las ratas: syncretism in the reflexive paradigm in Spanish and Catalan

In this paper I provide a description of the reflexive syncretism found in some Spanish and Catalan neighbouring varieties. In these varieties, the 3 rd person reflexive pronoun se can also appear with 1 st person plural and 2 nd person plural verbs. With dialectal data from the Audible and Spoken Corpus of Rural Spanish (COSER), the Linguistic Atlas of the Iberian Peninsula (ALPI) and from works of other scholars I analyse the distribution of the syncretism in order to establish what person (the 1 st plural or the 2 nd plural) it affected first. I also analyse this syncretism in other Romance languages, in order to see if the person hierarchies previously proposed in the literature are confirmed by these varieties.

In September 2013, during the last large COSER interviews campaign, 1 a team of 23 linguists did dialectology fieldwork in the Spanish provinces of Castelló, València, Alacant and Murcia. At the very first interview I took part in, the speaker showed a regular use of the reflexive pronoun se not only in the 3 rd person, but also in the 1 st and 2 nd person plural.
(1) Lucena del Cid, Castellón, COSER 1308 (Spanish) 2 El novio le decía al padre: The groom him.DAT said to.the father 'The groom used to tell the father:' "Mira, hamos pensao de casarse, ¿a usté qué le parece?" look have.1PL thought.PTCP of marry.INF-3.REFL to you what him.DAT seems '"Look, we want to get married, what do you think?"' "Ay, pues mira, a mí me parece, si vosotros se queréis, oh well look to me.OBL me.DAT seems if you.PL 3.REFL love.2PL '"Oh, well, I think, if you love each other,' pues me parece muy bien que se caséis" well me.DAT seems very well that 3.REFL marry.2PL 'then I think you should get married"' During the following days, we found this usage in several other villages, making it clear that it was not an isolated phenomenon, but a regular one in the area. In this work I will present a first approach (the first one, to my knowledge) to this phenomenon, analysing both data from the COSER interviews and the ALPI. In section 2 I will relate the phenomenon to the typological difference between functionally streamlined and strategically streamlined reflexive paradigms (Faltz 1985). In section 3 I will present a brief summary of parallel phenomena in other Romance languages. Section 4 will be devoted to describing and analysing the phenomenon in Eastern Spanish and in Catalan, in order to propose a historical evolution across the different verbal persons. Lastly, I will provide some conclusions in section 5.

Reflexive systems across the world
In Faltz's (1985: 118-119) typological survey of reflexive systems, he notices that true pronominal reflexive systems may occur in two different forms in the languages of the world. 3 On the one hand, some languages only have a reflexive pronoun in the 3 rd person, using the regular pronouns for every other person. Standard Spanish is a good example: the 3 rd person (both singular and plural) reflexive marker, se, contrasts with the 3 rd person non-reflexive pronouns (le(s), la(s), lo(s)), while in the 1 st and 2 nd persons (singular and plural) the reflexive and the non-reflexive pronouns are the same (me, te, nos, os).
(2) a. [ Faltz considers these systems to be "functionally streamlined", since the reflexive form is found in the only context where ambiguity may be caused by the use of a non-distinct marker, since both 1 st and 2 nd persons have unequivocal reference.
The other typological possibility is to have a system with a distinctively reflexive marker in every person. Russian is a good example of this, with the reflexive form sebja used in every person. He considers these systems to be "strategically streamlined", since they mark co-reference in every case, regardless of person.
(4) videt' sebja 'to see oneself' (Russian) ja vižu sebja 'I see myself' ty videš sebja 'You see yourself' on/ona/ono videt sebja 'He/she/it sees himself/herself/itself' my videm sebja 'We see ourselves' vy videte sebja 'You see yourselves' oni vidjat sebja 'They see themselves' Faltz (1985: 120) finds just a few examples of reflexive systems that cannot be classified within these two types, that is to say, languages where the reflexive marker is used in more persons than only the third, but not in all. He hypothesizes that these are intermediate stages of the process whereby a functionally streamlined system turns into a strategically streamlined one. Although he admits to having little evidence for this, he also proposes that, in this extension, the reflexive marker spreads first to the 2 nd person and only afterwards to the 1 st .
This typological prediction has been challenged by Puddu (2010). In a very interesting work, she looks at a larger sample of languages with "intermediate" reflexive 3 Faltz opposes true pronominal reflexives to fused adjunct reflexives such as English myself, yourself, etc., which, albeit pronouns, still show their origin of pronominal head + noun. For a more detailed description of the formal possibilities of reflexive systems, cf. Faltz (1985: chapter II). Isogloss 2015, Vol. 1 No. 1 Carlota de Benito Moreno 98 paradigms and shows that Faltz's hierarchy is not supported by evidence. According to her, although it is true that reflexive forms normally spread from the 3 rd person singular to others, it is not always the case that this extension always affects the 2 nd person before the 1 st . In Javanese Ngoko (Trans-New Guinea), for example, the reflexive marker spread first to the 1 st person plural (Cole, Hermon, Chonghyuck, Chang-Yong, & Yaping 2003, apud Puddu 2010. On the other hand, she claims that, from a diachronic point of view, when reflexives spread from the 3 rd person to the others, they show up first in the plural and only later in the singular, whereas the opposite situation is unattested. 4 On the basis of data from Italian dialects, Benincà & Poletto (2005) propose that in the evolution from one reflexive system to the other, Romance languages follow the hierarchy in (5). 5 As can be seen, this hierarchy is also in contradiction with Faltz's proposal. (5) 3 rd /6 th > 4 th > 5 th > 2 nd > (1 st )

Reflexive systems in Romance
Although all the major Romance languages have a functionally streamlined reflexive system in their respective standards, as Latin itself did, the strategically streamlined paradigm is not unknown in the family. For starters, Surselvan (Rhaeto-Romance) shows a fully strategically streamlined reflexive paradigm in which all persons are syncretic with the 3 rd : se- (Stimm 1973, Meyer-Lübke 1906, Meneghin 2008 Some other Rhaeto-Romanic and Northern Italian varieties also present a fully formed strategically streamlined reflexive paradigm. According to Meneghin (2008), that is the case in Sutsilvan (Romansh/Rhaeto-Romance), Bergamasque and Bressan (East Lombard) and the variety spoken in Collina d'Oro (West-Lombard). 7 4 From now on, I will use numbers from 1 to 6 to refer to the verbal persons, so as to keep clear the difference between singular and plural persons. Unless otherwise specified and for the ease of explanation, 3 rd person will also be meant to include the 6 th , since they are syncretic in all the varieties mentioned in this paper. 5 Although Benincà & Poletto (2005) do not specify the role of the 6 th person in their proposal, it is clear from their paper that it aligns with the 3 rd . 6 As can be seen in (6), in Surselvan the reflexive marker is not a pronoun anymore, but a verbal prefix. According to Faltz (1985), pronominal systems are one of the possible diachronic sources of verbal reflexive systems, as undoubtedly happened in Surselvan. 