A mme e tte : An Associative Plural of the First Person Pronoun in Abruzzese

In Abruzzese, the structure a mme e X is used to convey the notion that the speaker and someone closely related to them act jointly. The order of conjuncts is fixed: the first one is always a first person pronoun, while the second one, usually a second person pronoun, strictly denotes a close acquaintance of the speaker. The present ar- ticle argues that this construction is an associative plural of the first person pronoun, something that has not been previously identified in Romance. On top of providing new empirical data, the evidence presented here bears on whether there is a formal differ-ence between associative plural pronouns and regular plural pronouns, a question that has been debated in the literature for quite some time.


Introduction
In Abruzzese, an upper-Southern Italo-Romance language, the semifixed construction a mme e X (in italics in (1)) is used to convey the meaning that a speaker and a close associate of theirs act together.
(1) Parləmə a mme e tte. talk.IMP.1PL to me and you 'Let's talk, me and you.' The data in (1) have not been previously described in the literature, and constitute an interesting puzzle for their peculiar syntactic, semantic and pragmatic characteristics. First of all, the structure a mme e X constitutes the subject of the sentence in which it occurs but it is not in nominative case. Secondly, the second conjunct in the structure, the X, always denotes a close acquaintance of the speaker. Thirdly, this structure is used to convey intimacy between the speaker and one of its closest friends/relatives, and the fact that they act jointly. Building on work by Ackema and Neeleman (2018), I argue that this structure is an associative plural of the first person pronoun, showing that this kind of pronoun is attested in Romance. Indeed, the a mme e X construction in Abruzzese fit Ackema and Neeleman's (2018) prediction on associative plural pronouns. They maintain that if a language has associative plurals of the first and second person pronouns, denoting the speaker or the addressee and their permanent associates, they will always be formally distinct from regular plural pronouns. The data from Abruzzese confirm their prediction: a mme e X is used to refer to the speaker and a close friend or family member of theirs, while the regular plural nu 'we' is used to denote a set made up of the speaker and any other individual salient in the context. More specifically, I argue that Abruzzese resorts to Differential Subject Marking to distinguish between associative plurals and the regular plural of the first person.
The paper is structured as follows: in section 2, I review empirical data from Abruzzese, providing a description of the syntactic and semantic characteristics of the a mme e X construction. In section 3, I briefly review the literature on plurals of first and second person pronouns: I follow Ackema and Neeleman (2018) in arguing that they refer only to contextual associates of the speaker and the hearer, and that they are formally distinct from associative plurals of the first and second pronouns. Section 4 highlights (dis)similarities between associative plural pronouns in Abruzzese and other phenomena marking close familiarity with the speaker that have been identified in the typological literature. Section 5 concludes the paper, pointing out some of the questions that remain open, and hinting at future directions of research.

Syntactic Characteristics
Abruzzese is an upper-Southern Italo-Romance language spoken in the Italian region of Abruzzo. 1 It is mainly an oral language, and shows a huge degree of microvariation even among towns and villages that are relatively close to each other. Therefore, even though I use the term 'Abruzzese' in this essay, I actually bring data from Adriatic Eastern Abruzzese (Loporcaro 2009, 68-9) and, more specifically, from Lancianese, the variety spoken in the town of Lanciano. It is worth pointing out that there are no monolingual speakers of Abruzzese (and more generally of Italo-Romance languages) anymore, and all of them also speak Italian (or, more precisely, a regional variety of Italian), which inevitably influences their Abruzzese. The a mme e X construction is used in Abruzzese to convey the idea that the speaker and a close acquaintance of theirs act jointly. It is found with verbs of any kind of thematic structure. In (1) there was an unergative verb, while in (2) and (3) we can see an unaccusative and a transitive respectively.
(2) Ieməcə a mme e tte. 2 go.IMP.1PL-CL.loc to me and you 'Let's go together, me and you.' 1 Italo-Romance languages are sometimes referred to as 'Italian dialects'. However, they are not local varieties of Italian, but Romance languages that originated from Latin independently of Italian. Therefore, the use of the term 'dialect' in this contest may be misleading, especially for readers not acquainted with the complex socio-linguistic situation of Italy. Thus, in what follows, I refer to Abruzzese as either 'language' or 'variety', and not as 'dialect'.
