the data regarding 3PL lead us to suspect a strictly syntactic explanation, based on features, and invite us to look for more data, independently of its frequency.

Mare and Pato (2017, p. 85) mention a very interesting set of data in which the comitative introduces the speaker as new information because the unpronounced referent is recovered by an antecedent. In (27), the antecedent is Antorcha Campesina, a 3SG referent, and in (28), the antecedent is the group formed by Quilco and Agiali, both PART (2PL). Examples (27) and (28) are adapted from Mare and Pato.

(27)	Quiero una relación cercana y de entendimiento con Antorcha Campesina;
	‘I want a close relationship and understanding with Antorcha Campesina’
	con	-migo	nos	vamos	a entender bien.
	with	-1SG	CL.1PL	AUX.1PL	to undertand well
	‘we will understand each other well’.
	antorchacampesina.org.mx. 26/09/2015, Mexico

(28)	–Oye –dijo Quilco a Agiali– vamos a coger cañas.
	‘-Hey- said Quilco to Agiali- let’s go get some reeds.
	–¿Y si nos ven? –objetó Agiali (…).
	-What if they see us? -said Agiali (…).
	–No hay nadie por ese lado.	Vamos	con	-migo
	There’s nobody on that side.	go.1PL	with	1SG
		- Let’s go with me.
	Raza de bronce, Alcides Arguedas, 1919, Bolivia

In brief, following Leonetti’s (2022) and Leonetti and Escandell Vidal’s (2020) way of reasoning, we claim for a division of labor between syntax and pragmatics. Syntax must explain the abstract structure of the PRO_[PL] with-DP, the mechanisms of agreement and the reasons why plural number is obtained, among other things. On the other hand, pragmatics should deal with referent identification, i.e., which is the most salient accessible antecedent in the discourse. Interestingly, this labor exceeds 3PL: 1PL and 2PL are also ambiguous between the dual reading (PRO_[PL] with-DP) and the plural reading (GCC), but it is the context that is going to define the optimal antecedent for the referent accompanied by the with-DP phrase.

With this information in mind, let us remark on the extent to which syntax and pragmatics dialogue. As discussed in Section 3, some features’ combinations trigger plural number. We focus on person features, but, as is well known, first and second person are dependent on the discursive context, so, accordingly, features like [Auth] and [Part] are necessarily related to pragmatic issues. Moreover, these features guarantee the heterogeneity needed for plural number: [Auth] needs to combine with other person features to obtain plurality and [Part] also opens this option.

Nevertheless, there are two facts that draw our attention: the first is the behavior of 3PL just mentioned and the examples in which the p-CON introduces a 1SG pronoun (27 and 28); the second refers to the diagnostics developed in Section 2, specifically, the data where some kind of focalization takes place and the dual reading is lost. Section 4.1 focuses on the difference between the two referents retrieved by PRO_[PL] with-DP in terms of accessibility, while Section 4.2 discusses the cases in which the dual reading is lost.

4.1. Codifying Salience

The analysis developed in Section 3 moves the debate to the distinction between overt subject pronouns and null subjects. As proposed, PRO_[PL] with-DP in Spanish is a peculiar kind of element which presents a null part and an overt part at the same time. The null part is recovered from the discourse situation, while the comitative introduces a referent that is otherwise ambiguous or hardly accessible. Moreover, PRO_[PL] with-DP is compatible with overt plural pronouns, a fact that leads us to pay attention to the distribution of null subjects and the structure under study.

Some authors point out that null subjects are topic oriented (see Samek-Lodovici 1996, p. 46; Camacho 2013, p. 146) and that overt preverbal subjects must be licensed by topic antecedents. On the other hand, Leonetti (2022) notes that the relevant notion is not topicality but salience. In fact, the author proposes that, according to the data, a null subject may be linked to a non-topical antecedent if it is salient enough. According to the Accessibility Theory (Ariel 1990, 2001), human languages provide speakers with the means or mechanisms to encode the degree of accessibility of referents, i.e., complex systems of markers that indicate whether or not the mental representation of the referent being alluded to is immediately accessible to the listener. The less accessible a referent is, the more elaborate the strategy used to make it explicit is. Ariel (1990) puts forward the proposal that referential expressions have the function of accessibility markers that signal the addressee which is the mental representation of the referent. As mentioned, in 1PL pronouns, for instance, the accessible referent is the speaker, but the rest of the reference is not accessible enough unless it can be recovered from the previous context. Moreover, it seems that the inclusive interpretation is the most accessible by default, as can be observed in the 1PL imperative construction: ¡Hagámoslo! ‘Let’s do it’.

In terms of accessibility, PRO_[PL] with-DP seems to be an intermediate option between a null plural pronoun and the coordination of DPs. The former is not informative enough, while the latter places the two coordinated DPs at the same informative level. The PRO_[PL] with-DP strategy allows the recovery of a hierarchy of accessibility between the referents involved. In fact, inside this structure, the two poles of the scale proposed by Ariel (1990, p. 73; 2001, p. 21) can coexist: a null element and a full name.