7 Mendrisiot and Luganese (West Lombard) are not completely there, but close enough. They show some variation between simple and double forms: 5 th person is always marked by the originally 3 rd person form (sa), but the original form va may be added to it: vialtri SUBJ sa REFL làvuf/vialtri SUBJ sa REFL va REFL làvuf 'you wash yourselves'. The 1 st Pero se escondíamos como las ratas: Isogloss 2015, 1/1 99 Syncretism in the reflexive paradigm in Spanish and Catalan Quite interestingly, intermediate situations between fully functionally and fully strategically streamlined reflexive systems are found in several Romance varieties. This is exactly the case in Rhaeto-Romance and in the Northern Italian dialects, where a considerable amount of variation is found (and, to a lesser extent, in some other substandard Romance varieties). 8 According to the typological research reviewed in the previous section (Faltz 1985, Benincà & Poletto 2005, Puddu 2010), we should expect the following possibilities shown in (7), while any others are unpredicted. (7) a. 3 rd /6 th = 5 th a'. 3 rd /6 th = 4 th b. 3 rd /6 th = 5 th = 4 th c. 3 rd /6 th = 5 th = 4 th = 2 nd c'. 3 rd /6 th = 5 th = 4 th = 1 st 9 d. 3 rd /6 th = 5 th = 4 th = 2 nd = 1 st Both possibilities in (7a, 7a') are attested in Romance, but (7a') seems to be the most frequent by far. Syncretism between the 3 rd person and the 4 th is found in Gherdëina (Ladin), 10 Alessandrine, Turinese and the dialect of Ceva (Piedmontese), Bolognese, the dialects of Bobbio and Roccabianca (Emilian), Genoese (Ligurian), Cremonese (Lombard) and in Supradialectal Venetan (Meneghin 2008). Marco Tamburelli (p. c.) confirms that most Lombard varieties either allow this 3 rd /4 th person syncretism or show the syncretism in all persons in the plural. Puddu (2010) also attests the 3 rd /4 th syncretism in Piedmontese and Venetan. Meyer-Lübke (1906: §380) notices that the use of se in the 4 th person is common all across Northern Italy and in Southern Provençal. Ronjat (1937) corroborates this for Provençal and Blanchet (1992) considers the syncretism to be the norm in that variety. Quint (1998) also attests the syncretism in the Vivaro-Alpine Occitan dialect (to the North of the Provençal domain), but other works on the Occitan language mention that this kind of syncretism is rare (Carrera 2011;Quint 1999). This apparent discrepancy seems to suggest that the reflexive syncretism is not the norm in Occitan, although definitely common in the East of the domain. One example of this syncretism is displayed in (8), from Genoese. person is obligatorily marked by ma, but admits doubling with sa: mi SUBJ ma REFL lavi/mi SUBJ ma REFL sa REFL lavi 'I wash myself' (Meneghin 2008). 8 However, not all Rhaeto-Romanic and Nothern Italian varieties show signs of a strategically streamlined paradigm. Some varieties, such as Surmiran, Rumantsch Grischon (Romansh/Rhaeto-Romance), the dialect of Riva (Lombard) and Orbasque (Piedmontese), show a fully preserved functionally streamlined reflexive system (Meneghin 2008). 9 Since Faltz (1985) only predicted the order 3 rd >2 nd >1 st , (6a') and (6c') are unexpected in his account. However, as said before, Puddu (2010) and Benincà & Poletto (2005) already found counterexamples to Faltz's person hierarchy. On the other hand, (6a) and also (6c') are unexpected according to Benincà & Poletto (2005). Puddu does not make any prediction on whether the same person hierarchy should be followed in the plural and in the singular, but from Northern Italian varieties it is clear that this is not the case, since the most common evolution in the plural is 3 rd /6 th >4 th >5 th and afterwards it reaches the 2 nd singular. 10 The syncretism is only found with reflexive meaning, while reciprocal verbs preserve nes in Gherdëina (Meneghin p. c.). Contrary to what Benincà & Poletto (2005) claim, the syncretism between only the 3 rd and the 5 th persons, although more rare, is also possible in Romance. They actually had a counterexample to their own claim in Trieste, which they attribute to Slavic contact. 12 But that is not the only exception attested in Romance: both Vallader and Puter (Romansh/Rhaeto-Romance) present the solution in (7a): as lavar 'to wash oneself' (Meneghin 2008 Syncretism in all the plural persons, (7b), is also attested in several Romance varieties. Meneghin (2008) finds it in Val Badia, Ladin Dolomitan, Standard Friulian and in Bregagliot (Lombard). 13 As said above, Marco Tamburelli (p. c.) also confirms the existence of this paradigm in Lombard. Meyer-Lübke (1906: §380) also documents it in Friulian. According to Puddu (2010), Campidanese Sardinian can be added to the list. Herzog (1914) states that, in French dialects, se is sometimes used in all the persons of the plural. 14 This solution is also found in some substandard varieties of Catalan (see 11 For the ease of interpretation, I will gloss the paradigms using subscript abbreviations. SUBJ stands for 'subject', REFL for 'reflexive' (which may appear twice, as some varieties have double forms) and SUBJ.CL for 'subject clitic'. 12 Unfortunantely, they do not elaborate this hypothesis further. Since Standard Slovene has a fully strategically streamlined reflexive paradigm with a unique reflexive pronoun sêbe for all persons (Herrity 2000), it is hard to see how the contact hypothesis could explain the allegedly anomalous behaviour of Trieste's dialect (unless it were contact with a Substandard Slovene variety that had se only in the 3 rd and the 5 th persons, but nothing of the kind is alleged in Benincà & Poletto's paper). 13 Bregagliot admits both the etymologically 5 th person form uv and the syncretic us (u SUBJ.CL as REFL ) (Giacometti 2003: 65-68), suggesting that the extension of the spread was 3 rd /6 th >4 th >5 th . Dolomitic Ladin admits variation between se and both the 4 th and the 5 th person forms nes, ves. 14 However, Herzog only provides examples of syncretism in the 5 th person: vous se portez 'you.PL 3.REFL go' (Champlemy, Bourgogne; Nivernais dialect) and vous s'imaginez 'you.PL 3.REFL imagine' (Saint-Pol, Faubourgs; Picard dialect). Johannes Kabatek kindly provides me with the following example (also a 5 th person), vous voulez aller se coucher-là [you.PL want.2PL go 3.REFL lay.down there 'you want to go lay down there'], from the French film Bienvenue chez les ch'tis, which features the "strange" speech in the North of the country (specifically the region Nord/Pas de Calais, right to the North of the Picardy region). Of course, this example cannot be seen as proof of the construction in this area, but it definitely shows that the phenomenon exists in French Pero se escondíamos como las ratas: Isogloss 2015, 1/1 101 Syncretism in the reflexive paradigm in Spanish and Catalan also section 4.2). Fabra (1913-14) attests it in the speech of (mostly the lower classes of) Barcelona and Badia i Margarit (1962) considers this use a pervasive feature of Substandard Catalan. Alcover & Moll (1968) show many examples of this syncretism in Catalan, noticing that it seems more frequent in Valencian, whereas it is unattested in Balearic. Several grammars on Valencian confirm this suspicion (mostly by condemning the use): Guinot i Galan (1987), Salvador (1974) and Sanchis Guarner (1950) also attest (and disapprove of) it. In his sociolinguistic study of Sagunto (Castelló), Gómez Molina (1986) indicates that this use of se is especially common in informal speech. All these Catalan varieties show variation, accepting both the unequivocally reflexive form se and the original forms (nos, vos or its dialectal allomorphs) for the 4 th and 5 th reflexive pronoun. An example of this syncretism is presented in (10) Of the possibilities in (7c, 7c'), only (7c) (all the plural persons and also the 2 nd are syncretic with the 3 rd ) is attested in Romance, as far as I know. This pattern is found in some Northern Italian varieties, namely, the dialect of Medeglia (West Lombard), in Poschiavino (Lombard) (Meneghin 2008) 15 and Milanese (Puddu 2010). I give an example of this distribution in (11), from the variety spoken in Medglia.