2 There are no fixed writing conventions for the written rendering of Abruzzese. Therefore, following previous literature (e.g., D'Alessandro 2014; Garzonio, Poletto 2013, a.o.), I use Italian orthographic rules to write Abruzzese. However, I also make use of the IPA symbol schwa /ə/ since this phoneme is present in Abruzzese but not in Italian. Finally, I overtly mark raddoppiamento fonosintattico (gemination of the ini-(3) I lə magnəmə a mme e tte. CL.refl.1PL CL.it eat.1PL to me and you 'You and I will eat that (thing) together.' As shown by (1) and (2), the main context of occurrence of this construction is with hortatives, but at least some speakers allow it in other moods and tenses, as in (3) and (4).
(4) Emə itə a mme e essə. have.1PL gone to me and him/her 'S/he and I went (together).' The main syntactic characteristics of associative plural of the first person in Abruzzese can be deduced from the previous examples. First, the main verb or auxiliary shows first person plural agreement (signalled in italics in (4)), and the subject is constituted by a conjoined structure, in this case a mme e essə 's/he and I'. The fact that this is actually the subject of this sentence is shown by the ungrammaticality of (5) where another overt subject is present. 3 (5) *Nu emə itə a mme e essə. we have.1PL gone to me and him/her (intended) 'S/he and I went (together).' The conjoined structure that characterised all of the examples can be schematised as in (6).
a mme e X The first element in the structure is the preposition a 'to', which I will argue is a differential case marker of the subject (see § 3). It is also homophonous with the accusative marker of the differential object marking, and it engenders raddoppiamento fonosintattico, i.e., gemination of the of initial consonant, on the following word (see Fanciullo 1997;Loporcaro 1997, a.o.). The second element in the structure, tial consonant of a word in specific phono-syntactic contexts; see Borrelli 2002;Fanciullo 1997;Loporcaro 1997;Napoli, Nespor 1979, a.o.) with a reduplicated consonant in word-initial position.
3 Abruzzese is a pro-drop language, and the restrictions on the realisation of an overt subject are pragmatically similar to that of other pro-drop languages, such as Italian and Spanish.
i.e., the first conjunct, is mme, the first person pronoun in the oblique case. Then, there is the conjunction, e, which also engenders raddoppiamento fonosintattico on the following word. Finally, we have the second conjunct, marked as X in (6). The great majority of speakers allow only a second person pronoun in this position, but some speakers also allow a third person singular pronoun (as in (4)), and a personal name or a full DP as in (7). Despite intraspeaker differences, the X always denotes a close acquaintance of the speaker, either a family member or a close friend of theirs, as shown by the ungrammaticality of (8). The second conjunct too is in oblique case, but this is overtly marked only on the second person pronoun. 4 Moreover, it is not possible to have a non-human DP as the second conjunct, as clearly shown by the ungrammaticality of (9). It is clear from this data that only DPs that are found high in the Animacy Hierarchy (Silverstein 1976) can occur as the second conjunct in this structure.
(10) addressee > 3rd person > kin > human > animate > inanimate Moreover, as we have seen, some speakers only accept a second person singular pronoun as the second conjunct. Interestingly, the speakers who accept a third person singular pronoun or a DP as the second conjunct always also accept a second person singular pronoun in the same position. In other words, there is an implicational relation that allows the occurrence of third person pronoun/kin terms in the second conjunct only when the second person is already allowed. This behaviour is expected given that crosslinguistically many linguistic phenomena, such as for instance Differential Object Marking (DOM) in many languages (Silverstein 1976;Dixon 1979, among many others) or morphological marking of the dual in Arapesh (Corbett 2000, 91), do not apply to the whole range of DPs, but only to DPs high in the Animacy Hierarchy. 5 The a mme e X construction is strictly dual, i.e., both of the conjuncts need to be singular, as shown by the ungrammaticality of (11). Moreover, it is not possible to have more than two conjuncts in this structure, as visible from the ungrammaticality of (12). Finally, even though the a mme e X construction is usually found postverbally, it can occur preverbally, as in (13). 8 5 There are however other accounts of DOM that analyse it as a reflex of definiteness (Lyons 1999;Aissen 2003), or specificity and topicality (de Swart, de Hoop 2007;Leonetti 2008). 7 It is possible to convey the meaning of this sentence, but simply not with this construction. One needs to revert to the normal nominative subject: 8 It would be necessary to investigate further where a mme e X is in the clausal structure when it occurs postverbally, and whether it occupies the same position with all kinds of verbs. It might be that the subject stays in situ or moves to some other low position. However, it seems implausible that a mme e X is right dislocated since it is not de-accented nor pronounced with a comma intonation.
