Forms of address in the south-western Sprachbund of the Iberian Peninsula One hundred years of evolution in western Andalusian Spanish

: South-western Peninsular Spanish (Andalusian) and European Portuguese rely on a single plural pronoun to address a group of people ( ustedes / vocês respectively). However, this can induce two different agreements in the verb, in the object pronouns and in the possessive: (i) second person plural (2PL) and (ii) third person plural (3PL). This chapter studies the linguistic spread of these agreement patterns during the last hundred years as well as the theoretical aspects that led to this variation in use, and it also confirms the Sprachbund theory that has been recently put forward regarding western Andalusian and southern European Portuguese, since both varieties share a series of linguistic behaviours and developments in phonetics, lexicon and morpho-syntax.


Introduction
The evolution of the plural systems of address throughout the Iberian Peninsula has undergone quite a similar process in all its Romance languages. Catalan, Galician and Spanish have two different pronouns: one for informality (vosaltres, vosoutros, vosotros respectively) and another one, for formality (vostès, vostedes, ustedes respectively) (Wheeler et al. 1999;Álvarez & Xove 2002;RAE-ASALE 2009: § 16.3). Standard European Portuguese, on the contrary, possesses a single pronoun to address a group of people both in an informal and a formal context (vocês). However, the northern part still maintains an older system based on the dichotomy of two pronouns: vós for informality and vocês for formality; vós is also resorted to in Church or military speech all throughout the country (Raposo et al. 2013). The levelling attested in standard European Portuguese extends throughout southern and central areas of Portugal and coincides with an analogous levelling in the Spanish of western Andalusia (in southern Spain). This region has eliminated the opposition vosotros -ustedes by favouring the use of ustedes both for formal and informal contexts. The fact that these two instances of levelling occur in the same area in which other linguistic features emerge as a consequence of the geographical proximity of Andalusian Spanish and southern European Portuguese has led scholars to put forward the existence of a Sprachbund, that is, a group of languages (in this case, Spanish and Portuguese) spoken within a specific area sharing several linguistic features as a consequence of their geographical proximity. Map 1 depicts the Sprachbund in the domain of plural forms of address. This proximity is also attested in other phenomena regarding lexicon and phonetics. According to Fernández-Ordóñez (2011), the word borrego 'lamb' to refer to the baby sheep is found throughout the west-southern area of the Spanish Peninsula and spreads to the centre-southern part of Portugal, to the detriment of the northern word cordeiro. Furthermore, the word chivo 'goat' is also attested uninterruptedly in western Peninsular Spanish and in the centre and the south of Portugal (chibo), as is the term mazorca -maçaroca 'corncob' which shares a similar geographical distribution. It is noteworthy that these words do not simply mirror lexical variation: they belong to agriculture and cattle breeding and, consequently, reflect a close cultural proximity. Cintra (1961Cintra ( , 1962 splits the centre-north and the centre-south of Portugal, following the origin of the differences between the words ordeñar 'to milk' or ubre 'udder', among others, and he notes that the southern area possesses a certain tendency to diffuse innovations although not systematically. As for phonetics, Cintra (1971) establishes the distinction centre-north and centre-south when he compares the realisation of the sibilant that corresponds to the spellings <s> and <ss>. According to the author, the northern pronunciation is apical-alveolar, whereas the southern one is pre-dorso-dental. It is precisely the southern pronunciation that coincides with the seseo pronunciation (the non-distinction between the phonemes [s] and [θ] by favouring the former) of western Andalusia.
In Section 2, I detail the evolution of Peninsular Spanish and European Portuguese regarding their plural forms of address system and the information available up to now about the situation in the abovementioned Sprachbund. Later in Section 3, I describe the methodology employed to elicit the study's data. Afterwards, I describe the results firstly from a synchronic perspective (4.1) and then from a diachronic view (4.2). In the following (4.3), I analyse the results from a theoretical point of view. I then discuss the historical evolution of forms of address in Portuguese and Spanish,(4.4), and in Section 5 I present the study's conclusions.

Spanish
Standard Peninsular Spanish possesses four pronouns of address: two for informal contexts (tú and vosotros, singular and plural respectively) and two for formal contexts (usted and ustedes, singular and plural respectively). Formality is expressed through the third person while informality chooses the second one (Table 1). However, the western part of Andalusia (in southern Spain) eliminated this distinction around the 18th century (Fernández 2012) and levelled any plural form of address in the pronoun ustedes. Nevertheless, despite the fact that the syntax requires the elements that refer to ustedes to agree in 3PL, it induces both 2PL and 3PL inflections (Alvar 1996;Cano 2004Cano , 2008Carrasco Santana 2002;De Jonge et al. 2012;Lapesa 1981;Menéndez Pidal 2005;Penny 2004 or RAE-ASALE 2009). This situation is identical to the one attested in the Spanish spoken in Latin America, with the difference that in Latin America ustedes systematically agrees in 3PL (Fontanella de Weinberg 1999). The disappearance of vosotros in Latin America is said to have occurred around the 19th century, when vosotros was relegated to rhetoric, the army and the church (Vázquez Laslop 2010;Bertolotti, this volume). Regarding western Andalusian, the available data (especially the studies by Alvar et al. 1961Alvar et al. -1965Carricaburo 1997;Lapesa 2000;Hummel et al. 2010) only suggest that the reflexive pronoun and the verbs in the past simple adopt the 3PL whereas the rest of the elements adopts the 2PL, except the possessive, which in principle is construed with the prepositional phrase de ustedes ( Table 2). Notice that the subject form and the form inside a prepositional phrase are identical in Spanish; this is why I use stressed pronoun to refer to both elements. Thanks to the data of the Linguistic atlas of the Iberian Peninsula (ALPI), collected between 1931 and 1954, Lara (2012) has shown that the levelling in ustedes ran throughout the Andalusian provinces of Huelva, Seville, Cádiz, Málaga (except the most eastern part) and Córdoba (except the most northern part) and that it could induce either the 3PL or the 2PL, based on the hierarchy rep resented in (i).
(i) Stressed pronoun > reflexive > accusative > embedded verb The continuum set out in (i) shows the extension of the 3PL throughout the different elements that refer to ustedes: if the 3PL emerges in the accusative (los), it also arises in the reflexive (se) and the stressed pronoun (ustedes), but not yet in the embedded verb. In (1 a -d) the evolution of the agreement is shown based on the ALPI data.
( These examples show that the 3PL gradually spreads throughout the syntax. While (1b) only induces the 3PL in the stressed pronoun and the reflexive pronoun, (1d) already adopts the 3PL in the accusative pronoun and the verb of the embedded sentence, as (i) illustrates.
The methodology employed in the ALPI, based on a questionnaire with pre-established sentences and words that the informants had to repeat in their vernacular varieties, did not provide information about the agreement of all the syntactic elements governed by ustedes, since there were no questions eliciting the dative, the possessive or other verb tenses. Likewise, this atlas has limited quantitative data, for it only collected one response per sentence and informant (who was always male, non-mobile, rural, with a limited educational background and over fifty years old) (Sanchís Guarner 1962).