(11) laváss 'to wash oneself' (Meneghin 2008 To my knowledge, syncretism in all persons but the 2 nd (the solution in 7c') is not attested in any Romance language. But the Northern Italian dialects also exemplify a solution not completely predicted by Puddu's typological generalisation (plural > singular): the syncretism between the 3 rd , the 4 th and the 2 nd persons, but not the 5 th . According to Meneghin (2008), this is exactly the situation in Milanese and Airolo (although in Airolo the 2 nd person has a double form: ti ti t s làvat). However, Puddu (2010) claims that in Milanese the 3 rd person reflexive has also extended to the 5 th person, and that it is considered vulgar. 15 In Poschiavino (Lombard), there is variation between simple forms and double forms, with se and the original pronoun (ta, ma, va) in all these persons: ti SUBJ ta SUBJ.CL sa REFL làvas/ti SUBJ ta SUBJ.CL sa REFL ta REFL làvas 'you wash yourself'; nualtri SUBJ ma REFL làvum/nualtri SUBJ ma REFL sa REFL làvum 'we wash ourselves', vualtri SUBJ va REFL làvuv/vualtri SUBJ va REFL sa REFL làvuv 'You wash yourselves'. Interestingly, in such a paradigm the only person other than the 3 rd that always shows sa is the 2 nd (singular). 102 so more data on this variety would be very welcome. On this regard, it is interesting to recall that Meneghin (2008) attests optional double forms in the 1 st (ma/ma sa) and the 5 th (sa/ sa va), but not in the 2 nd (where only the corresponding se form is used) in Mendrisiot and Luganese, which could also suggest that the syncretism developed first in the 2 nd than in the 5 th (cf. note 6) -although it may also be the case that the doubling and the syncretism evolve independently. Airolo's paradigm is given in (12).
From a typological point of view, then, it seems that the generalization proposed by Puddu (2010) may need to be qualified. So far, the patterns attested in the Romance varieties I just reviewed allow for the reconstruction of the evolutions in (13)-but note that (13c) seems to be quite exceptional in our sample.
The syncretism has also been attested in Ibero-Romance varieties across the Atlantic Ocean. According to Vidal de Battini (1964), se is used as the 4 th person reflexive pronoun in the speech of uneducated people from several areas in Argentina (in Río de la Plata, in the Northern region and in the Western Cuyo region). It is, of course, also used in the 5 th , since the 5 th person subject pronoun vosotros/as has been replaced by ustedes. Ustedes was originally (and still is in most dialects of Peninsular Spanish) a polite address plural pronoun, which triggers 6 th person agreement (since it derives from the plural noun phrase vuestras mercedes 'your graces'). In these varieties, then, se is generalised as the only reflexive form in the plural. Kany (1945) explains that Lafone Quevedo (1898) attributes this development to Quechuan influence in Catamarca, while he proposes that the cases found in Buenos Aires may be due to Italian influence. He thinks nevertheless that the syncretism was most likely inherited from Spain, since se is frequently used instead of os (the 5 th person object pronoun) in Andalusia and Murcia. 16 However, the fact that the syncretism is found in different persons in Spain and in Argentina poses some problems for such an explanation. I have only managed to find one reference to the use of se instead of nos (4 th person) in Peninsular Spanish: Blas Arroyo et al. (1992) describe it in Castelló (Valencian Country) and consider it a Catalan (Valencian) interference. 17 In Brazilian Portuguese, the use of the 3 rd person reflexive pronoun se in other persons can be seen as a Pan-Brazilian phenomenon, according to de Oliveira (2005). The situation there, as is usually the case in Brazilian Portuguese, is quite diverse. In many (most) varieties, every subject pronoun but the 1 st person triggers 3 rd person agreement (singular or plural), as a result of the reorganization of the pronominal paradigm. On the one hand, the 2 nd person forms have been replaced by the previously formal pronouns você(s) and, on the other, the 4 th person pronoun has been replaced by the form a gente. All these forms historically derive from etymologically 3 rd person noun phrases, which trigger 3 rd person (singular or plural) agreement. Thus, in these varieties, all persons except for the 1 st agree syntactically with a 3 rd person reflexive pronoun (se). It does not seem unlikely, then, that in some varieties se could appear also in the 1 st person, a development that would imply an absolute regularisation of the reflexive paradigm. This is the case in Afro-Brazilian Portuguese, for example, where se is also attested as the 1 st person reflexive pronoun, alternating with me (Alves de Souza 2011).
However, the neutralization of the reflexive form is also found in varieties that preserve some of the original pronouns. For example, the São Paulo dialect preserves the 4 th person pronoun nós, but se is attested in all persons, including the 1 st (eu) and the 4 th (da Silveira 2008;de Oliveira, 2005). In Northern varieties, where both nós and the 2 nd person pronoun tu are preserved, reflexive se is attested with both of them, meaning that the only person that escapes the syncretism is the 1 st (de Oliveira, 2005). In Mozambican Portuguese, according to de Oliveira (2005), se is also attested in the 4 th person (with the 4 th person subject pronoun nós).