(13) A mme e sorəmə n'i parlemə cchiù. to me and sister-my not-CL.REFL speak.1PL any-more. 'My sister and I don't speak to each other anymore.' Summarising, we have seen that a mme e X has a fixed structure, is the syntactic subject of the sentence in which it occurs and triggers first person plural agreement on the verb. It can be used with verbs of any thematic structure as long as they denote actions that are performed jointly, and it usually surfaces postverbally.

Associative Plurals and Plural Pronouns
In the typological literature, a distinction is drawn between additive plurals on the one hand and associative plurals on the other. Additive plurals denote homogenous sets. Regular plurals of R-expressions are additive plurals: indeed, the plural 'girls' denotes a set in which every member is a girl (cf. Barulin 1980). On the other hand, associative plurals are defined as plurals that refer to an individual and their closest associates, i.e., relatives, friends and more in general close acquaintances of an individual (Corbett 2000, 101). Associative plurals are found in many different languages, such as Hungarian, Central Pomo (Pomoan family, Northern California), Central Alaskan Yup'ik (Eskimo Aleut), Sanskrit, Old Egyptian, Basque (Corbett 2000, 101-10). For instance, Hungarian has two different plural endings, one for regular plurals and one for associative plurals, as clear from (14).
(14) a. János-ok John-PL 'Johns' (more than one person called John) b. János-ék John-ASSOC.PL 'John and associates', 'John and his group', 'John 'n' them' (Hungarian; Corbett, Mithun 1996, 5) (14a) is an example of an 'additive plural' since Jánosok denotes a homogenous set, i.e., a set made up of people called John (cf. Barulin 1980). On the other hand, associative plurals denote an heterogenous set, which, in the case of (14b), is made up of John and his associates (Daniel, Moravcsik 2013). At least prima facie, there are similarities between associative plurals of R-expressions and plural pronouns of the first and second person, in that they both denote heterogenous sets. Moravcsik (1994) argues that first and second plural pronouns are associative plurals, denoting sets made up of the speaker and the addressee and their permanent associates. However, Daniel (2020, 4) notes that even though plural pronouns denote heterogenous sets, they do not impose restrictions on the other members of the set. In other words, the plural pronoun 'we' denotes a set made up by the speaker and some other individual salient in the context, but there is no requirement for these individuals to be close associates of the speaker. Indeed, in a sentence like (15), 'we' refers to the speaker and someone that the speaker has just met, surely not someone they have a close relation to.
A: Who is that girl? B: I don't know. We met in the shop and started talking.