The high frequency of PRO_[PL] with-DP with 1PL agreement is expected for two reasons. First, this construction is related to informal interchanges, something already observed by Kany ([1945] 1970, p. 314), Camacho (1999, pp. 2669–70) and RAE-ASALE (2009: §33.7i). Second, the speaker is clearly an accessible referent. Beyond this, PRO_[PL] with-DP can present different features and recover different referents: processing works insofar as the unpronounced referent is salient enough for the interlocutor to be able to retrieve it.

Inspired by López (2009), we propose that some of the properties mentioned above can be codified in syntax by using two features: [+/−a(naphor)] and [+/−c(ontrast)]. However, we follow Leonetti (2022) in considering that not all sentence topics, i.e., elements marked as [+a], are equated with left dislocated phrases. Accordingly, null subjects in general are going to be marked as [+a]—in fact, they are definite—and not anaphoric elements, as [−a]. On the other hand, new information is going to be related to the feature [+/−c]. As López (2009, p. 37) proposes, fronted focus constituents and wh-phrases are [+c], while regular focus is [−c] (and [−a], of course). In the light of these distinctions, PRO_[PL] with-DP can be understood as a complex DP in which the difference—in terms of accessibility—between the two participants it refers to is codified by the combination of the features [+a] and [−a, −c]. The former ([+a]) is responsible for the topicality of one of the referents, while the latter set ([−a, −c]) marks the with-DP phrase. Interestingly, the fact that two opposite features coexist inside the same DP necessarily triggers plurality and codifies the interpretation of a heterogeneous group developed in Section 3.1.

The structure in (29) represents the situation in which there is an associative group that refers to two entities that do not participate in the speech act (example 26). The configuration is similar to 2PL (exclusive), but the duplication of the head Part in (29) is due to the difference between the referents in terms of accessibility.

(29) a. 3PL- associative reading	b. PRO[PL] with-DP, 3PL

Finally, the structure in (30) would represent the rare cases where the verb shows 1PL agreement but where the antecedent is not the speaker but some other referent (examples 27 and 28). The base is a 1PL configuration, but the features [+a] and [−a] are distributed differently from the most regular cases. This causes that p-CON merges in the structure in relation to AuthP, resulting in the expected introduction of a first-person pronoun.

(30)

In brief, the frequency of PRO_[PL] with-DP that triggers 1PL agreement and where the recovered referent is the speaker is not due to a kind of person hierarchy but to discursive factors. Leaving this aside, the contribution of syntax in the inference process is restricted to features that allow us to explain the absence of phonological material for the salient referent, on the one hand, and the mandatory nature of plural number, on the other. In the next subsection, we will present some arguments in favor of a syntax that codifies this kind of information in some way.

In Section 2.1, we discussed a set of data in which the dual interpretation was lost and the comitative phrase cannot be part of the plural subject. The contexts involve focalization and wh-phrases. The relevant examples are repeated below.

(31)	a.	#Con SEVERINO	bailamos	toda la noche.
		with SEVERINO	danced.1PL	all the night
		‘We danced the whole night with Severino.’
	b.	#¿Con quién	bailaron	toda la noche?
		with whom	danced.2PL	all the night
		‘Who did you dance the whole night with?’

According to the hypothesis that the PRO_[PL] with-DP construction is a partially null pronoun in which both [+a] and [−a] coexist, the result of focalization is expected and reinforces the proposal that the with-DP phrase is part of a complex DP. As López (2009) remarks, focus constituents and wh-phrases are [−a], but, most importantly, they are [+c], a feature connected with the left periphery. Even if we were to assume with López that [+a] constituents are also left dislocated, there is a clear incompatibility regarding the final locus in each case and, more obviously, regarding the materialization of the referent’s information. The feature [+c] not only implies movement, but also phonological materialization, i.e., it is incompatible with null information. In brief, the coexistence of the features [+a] and [+c] in a DP seems to be impossible.

Furthermore, the construction under study can be doubled by an overt pronoun, as in (32).

(32)	Nosotros	nos	casamos	en abril	con Ana.
	we	CL.1PL	married.1PL	in April	with Ana
	‘Ana and I got married in April.’

The overt pronoun can present the features [−a, −c] and constitutes a proper antecedent for the [+a] feature of PRO_[PL] with-DP. However, when the overt pronoun presents a contrastive reading, the dual interpretation (Ana and I) is lost, as shown in the translation of B’s answer in (33).

(33)	A:	Juan	y Julia	se	casaron	en abril.
		Juan	and Julia	CL.3PL	married.3PL	in April
		‘Juan and Julia got married in April.’
	B:	#NOSOTROS	nos	casamos	en abril	con Ana.
		we	CL.1PL	married.1PL	in April	with Ana
		‘It was us who got married in April with my wife.’

The relevance of this observation is that the harmony among features when a pronoun is overt explains the PRO_[PL] with-DP in non-pro-drop languages like French (remember (1a)): the obligatory overt subject is a plural pronoun which becomes more accessible owing to the presence of a structure with compatible discursive features.