It is especially interesting that, although the syncretism is found in so many Romance varieties, quite diverse origins have been proposed for the different varieties. Meyer-Lübke (1906) believes that phonetic proximity between the reflexive and the non-reflexive pronouns may have originated the syncretism in Rhaeto-Romance and Southwest France varieties. Further evidence supporting this would be the fact that, according to Ronjat (1937), in several Provençal dialects se can be used as a nonreflexive pronoun, as in apère-se 'call us'. Although he believes that the use of se instead of reflexive nos is the original context, from which se would have extended to non-reflexive environments, he also thinks that phonetic weakening of nos may have played a role in this development. Meneghin (2008) also finds plausible that phonetic reduction of the 4 th person clitic (after having crossed with the subject clitic: a nos > a nes > ans > as) may have caused the syncretism with the 3 rd person (after a similar (To you.PL 3PL.ACC saw.3PL 'They saw you') and A ustedes os vieron (To you.PL 2PL.ACC saw.3PL 'They saw you') are possible (cf. Lara Bermejo 2012). 17 Unfortunately, it seems that the possibility of resorting to language contact has some responsibility in the lack of studies of these developments in Spanish. While Lafone Quevedo does support his contact hypothesis by comparing the Spanish reflexive with the Quechuan reflexive suffix -ku (he transcribes it -cu), which is identical in every person, Kany adds no more to his hypothesis. Whether this Italian influence is due to contact with Northern varieties that also show the extension, to the phonetic similarity of the 4 th person object pronoun in Italian, ci or to some other possibility, it is to the reader to decide. On the other hand, Blas Arroyo et al. (1992) attribute this feature to Catalan influence because the phenomenon is described in Valencian grammars, although this only raises more questions: how did it happen in Valencian?; how can we know the direction of the interference (if it is actually one)? 104 development: /al səә/ > /als/ > /as/) in the Rhaeto-Cisalpine domain 18 . As he notes, this hypothesis would be reinforced by the fact that Emilian dialects show variation between als and as in the 4 th person.
In some varieties, however, phonetic reasons must be necessarily ruled out. Meyer-Lübke (1906) finds them inapplicable in Veglia and proposes that the situation there may be due to Slavic influence, as Benincà & Poletto (2005) do for the variety spoken in Trieste. A different kind of contact, that with the Standard Italian form ci diamo (we.REFL give.1PL), could explain the confusion in some Italian dialects (like Venetian or Genoese), but not in others (Lombard, Emilian, Tuscan) or in Provençal, also according to Meyer-Lübke (1906). He similarly discards a possible Genoese influence on Provençal, for example, because in the latter the phenomenon is too widespread. As we saw, Blas Arroyo et al. (1992) also attribute the syncretism in the Spanish of Castelló (Valencian Country) to contact with Catalan.
A different kind of explanation is proposed by Meneghin (2008), concerning the extension to the 2 nd person. According to him, the use of se in the 2 nd person in some Northern Italian varieties, like Ticinese, is best explained by the fact that 2 nd person verbs are compulsorily preceded by at least the unstressed subject clitic, which is identical to the 2 nd person object pronoun. The use of the unequivocally reflexive form se would hence avoid the repetition of the 2 nd person pronoun two times and make clearer the reflexive interpretation. He finds further support to this hypothesis in varieties like Ticinese, where (ti) ta nètat/nètas 'you.SG clean' and (ti) ta sa nètat/nètas 'you.SG clean yourself', but *(ti) ta ta nètat/nètas 'you.SG clean yourself'.
Although internal pressure from a pronominal paradigm where most pronouns trigger 3 rd person agreement may explain the syncretism in some varieties of Brazilian Portuguese, we saw that not all of them show such a homogenous paradigm. I have not found any explanations for the extension of se in Portuguese in the literature, except for Castilho (1997), who states that reflexives tend to generalise in all persons, especially in non-standard languages, which is of course more of a description of a common crosslinguistic situation than an actual explanation.
I would like to end this overview of the reflexive paradigms in Romance by mentioning a somehow similar phenomenon that has been attested in European Portuguese. Martins (2003) finds that in some varieties of Madeira and Porto Santo, se may appear in sentences with non 3 rd person verbs. These sentences do not contain, however, reflexive verbs, but this seems to be an indefinite se indicating that the sentence has a generic or indefinite subject (which may or may not be overt). As can be seen in the examples in (14), this indefinite se also appears with 3 rd person (nonreflexive) verbs (14b, c). 18 Such a hypothesis or a slightly broader one in which the fusion of the subject clitic and the reflexive pronoun would have created a syncretic reflexive reduced form -s for two or more persons in the plural can be alleged for several Northern Italian varieties: Airolo (West Lombard), dialect of Ceva, Orbasque, Alessandrine (Piedmontese), dialect of Bobbio, dialect of Roccabianca and Bolognese (Emilian).
Pero se escondíamos como las ratas:  [Talking about the celebration that took place before the young men left to the military service] (Urriés, ZA 4720) Pues cada, pues se, se invitaban uno en cada casa, Well every, well 3.REFL, 3.REFL invited.3PL one in every house 'Well, every…, well, they invited one to every house,' uno, uno en cada | con los que había one, one in every with the.M.PL that were.3SG 'one, one to every…, with those that were there' These Spanish and Portuguese examples could be the result of fusing two impersonalising devices. In Spanish, non-specific subjects may be indicated by two main means. On the one hand, the reflexive passive and the reflexive impersonal constructions, marked by the reflexive pronoun se. On the other, 2 nd , 4 th and 6 th person verbal morphology can also form generic or indefinite agent constructions. Although the examples in Spanish are quite sporadic, the fact that they were found in Eastern Spanish may suggest some relationship with the extension of se to other persons that should be further explored when more data are available.
I will now turn to describe the reflexive syncretism found in Peninsular Spanish and Valencian Catalan and try to see how it fits in the hierarchies and developments found in the Romance varieties reviewed in this section.

The reflexive paradigm in the Valencian Country and Murcia
To my knowledge, there are no studies focusing specifically on the extension of reflexive se to other persons in either Spanish or Catalan, although, as seen above, there are some mentions in the literature. In this section I will be looking at this syncretism in the Valencian Country and Murcia in both Spanish and Catalan.
Most of the data in the present study come from the COSER. The corpus consists of semi-directed interviews (also known as sociolinguistic interviews) targeting elderly, rural, non-mobile and low educated speakers from all around the Spanishspeaking area in Spain. The interviews' duration is variable, but the mean is around 75 minutes per village.