Obviously, 'we' can also be used to refer to a group made up of the speaker and their close associates, but, differently from associatives, it does not have to. In other words, first and second plural pronouns are nonhomogenous plurals, but they do not impose any restrictions on the other elements of the set which are not the speaker and the hearer. Clearly then, first and second plural pronouns are neither additive plurals (they do not denote a set containing many instances of the speaker or of the hearer) nor associative plurals (they do not denote a set which necessarily refers to the closest associates of the speaker or of the hearer). On top of the semantic reasons we have just seen, there are also morpho-syntactic reasons to doubt that first and second plural pronouns are associative plurals (pace Moravcsik 1994). Ackema and Neeleman (2018, 88-98) analyse a few languages (Hungarian, Talitsk Russian, Turkish, Central Askan Yup'ik) that have associative plurals and that mark them in various morphological and syntactic ways. In none of these languages, the morphological or syntactic strategy used to mark associativity on R-expressions is employed to create regular plural pronouns. Thus, Ackema and Neeleman (2018, 96) predict that if a language has associative plurals of pronouns, these will be 'formally distinct' from regular plural pronouns, and will be used only to denote groups made up by the speaker/hearer and their permanent associates, i.e., their close acquaintances. 9 In other words, the associative connotation of these plurals must be overtly marked in order to distinguish them from regular plural pronouns (which, as we have seen, denote a set of the speaker/hearer and any other individual salient in the context). There are indeed languages that mark associativ-9 By 'permanent' Ackema and Neeleman (2018) do not necessarily entail life-long or ever-lasting bond: they use this term to refer to some amount of time long enough to establish a stable and strong relationship. ity on plural pronouns. For instance, Dutch has a regular first person pronoun as in (16a) and one referring only to the speaker and their family or other permanent associates (Ackema, Neeleman 2018, 96-7). The latter is exemplified in (16b), and it is made up of two conjuncts: the first is a first person pronoun or a personal name and the second consists of a definite determiner and possessive. Thus, Dutch employs a syntactic strategy to distinguish associative plurals of pronouns, while other languages resort to morphological differentiation. Indeed, in Haruai, a Piawi language of New Guinea, the regular first plural pronoun is an and triggers plural agreement on the verb, as in (17a). However, when the singular first person pronoun occurs with plural agreement on the verb, the sentence acquires an associative reading as in (17b). 11 (17) a. An hön pay-n-ŋ-a. we pig hit-FUT-1PL-DECL 'We will hit the pig.' (Haruai, Comrie 1998, 146 quoted in Ackema, Neeleman 2018, 98) b. N dy-n-ŋ-a. I go-FUT-1PL-DECL 'I and some others will go'. (with associative reading) 12 (Haruai, Corbett 2000, 192 quoted in Ackema, Neeleman 2018 Now that we have a better understanding of the crosslinguistic picture on associative plurals and associative plural pronouns, let's return to the Abruzzese data. The regular plural of the first person pronoun in Abruzzese is just nu, in italics in (18).
10 Obviously, English 'you and yours' closely resembles the Dutch data in (17b), but it is attested only in the second person, while in Dutch it is much more productive.
11 This strategy is used to mark associativity on R-expressions in other languages.
For instance, in Talitsk Russian, lack of number marking on a noun and plural agreement on the verb engenders an associative reading.
12 As pointed out by Ackema and Neeleman (2018, 98), Comrie's (2000) translation of (17b) does not make the associative reading salient, but in the surrounding text he comments that the other people who take part in the action must be permanent associates of the speaker. However, as we have seen, Abruzzese has also a different plural of the first person, as in (19).
Ci iemə a mme e tte. Associative plural CL.LOC go.PRES.1PL to me.OBL and you.OBL 'We will go there (together), you and I.' This second plural, what I have called the a mme and X construction throughout the paper, is really just an associative plural of the first person pronoun. This is line with Ackema and Neeleman's (2018) prediction that if a language has associative plural pronouns, these will be formally distinct from regular plural pronouns. Analysing the a mme and X construction as an associative plural allows us to explain two empirical facts: first of all, why there are restrictions on the second conjunct (as we have seen, it can only be a close acquaintance of the speaker) and, secondly, why these pronouns do not bear nominative case despite being the subjects of the sentence. Indeed, oblique case is the way in which Abruzzese marks the distinction between regular plural pronouns and associative plural pronouns. 14 Therefore, Abruzzese resembles Dutch in using a conjunction structure to mark associativity, but, on top of conjunction, it also resorts to Differential Subject Marking (DSM) to highlight the distinction between regular and associative plural pronouns. Indeed, subjects in Abruzzese usually receive nominative case, as visible from (20). 13 Central and Southern Italo-Romance languages, among which Abruzzese, do not have a synthetic form for the future and rely on present indicative or deontic modals to encode futuricity (Rohlfs 1968, 333;Schwegler 1990, § 5.3.2;La Fauci 2006;Ledgeway 2012, § 4.3.1.5;Ledgeway 2016).