The last point we would like to include here regards the occurrence of the overt plural pronoun. As has been mentioned, it does not have a contrastive reading. In addition to the constituent order in (34), the one represented in (35) is also possible. In both orders, the overt pronoun appears to the left of PRO_[PL] with-DP. Remarkably, the relevant interpretation is lost when this order changes (32).

(34)	Nosotros	con Ana	nos	casamos	en abril.
	we	with Ana	CL.1PL	married.1PL	in April
	‘Ana and I got married in April.’

 

(35)	#Con Ana	nos	casamos	en abril	nosotros.
	with Ana	CL.1PL	married.1PL	in April	we
	‘My wife and we got married in April.’

Leonetti and Escandell Vidal (2020) note that strong pronouns as subjects are in competition with null subjects, because the former are used only when certain specific information-structure factors justify them. This observation involves preverbal as well as postverbal strong pronouns, and a factor that justifies them, beyond contrast or emphasis, is referent identification. We hypothesize that in cases like (32) and (34), the strong pronoun is a topic originally merged at the left periphery of the sentence and that in most contexts it is used to change the reference from 1SG to 1PL. This means that the subject in (32) and (34) is just PRO_[PL] with-DP. Although this idea should be explored further, we believe that this change of reference is the key to understanding why the order in (35) blocks the dual number interpretation for 1PL.

The hypothesis developed here opens an interesting panorama for language variation. Syntax can combine features in different ways, but not all languages can lexicalize each combination. The Spanish varieties discussed in this paper lexicalize the combination [+a]/[−a] with a PRO_[PL] with-DP, but this option is not spread through all Spanish varieties. Needless to say, further research on this perspective could offer an explanation for the restrictions observed in Section 2.2 regarding syntactic functions.

4.2. Summary

This section delved into PRO_[PL] with-DP from an information-structure perspective. This construction has the same distribution as a null subject and, accordingly, it is unable to convey contrast. The covert referent is contextually determined, and the with-DP constitutes new information. These two properties are codified by the features [+a] and [−a], respectively, which allow us to codify in the syntax the accessibility proposed for the interpretation of the referents. Finally, we argued that the presence of these features in the syntactic domain explains the mandatory presence of plural number for the complex DP and the impossibility of maintaining the dual number reading when the with-DP is involved in a fronting focus or in a wh-constituent.

5. Final Remarks

The analyses of PRO_[PL] with-DP, at least in Spanish varieties, cannot be complete if we do not take into consideration syntactic aspects as well as informative characteristics. The key feature is that this structure is the strategy some languages offer to materialize an asymmetric informative relation between the two referents involved. This asymmetry triggers plural number because the syntax of PRO_[PL] with-DP involves hypermarking: the combination [+Auth]/[−Part], [+Part]/[−Part] or [−Part] [+a]/[−Part] [−a]. Moreover, the possibilities found are not restricted to any kind of person hierarchy, as a syntactic property of languages, but to discursive patterns which affect the frequency of use for each possibility. As we have seen, a 1SG pronoun can be introduced as new information and so is the annex for an anaphoric 3SG, but the most frequent discursive situation is the opposite, where the most accessible referent is the speaker, and, consequently, it can be null.

In syntactic terms, we propose that with-DP is part of a pronominal structure in which plural number is triggered by feature hypermarking. The general interpretation of 1PL as an associative plural is a cue to understand the syntax of PRO_[PL] with-DP and its referential properties. The dialogue between syntax and information structure offers the adequate frame to analyze the sequence under study.

A few words need to be said about variation in Spanish. As was pointed out throughout the paper, associative/heterogeneous plural pronouns can be found in all languages, but PRO_[PL] with-DP is restricted to certain languages/varieties. As shown in (1), this property is not conditioned by being a pro-drop language or a non-pro-drop language. Our hypothesis is that the merging of a with-DP in the context of pronominal projections is not allowed in the grammar of many languages as a strategy for identifying a less salient referent. In fact, Spanish varieties differ in this aspect, because in some varieties the PRO_[PL] with-DP is not observed and speakers do not make the dual interpretation when exposed to these data. At the opposite extreme, there are varieties in which the p-CON can merge in relation to AuthP, as discussed in Section 4.1.

This description could open relevant research regarding microvariation and containment: if a variety presents PRO_[PL] with-DP, 3PL—the structure represented in (29b)—it also presents PRO_[PL] with-DP, 2Pl and 1PL; if a variety allows p-CON in relation to AuthP, it also allows it in relation to PartP. All in all, the structures and features proposed in this paper are consistent with current studies of (micro)variation (see Bobaljik 2012 and the Nanosyntax literature inspired in his proposal). A final remark on variation concerns the distinction between syntactic possibility and pragmatic oddness. The lower frequency of a particular combination of features does not mean that this combination is avoided by the syntax. This is a point that is relevant to the way our analysis was conceived.

Of course, there are some aspects that deserve more research. The most remarkable one regards syntactic functions: PRO_[PL] with-DP cannot appear as an accusative or a dative object. This is an important difference from the Slavic version of PRO_[PL] with-DP, and it also contrasts with the fact that it can be the subject of passive constructions (i.e., it can be an internal argument) and materialize quirky subjects (dative). Probably, further research on information structure could give us some answers to this puzzle.