For this paper I use data from 19 different villages in València, Castelló, Alacant and Murcia. Since these provinces were only enquired in September 2013 and most of the interviews have not been transcribed yet, I first checked the interviewers' field notes in order to see in which localities the syncretism had been attested. I afterwards heard those 19 interviews and transcribed all the examples of 4 th and 5 th person reflexive verbal forms, since in the field notes there was no notice of the syncretism appearing in the singular. In the end, I got a corpus of 385 examples of 4 th and 5 th person reflexive verbal forms.
I also use data from the Linguistic Atlas of the Iberian Peninsula (ALPI from now on, for its acronym in Spanish). 21 Most of the ALPI interviews took place in the 1930s, but the Spanish Civil War interrupted the project, which had to be completed between 1947-1956 (for more information on the ALPI's history, cf. the introduction to the only published volume, ALPI (1962) and also Navarro Tomás (1975) and Heap 21 Only the first volume of the ALPI was published, but the materials are both available online (http://westernlinguistics.ca/alpi/) and at the Universidad Autónoma de Madrid, where I could complete some information that was missing from the website materials.

Pero se escondíamos como las ratas:
Isogloss 2015, 1/1 107 Syncretism in the reflexive paradigm in Spanish and Catalan (2003)). Hence, there is a gap of around 80 years between the earliest ALPI campaigns (where all the Spanish and most of the Catalan data was obtained) and the COSER interviews I will be looking at, which allows for some historical perspective.
Unfortunately, the interviews collected in the Dialectal Spoken Corpus (from now on, COD, for its acronym in Catalan) do not attest any example of the syncretism in the variety of Catalan spoken in the Valencian Country. This contrast with the COSER data may be due to two different reasons. On the one hand, the interviewees were younger and more educated and the settings larger than those in the COSER, which could hinder the occurrence of a substandard phenomenon like this one. On the other hand, the available transcriptions correspond to much shorter recordings (of around 10 minutes). Actually, out of 8 interviews (Elx, La Vila Joiosa, Alacant, Sueca (Alacant), L'Alcora, Morella (Castelló), Llíria and València (València)), only one case of 5 th person reflexive verbal forms and 13 cases of 4 th person reflexive verbal forms were found (none of them with the syncretism). I will thus be using the data from the COD questionnaires (not available online) as presented in Perea (2012) and from Alcover's materials as presented in Perea (2012) and Alcover & Moll (1968). 22 In the remainder of this section I will describe these data and propose a hypothesis regarding the diachronic extension of se to other persons in Spanish. I will also connect this development to the situation in Catalan, although I will not propose a diachronic hypothesis for this language.

The syncretism in Spanish
Some, although very scarce, references to the syncretism can be found in dialectal works. Nebot Calpe (1984) notices that se instead of 5 th person os occurs (but not regularly) in the Aragonese dialect of Spanish spoken in the provinces of València and Castelló and that it is most common with reflexive verbs (implying that it may also appear in non-reflexive contexts) and so are all her examples. She attributes it to Valencian contact, without further explanation. Some of her examples are given in (16).

(16)
Nebot Calpe (1984) (Spanish) a. Se laváis to los días a desgusto 3.REFL wash.2PL all the days to unpleasure 'Every day you guys take a shower while complaining' b. ¿se vais a casa la mare? 3.REFL go.2Pl to house the mother '¡Are you guys going to your mum's?' c. se lo topetastis a mi primo en el embolau d' Argelita 3.REFL him.ACC found.2PL to my cousin in the bulls.on.fire of Argelita 'You guys ran into my cousin in the "Bulls-on-fire" feast of Argelita' Guillén García (1974) and García Cotorruelo (1959) confirm the use of se as a 5 th person pronoun in Orihuela (Alacant) and Cartagena (Murcia), two areas generally conceived as belonging to the Murcian Spanish dialect. They do not mention contact with Catalan as a possible explanation, unsurprisingly, since these areas have been monolingual in Spanish for seven centuries. 22 These materials include Alcover's fieldwork notebooks  and the data in "La flexió verbal en els dialectes catalans" and "Estudis de Sintaxis dialectal catalana" (edited in two parts in 1916, Bolletí del diccionari de la llengua catalana (BDLC), IX (11-37 and 49-63), see Perea (2012)).
Blas Arroyo et al. (1992) mention the use of se instead of 4 th person nos as a typical feature of the speech of Castelló (Valencian Country) and, as said above, they consider it a calque from Valencian (Catalan). 23 In the COSER interviews the syncretism was found both in Spanish monolinguals and in Spanish-Catalan bilinguals. This does not necessarily rule out the contact hypothesis as the origin of the innovation, but would imply that the interference settled well enough so as to spread to the monolingual contiguous area.
The geographical distribution of the phenomenon in Spanish according to the COSER data is presented in map 1. 24   23 The same interpretation is attributed in a commentary to the only example found in the PRESEEA (Project for the sociolinguistic study of Spain's and America's Spanish) corpus from Valencia: la gente joven pensamos de otra manera se pensamos que somos los reyes del volante (the young people, think.1PL different, 3.REFL think.1PL that are. The COSER data show that, although the extension of se is quite common, it is definitely not the norm and the standard forms are usually preferred (in a 69,09% of the total cases). In only two villages-Lucena del Cid (COSER 1308, Castelló) and Benimarfull (COSER 303, Alacant)-the speaker presented a consistent use of se in both the 4 th and the 5 th persons. Some others only showed a few sporadic examples and still some others presented quite a lot of variation. 25 It also needs to be said that when there is more than one speaker in the same place, the situation can dramatically change from one speaker to another, going even from an apparent absence of the syncretism to an apparent exclusive use of se. 26 Table 1 presents the global data obtained from the COSER. The data in table 1 indicate that the use of se is much more common in the 5 th person than in the 4 th and that this difference is highly significant, as the very small p value (7.409e -12 ) indicates. 28 This would suggest an evolution 3 rd /6 th > 5 th > 4 th , the least common in the Rhaeto-Romanic and Northern Italian varieties-and rejected as a possible Romance evolution in Benincà & Poletto (2005). The geographical distribution of the COSER data supports such a hypothesis, since 4 th person se occurs in a more restricted area (it is almost absent in Murcia). This could suggest that the focus of the phenomenon is in the Valencian Country.  All the p values were calculated with Pearson's Chi-squared test with Yates' continuity correction, using the software R. 28 The big difference between the total occurrences of the 4 th and the 5 th persons is explained by the fact that our data come from interviews and not free conversation samples, hence the speakers refer more often to themselves and their peers than to the interviewers. Although more data (in any of the slots) would be very welcome and insightful, the robustness of the p value (which controls for these differences) suggests that our results are on the right track.  29 Map 3. Se in the 4 th pl

Map 1. The reflexive syncretism in Eastern Spanish
Such an evolution seems especially plausible in Eastern Spanish, where another syncretism in the pronominal paradigm is attested. In a bigger area, but that also includes the Valencian Country and Murcia, the non-reflexive masculine plural accusative clitic (los) can be used also as the 4 th and 5 th object clitic (both reflexive and non-reflexive) 30 , as shown in (17)  By studying the geographical extension of the phenomenon, Enrique-Arias (2011) has shown quite convincingly that this syncretism started in the 5 th person and spread later to the 4 th : all the villages with los in the 4th also have the syncretism in the 5th, while the opposite is not true. It seems likely, then, that if a new syncretism in the 29 The dots in maps 2 and 3 are pie charts illustrating the proportion between se and other forms in each interview (in raw numbers). Its size varies according to the total number of examples found in each interview. An indication of the equivalence between size and number of examples is given in the legend under the title "sum of fields". 30 The existence of such a syncretism is also an important counterexample to Benincà & Poletto's claim "that Kayne's hypothesis concerning the fact that s+V is in a single paradigm with the m,t,n,v+V forms, while l+agreement is excluded from it, is confirmed by the fact that the extensions found in Romance languages involve the s+V form frequently substituting for the m,t,n,v+V forms, but never for the l+V forms" (Benincà & Poletto 2005: 269).