14 The fact that the distinction between the associative and the standard plural of pronoun is present only on subjects might therefore be due to a morphological accident: subjects, usually marked in nominative case, bear oblique case when they are associative plurals, but direct and indirect objects would anyway receive oblique case and so there is no morphological way to distinguish between the associative and the standard plural on them.

Clara Cuonzo A mme e tte: An Associative Plural of the First Person Pronoun in Abruzzese
Thus, the occurrence of the case marker a and of the oblique form of pronouns in the a mme e X construction is unexpected, and constitutes a case of DSM. According to Woolford: a language may be said to have Differential Subject Marking (DSM) if some subjects have a different Case, agree differently, or occur in a different position than others [:] […] such differences are thought of as DSM effects only if they depend on the features of the subject in some way. (2008, 17; emphasis added).
Here, what triggers the DSM is precisely the associative nature of this plural pronoun, that needs to be distinguished from the regular plural. 15 Therefore, the Abruzzese data conform to the general crosslinguistic picture in distinguishing between regular and associative plural pronouns, and do so by differentially marking the subject. The fact that the distinction between the associative and the standard plural of pronoun is present only on subjects might be due to a morphological accident: subjects, usually marked in nominative case, bear oblique case when they are associative plurals, but direct and indirect objects would anyway receive oblique case and so there is no morphological way to distinguish between the associative and the standard plural on them.

4
More on the Crosslinguistic Picture As we have seen, the data from Abruzzese can be easily explained when adopting a framework à la Ackema and Neeleman (2018) who argue that regular plural pronouns are not associative plurals (pace Moravcsik 1994) and that associative plural pronouns need to be formally distinguished from the regular plural pronouns. This section will be devoted to reviewing other typological data on how languages encode associativity and related phenomena. In the typological literature, the existence of inclusory pronominals has been reported for various languages (see Bril 2010 for a recent overview, but also previous works by Lichtenberk 1983;Schwartz 1988;Aissen 1989;Haspelmath 2004, 25-6, a.o.). 16 Inclusory pronominals are (a family of) construc-15 As pointed out by an anonymous reviewer, in Abruzzese (differently from Ducth) the syntactic form '1st person pronoun & X' is not enough to distinguish the associative plural from the regular 1st person plural pronoun, and DSM is also required. Synchronically, there is no reason why this should be the case, but the DSM could be a relic of previous stages of the language in which a mme e X was just a PP and then got reanalysed as the subject of the sentences in which it occurred therefore acquiring its associative meaning.
16 For a formal analysis of inclusory pronominal in Icelandic instead, see Sigurðsson, Wood 2020. tions denoting groups. More specifically, in these constructions both the whole set and one of the members of the group are overtly specified. They are called 'inclusory' precisely because of this double reference to the whole group and some of its members. An example of inclusory pronominals, coming from Dan-Gweetaa of the Mande family, is provided in (20). In this example, the pronoun yāā 'we' refers to the whole set, while Gbȁtȍ is overtly specifying one of the participants.
(21) yāāGbȁtȍyá nū. we Gbato we.AUX come 'Gbato and I, we have come. ' (lit.: 'we [including] Gbato have come') Dan-Gweetaa (Vydrin 2010, 165 as quoted in Khachaturyan 2019 In the typological literature, 'inclusory pronominals' is used as an umbrella term to characterise a wide array of constructions, including the Italian data in (22), which are ambiguous between an 'I-and-Mary' reading and 'we-and-Mary' reading. 17 Clearly, in (22) the overt pronoun noi 'we' is not overtly realised, but is entailed by the first person plural agreement on the verb. The Abruzzese data we have analysed in previous sections cannot be analysed as inclusory pronominals since the a mme e X construction is not 'inclusory': indeed, in this construction the first person pronoun is singular and so it does not refer to the whole set of participants. However, there are similarities between a mme and X and inclusory pronominals: first, the fact that they are used only with human DPs, but also the fact that they refer to sets of close acquaintances (Bril 2010, 362). Indeed, although inclusory pronominals are not examined in Ackema and Neeleman (2018), it may well be that (at least some instances of) inclusory pronominals are an instantiation of associative plural pronouns. However, in order to be able to argue this with certainty, it would be necessary to determine whether inclusory pronouns always and only refer to permanent associates of the speaker and addressee.