Pero se escondíamos como las ratas:
Isogloss 2015, 1/1 111 Syncretism in the reflexive paradigm in Spanish and Catalan pronominal system were to be born within this area, it would follow the same person hierarchy, which was already operative in their system. 31 From a diasystematic point of view, several pronominal systems can be identified in the area under study: 1) the standard system, with different pronouns for every person and a distinct reflexive pronoun for the 3 rd /6 th person; 2) the system described by Enrique-Arias, with a single pronoun in the 4 th and 5 th persons, common to one of the 6 th person forms-and a distinct reflexive pronoun for the 3 rd /6 th person, and 3) the system I am analysing here, that grammaticalises the difference between reflexive and non-reflexive pronouns in the plural. Some speakers used all the three different possibilities in a single interview, using, for example, nos, los and se for the 4 th person reflexive (but only nos and los for the 4 th person non-reflexive).
An evolution such as 3 rd /6 th > 5 th > 4 th can be readily explained if we take into account a dialectal phenomenon that could have acted as the main bridging context (Heine 2002) for the spread of se to the 5 th person in the first place. In most varieties of Peninsular Spanish, the imperative is syncretic with the infinitive. 32 That is to say, in most varieties of Peninsular Spanish the original imperative form in /-d/ has been replaced by the infinitive form in /-r/ (cf. Fernández-Ordóñez (2012) for a study of the geographical distribution of this syncretism). Crucially, in some varieties, this infinitive pro imperative takes a 3 rd person reflexive pronoun, as shown by the ALPI data, in map 4. Although this feature is a stereotype of Southern (Andalusian) varieties, the data from the ALPI shows that it has quite a larger extension, including our area of interest. According to these data, the use of se in the infinitive pro imperative is especially common in Andalusia, in a small but compact area to the South of Madrid and in Murcia and the Valencian Country. 31 Although slightly different, two other dialectal phenomena come to mind. First, the levelling of the oblique pronominal paradigm after a preposition, which is also typical of Eastern Spanish, is both more frequent and more widespread in the 2 nd person than in the 1 st : para tú 'for you.SG.NOM' > para yo 'for I' (Pato 2012). In these cases, the nominative forms tú 'you. SG.NOM' and yo 'I' are used after some prepositions instead of the oblique forms ti 'you.SG.OBL' and mí 'me.SG.OBL'. The result is a more regular pronominal system, since in all the other persons the nominative and the oblique are syncretic. Second, the regularization of the clitic linearisation in the combinations of a dative clitic + se is also more frequent and more widespread in the 2 nd person than in the 1 st . That is to say, the vernacular forms te se and me se, used instead of the standard orders se me (se me cayó '3.REFL me.DAT fell.down') and se te (se te cayó '3.REFL you.DAT fell.down'), are quite common in the Peninsula, but the vernacular order is significantly more spread in the 2 nd person than in the 1 st (Heap 2003). The vernacular forms also imply a regularisation, since se is the only 3 rd person pronoun that may appear in the first position in combinations of clitics in Standard Spanish (Martín Zorraquino 1979). All these dialectal phenomena, then, emerge earlier in the 2 nd person than in the 1 st , although the latter would be the only example where the singular preceded the plural in the regularisation (nos se and os se, although existent, are quite rare). 32 As said above, in America the 5 th person nominative pronoun vosotros/as has been replaced by the formal 6 th person form ustedes, so the plural imperative takes the 6 th person subjunctive form.

Map 4. Imperatives in the ALPI 33
That is to say, the apparent mismatch between the 5 th person in the imperative and the 3 rd in the reflexive is most likely explained by the fact that the imperative takes the form of the infinitive. Infinitives, being non-finite forms, take 3 rd person reflexive clitics as the default form with arbitrary, indefinite or generic subjects, for example. According to this hypothesis, this association of the infinitive and se would trigger the appearance of se together with the infinitive in other contexts, like the imperative. Why the imperative and not other controlled infinitives? Two reasons can be suggested. On the one hand, the imperative use would be the only one where infinitives have a controlled subject that can be (and normally is) absent in the clause, which makes the mismatch less salient: (18a) vs. (18b). On the other hand, in some contexts imperatives are semantically similar to generic instructions, which take 3 rd person reflexive clitics (18c The map corresponds to questions 344, ¡Arrodillaos! (Kneel.2PL-2PL.REFL down!), and 345, Bebeos este vaso de aguardiente (Drink.2PL-2PL.REFL that glass of liquor). The solutions in grey, labelled as C-os, include the varied substandard forms that the 5 th person pronoun os may take, all preceded by a consonant (los, sos, tos). The ALPI does not attest the solution seos, which was found in the COSER interviews. To figure out whether this form comes from a doubled pair se + os similar to those attested in some Lombard varieties (although seos is not restricted to reflexive contexts) or it has a different origin is beyond the goals of this paper. 34 Interestingly, in Sacile (Friulian), se is already used in the 4 th person, but in the 5 th is restricted to imperatives, although in this variety there is no imperative-infinitive syncretism (Benincà & Poletto 2005).

Pero se escondíamos como las ratas:
Isogloss 2015, 1/1 113 Syncretism in the reflexive paradigm in Spanish and Catalan c. A callarse de una vez. To shut.up-3REFL of one time 'Now (let's) shut up' The quantitative COSER data (see table 2), although confirming the relevance of the infinitive, seem to suggest that the infinitives that most favour the appearance of se are the non-imperative ones (93,31%). The few data in some of the slots, however, precludes us from taking statistical significance tests, which would not offer any garanties, so more data would be welcome to check if there are some significant differences between the two types of infinitives. On the other hand, the difference between infinitives and finite forms 35 is statistically significant (0.01887) and suggests that the syncretic pronoun is more frequent with the former. Nevertheless, in this particular case the infinitives coincide with enclitic environments and the finite forms with proclitic contexts, so there is no way of telling which factor is the determining one. 36 Enclisis is restricted in Spanish to non-finite forms and the imperative, but no gerunds or imperatives in -d appeared in our 5 th person data set, so all the examples are infinitives. 36 In the los syncretism the position of the clitic is also a relevant factor, although in the opposite direction: los (the syncretic and new form) is quite more often a proclitic than an enclitic. Enrique-Arias (2011) proposes that this is an indication of its phonetic origin, for the 5 th person pronoun is the only one beginning with a vowel. As can be seen, map 5 shows a radically different picture. Proclitic 5 th person se seems to be confined to Western Andalusia and to a few points in Murcia and the Valencian Country. On the one hand, this map proves that the Western Andalusian 5 th person se and the Eastern Spanish one have developed independently, since it is unattested in Eastern Andalusia in proclitic contexts. 38 On the other hand, the map shows a lower frequency of se in proclitic environments in Murcia and the Valencian Country. Although the significant visual difference between map 4 and map 5 is due to 37 These two questions could have actually been ideal for showing examples of enclitic se in a non-imperative context, since the reflexive verb appears in an infinitive periphrasis. However, the clitic appears in the first position, before the inflected verb, i.e. both of them are instances of the so-called clitic climbing. 38 Independently, but most likely from a common contextual usage (infinitive pro imperatives), as suggested by the continuous geographical distribution found along the Southern and Eastern coast in map 4.

Pero se escondíamos como las ratas:
Isogloss 2015, 1/1 115 Syncretism in the reflexive paradigm in Spanish and Catalan the fact that neither caure (Cat. 'fall down') nor callar (Cat. 'shut up') are reflexive in most of the Catalan speaking area, there is indeed a lower frequency of se in proclitic environments in the Spanish speaking area of the Valencian Country and in Murcia. Out of the 14 enclaves that had enclitic se in that area, only 7 had also proclitic se, 4 of which vary between se and os, los or sos.
Map 5. Proclitic se in the ALPI As mentioned above, in Western Andalusia this phenomenon is related to the loss of the 2 nd plural stressed pronoun vosotros and its replacement by the polite plural pronoun ustedes, which in principle requires 6 th person agreement. In Western Andalusia, however, the generalisation of ustedes in the 5 th person did not trigger automatic 6 th person agreement in these forms, which seems to be more developed in the pronominal system than in the verbal system (Lara Bermejo 2012). The agreement in the 6 th person is most widespread in the reflexive pronoun, followed by the nonreflexive pronouns and the verbal affixes. In Murcia and Alacant (and in Catalan) there is no such thing as the replacement of vosotros by ustedes, which means that the ALPI examples are unequivocal 5 th person reflexives.
Coming to the 4 th person se, the ALPI does not contribute any questions with reflexive verbs in that person. Some COSER examples can be seen in (21). Table 3 shows the COSER data according to whether the verb is an infinitive or not. 39 The high p value (0.8179) of these data clearly shows that its distribution is most likely due to chance and that this is not a relevant factor for the 4 th person, despite its importance in the 5 th -although collecting more data may change the picture, since once again we find important differences between the two possibilities. 39 In this case infinitives do not coincide fully with enclitic forms, since there were two gerunds (both with the non-syncretic form nos). Testing for the difference between enclisis/proclisis does not offer significant results either. As for the causes of the spread of the syncretism to the 4 th person, internal pressure within the paradigm to regularise at least plural morphology seems to be an obvious call. However, there might be independent causes for this development (as the frequent 3 rd /6 th > 4 th developments in Northern Italian varieties already suggest) that should not be disregarded. 4 th person pronominal subjects or verbal morphology function as one of the impersonalising devices available in Spanish. As shown by the European Portuguese and the Eastern Spanish data in (14, 15) above, there is a connection between the reflexive marker, which marks impersonal sentences in many languages (all Romance languages, as a matter of fact), and the verbal morphology also used to mark impersonality (6 th , 4 th and 2 nd persons in Spanish), to the extent that both of them may appear as impersonal markers in one sentence.
But this connection also implies that these persons, when used as impersonal markers (hence intended as generic or indefinite), may trigger 3 rd person agreement, which is the default one for syntactically impersonal constructions. Examples in (22) illustrate quite nicely this bridging context between the 3 rd and the 4 th persons in Spanish. They were uttered by a speaker in Badajoz (Extremadura), in Central Western 40 In table 3 only data from villages with at least one example of the syncretism in the 4 th person are considered.
Pero se escondíamos como las ratas: Interestingly enough, there were two examples of 2 nd person se in the COSER interviews and both of them appeared in impersonal uses of this person (23). Since I did not recollect all the 2 nd person reflexive verbs, I cannot give quantitative data as to how frequent or rare they are compared to the standard 2 nd person te.
(23) COSER (Spanish) a. Uh, porque el tío, tú te casa-… (La Romaneta, AL 308) Oh because the guy you.NOM you.REFL marry 'Oh, because the guy, you got marri-…,' antes de casar-se tú, con tu mujer o con tu novia, before of marry.INF-3.REFL you with your woman or with your girlfriend 'before you got married, with your wife or your girlfriend,' tenía que casar-se él had.3SG that marry.INF-3.REFL he 'he had to get married' b. L' has de sacar de las hojas, el maíz, (Turís, VA 4324) it.ACC have.2SG of remove.INF of the leaves the corn 'You have to remove it from the leaves, the corn,' para llevar-se el maíz to take.INF-3.REFL the corn 'to get the corn' To conclude this subsection, I should add that one example of se in the 1 st person was also found. It is given in (24). Quite interestingly, it was found in one of two speakers that showed a regular use of se in all the plural persons (although he did not show such a regular use in the 1 st person, as can also be appreciated in our example). It is also worth noting that all the examples in (23) and in (24) appear in infinitive contexts, reinforcing the idea of the close link between the se syncretism and this verbal form. Before considering the influence of contact, put forward by Nebot Calpe (1984) and Blas Arroyo et al. (1992), I will take a look at the situation in Catalan in the next subsection.