Another interesting comparison could be drawn between the Abruzzese a mme and X construction and what Daniel (2020, 4) calls "kin pronominal plurals", i.e., pronouns referring to a speaker and their family. According to Daniel (2020, 4), examples of kin pronominal plurals are found in Sursurunga. In Sursurunga, a South New Ireland/ West Solomonic language of the bigger Austronesian family, the trial/lesser paucal denotes small groups, or nuclear families (Corbett 2000, 27). On the other hand, the quadral/greater paucal is used to denote groups of four or more people (Corbett 2000, 26-9 building on Hutchisson 1986 andp.c. to Corbett). 18 However, there are two additional uses of the greater paucal: it is used when expressing kinship pairs (such as uncle/nephew-niece relationships), but also for hortatory discourse (Corbett 2000, 26-7). In particular, first person inclusive greater paucal can be used to entail joint action with the speaker (2000,27). In another Austronesian language, Marshellese the quadral/paucal is similarly employed to express 'intimacy' (Corbett 2000, 29-30 quoting Bender 1969. What seems to be happening in these languages is that the use of quadral/paucal conveys notions of joint action and close relation with the speaker (irrespective of group size). 19 Moreover, the hortatory use of the Sursurunga greater paucal is intriguingly similar to hortative use of the Abruzzese a mme e X construction. More generally, it appears that languages employ various strategies to familiarity with the speaker, but differ in the ways they mark these notions.

Conclusions
The main aim of this paper was to describe the characteristics of the a mme e X construction in Abruzzese, which has not been previously discussed in the literature. In particular, we have seen that a mme e X has a fixed structure, made up of the first person singular pronoun in oblique case and a second conjunct, which is usually the second person singular pronoun in oblique case. Importantly, the second conjunct in the structure, the X, always denotes a close 18 Sursurunga's number system was originally analysed as having a-five-way distinction: singular, dual, trial, quadral and plural (Hutchisson 1986). However, Corbett (2000, 26-9) argues that the trial and the quadral are respectively a paucal and a greater paucal since they do not denote sets of precisely 3 or 4 members, but more generally to small(-ish) groups of people. Moreover, the quadral/greater paucal is available only for pronouns.
19 As noted by an anonymous reviewer, these two notions, joint action and closeness to the speaker, are distinct and it's not clear why they should constitute a natural class. However, it might be that they tend to co-occur in many pragmatic contexts, i.e., groups of close relatives/friends tend to act together. This pattern is indeed attested crosslinguistically, for instance Turkish associative plurals only allow collective readings (Ackema, Neeleman 2018, 93-4).
acquaintance of the speaker. This conjoined structure is the syntactic subject of the sentence in which it occurs and triggers first person plural agreement on the verb. It can be used with verbs of any thematic structure as long as they denote actions that can be performed jointly, and it is usually used in hortative contexts. Indeed, as we have seen, the a mme e X construction is used to convey intimacy between the speaker and one of its closest associates, and the fact that they act jointly. Following Ackema and Neeleman's (2018) distinction of regular plural pronouns and associative ones, I argued that this structure is an associative plural of the first person pronoun used to refer to the speaker and their close associates. In particular, I showed that Abruzzese differentially marks this associative plural.
In the last section of this paper, I briefly reviewed the typological literature on inclusory pronominals and kin pronominals, pointing out (dis)similarities between the a mme e X construction in Abruzzese and other morpho-syntactic strategies to mark familiarity with the speaker. Two lines of research are left for future investigation. First, it would be useful to investigate the a mme e X construction further to understand better its full characterisation and secondly to determine whether similar data are attested in other closely-related Italo-Romance varieties. In this respect, a survey of historical texts would be necessary to determine the historical development that brought to the birth of this phenomenon. Secondly, it would be necessary to have a more comprehensive understanding of the ways in which familiarity with the speaker are marked crosslinguistically, and whether it is possible to formalise this wide range of phenomena under a single formal account.