The reflexive syncretism in Catalan
In Standard Catalan, the phonetic realization of object pronouns varies according to their position with respect to the verb and the phonetic environment, as shown in (25) However, the dialectal situation is (even) more complex. According to Perea (2012), reduced realizations of the 4 th and 5 th clitics nos/ens and vos/us as s (or z) or, with a support vowel, se, are frequently found in both positions. That is to say, in Catalan, phonetic reasons may cause the syncretism between the 3 rd person reflexive pronoun and the 4 th and 5 th pronoun in many varieties. That this phenomenon is caused by phonetic reasons is corroborated by the fact that the syncretism can be found in nonreflexive clitics: 41 41 Strangely, on the map SIN-79 of the Linguistic Atlas of Castile-La Mancha (ALECMan, García Mouton and Moreno Fernández 1988), which charts the nonreflexive 5 th person pronoun in the sentence Os dieron dinero (you.PL.DAT gave.3PL money 'They gave you guys money'), 5 points in Toledo offer se as a possible solution (these are La Calzada de Oropesa (TO 109), Talavera de la Reina (TO 301), Belvís de la Jara (TO 311), Ocaña (TO 410) and Yepes (TO 412)). These are quite surprising examples, since a phonetic reduction of sos, sus or /suh/ (the typical form that os takes in Toledo) to /se/ is difficult to explain. Unfortunately, the ALECMan only gives the pronoun form and not the complete sentences nor how the question was asked, so it is impossible to know the exact context of occurrence (may the sentence have been interpreted as reciprocal by the informants?). Even more strangely, none of these villages used se in the reflexive infinitive pro imperative context charted in the map SIN-34, Sentaros vosotros en las sillas (sit.down.INF-2PL.ACC you.PL in the chairs 'Sit down in the chairs'), although some of the neighbouring villages did. Martín Zorraquino (1979: 361-2) documents se venís (3.REFL come.2SG 'you come') in Ocaña (Toledo), but it seems that the syncretism only appears in such a sentence. All this together indicates that we need more data in this area. Since it is disconnected from our Eastern Spanish area, I think the Toledo examples cannot suggest a phonetic origin for the syncretism as found in the Valencian Country and Murcia.
Pero se escondíamos como las ratas: However, although these data clearly prove that phonetic processes have originated the syncretism in Catalan, there is also evidence that suggests that these reduced forms were later reanalysed and that 4 th and 5 th person se in Catalan is most commonly felt as reflexive.
First, although the form 's/s' can be easily described as a reduced form due to phonetic simplification processes (us > s, ens > s), the form se is harder to explain on phonetic grounds. Catalan accepts quite a number of final consonants and /-s/ is definitely not a problematic one. As Perea (2011: 442) points out, this casts some doubts on the epenthetic origin of the vowel; I concur. I believe, however, that the vowel in se could be explained if we do not consider it a reduced form but the expression of the reflexive 3 rd person pronoun.
Second, the non-reflexive examples of the syncretism are quite rare and a vast majority of the examples found in the literature are reflexive. As mentioned already, the phenomenon is condemned in several Valencian grammars. Salvador (1974), Guinot i Galan (1987) and Fabra (1913-14) document the phenomenon without specifying if the replacement of ens and vos by se is restricted to reflexive contexts, but all the examples they give are reflexive.
Sanchis Guarner (1950: 231), a native Valencian speaker, actually specifies that the syncretism is found in reflexive contexts: "A la capital i a gran part de la Regió Valenciana, s'usa vulgarment el pronom reflexiu se a indicar també l'acció reflexiva de 1.ª i 2.ª pers. plur. (nosatros es llavem; llavem-se; vosatros es llaveu, llaveu-se)". 43 Bonet (1984: 138) agrees that the syncretism is excluded when there is no subject-object coreference and that it can only appear in reflexive contexts. Perea (2012), as previously mentioned, considers the syncretism to be a phonetic phenomenon. She attests it (in the 5 th and in the 4 th persons) both in Alcover's material and in the COD questionnaire. 44 However, all the examples she gives are either reflexive verbs or combinations of two clitics (see (27)). Alcover & Moll (1968) consider that this 4 th and 5 th person se comes from an analogy with the reflexive pronoun, and they specifically claim that it has reflexive meaning in the 5 th person. 42 Maria Pilar Perea (p.c.) points out that the loss of the infinitive -r in the periphrastic past when it is followed by a 3 rd person clitic may have some influence in these reduced forms, since in such cases the clitic attaches directly to a vowel and there is no epenthetic vowel involved: va animar-los (go.3SG encourage.INF them.ACC 'S/he encouraged them') > va animà'ls. 43 "In the capital city and in a large part of the Valencian region, the reflexive pronoun se is vulgarly used to mark also the reflexive action in the 1 st and 2 nd person plural (we 3.REFL wash.1PL; wash.1PL-3.REFL; you.PL 3.REFL wash.2PL; wash.2PL-3.REFL)", my translation. 44 As said above, there were no instances of syncretism in the COD interviews. They, however, give a few non-reflexive examples in both persons, all of them in clitic clusters except for the two examples in (26). In all the works mentioning the phenomenon I have managed to review, those two are, actually, the only ones that do not present a reflexive meaning or are in a clitic cluster. In map 6 I have charted the geographical distribution of the reflexive examples compiled by Perea (2012) and Alcover & Moll (1968), the two sources with more data. 45 (27) a. Ç 'hu diré (Barberà de la Conca, TA, Perea 2012: 1053) (Catalan) 3.REFL it.ACC tell.FUT.1SG 'I will tell you guys' b. Ja s 'en daré (Vilabertran, GI, Alcover & Moll 1968: 149) Yet 3.REFL it.PART give.FUT.1SG 'I will eventually give you of that' Map 6. The syncretism in Catalan: Alcover's materials and the COD questionnaires Although with these highly unsystematic data I do not dare propose a diachronic evolutionary path, a few details are interesting to note. On the one hand, map 4 (which charts the ALPI data for imperative sentences) already presents some data on the syncretism in Catalan. Most interestingly, the distribution of s and se in map 4 is not the same. While the reduced s form is found in Eastern Catalan; Western Catalan and Valencian prefer se. There is no such thing as the infinitive pro imperative in Catalan, which means that these examples are perfectly transparent 5 th person verbal forms: begueu-se/beveu-se ('drink.2PL-3.REFL'), agenolleu-se ('kneel.down.2PL-3.REFL'). Although Perea's and Alcover's data do not draw such a sharp division in map 7, map 8 does show some difference between Eastern Catalan on the one hand and Western Catalan and Valencian on the other as regards the syncretism in the 4 th person. 45 Only the examples with a specific geographic adscription are charted in map 6. That is, examples that were said to be general in a certain dialect or were taken from popular songs are not represented in the map. last one to be affected by the syncretism? Can the syncretism develop from a person other than the 3 rd /6 th ? I do not have answers to any of these questions. Only a thorough typological research could shed some light into them, most likely in terms of frequency.