Portuguese
European Portuguese exhibits an analogous case to the Andalusian one. Before the 18th century, the whole area possessed the opposition of two different pronouns, vós and vocês, that expressed informality and formality, respectively (Table 3).
However, nowadays, the standard plural address pattern is levelled in vocês although it relies on a great many noun phrases that express kinship or social and professional differentiation in order to be more polite (Braun 1988;Carreira 2003). Amongst all the nominal formulas, the most common and least marked construction corresponds to os senhores/as senhoras 'sirs', 'madams' and, like vocês, it must agree in 3PL (Table 4). Nonetheless, the standard norm requires that certain elements take 3PL inflections and other elements 2PL inflections. So, the reflexive and any verb tense receive 3PL whereas object pronouns and possessives agree in 2PL (Table 5) (Cunha & Cintra 1992;Brito et al. 2006;Raposo et al. 2013). Again, Lara (2012) identified this phenomenon in the ALPI and realised that the standard system was only attested in the southern half of Portugal (Faro, Setúbal, Beja, Évora, Portalegre, Santarém, Lisbon, Coimbra and Leiria) while the northern half still maintained the previous system, represented in Table 3. This means that the paradigm of Table 3 was still valid in the northern region at least until the first half of the 20th century. With regard to the agreement of the levelling in vocês, as the Andalusian, the 3PL spread hierarchically based on (ii).
(ii) Stressed pronoun/reflexive/verb > accusative In other words, the hierarchy shows that if the 3PL is attested in the accusative, it is necessarily attested in the verb, the reflexive and the stressed pronoun (2a -b). As illustrated in examples (1a -d), example (2b) shows that the 3PL already emerged in the stressed pronoun, the reflexive pronoun and the accusative pronoun, while in (2a) it had not reached the accusative pronoun.
The ALPI data do not provide information about the situation of the dative and the possessive, as has been explained above. Moreover, the methodology of this atlas does not allow for the quantitative analysis of a given phenomenon, and the modus operandi could have tainted the informants' responses. Thus, the data collected in the ALPI have to be evaluated taking into account its methodological limitations.

Corpus and methodology
With the aim of investigating the social and linguistic reality of both levellings and compensating the shortcomings of other methods, I carried out fieldwork throughout western Andalusia and the centre-southern part of Portugal in 2012 and 2013. Such fieldwork consisted of a series of interviews in which the informants had to watch several scenes (without the audio track) of two famous sitcoms that usually show a character addressing a group of people. After watching them, the informants had to become the character and dub the scene. The scenes were chosen in a way to ensure that many tokens of 2PL/3PL would have to be produced. The informants were recorded while they carried out the activity and, later, the audio recordings were transcribed (Lara, 2016). Any occurrence that included a second person plural was classified on the basis of its syntax (subject, reflexive, direct object, indirect object, possessive, verb) and on the informants' extra-linguistic features (gender, age, educational background, origin, number of inhabitants of the place of origin). Moreover, within the category of verb, a distinction was made based on tense, mood, modality or the type of sentence (main or embedded). Lastly, the selection of the scenes was chosen, by taking into account different kinds of interlocutors (friends, family, strangers, elderly people, etc.) with the aim of finding possible pragmatic differences expressed in the agreement. Altogether, over 250 informants were surveyed and approximately 4,900 occurrences were obtained.
Two statistical tests were applied to the results: Pearson's chi squared and a logistic regression. The former gives the real significance of an independent variable (gender, age, etc.) and the latter orders the degree of affectedness of every significant variable. In the studied phenomena, the chi squared results highlighted the importance of the factors age, educational background and size of the population of the municipality. Hence, both in the Spanish and Portuguese areas, the tendency toward adopting the standard pattern (the distinction between vosotros and ustedes in Spanish, and the levelling in vocês in Portuguese) is usually related to middle-aged informants with a higher education background who, at the same time, live in urban environments (Lara 2015). However, the goal of this chapter is to account for the linguistic extension of the agreement throughout all the elements that refer to ustedes and vocês.

Results
Below, I show the results of my fieldwork. In the first place, in 4.1 I discuss the data from a synchronic perspective; later in 4.2, I deal with both phenomena from a diachronic view; then, I analyse them theoretically in 4.3; finally, I compare the common evolution of Spanish and Portuguese over time in 4.4.

Subject and verb agreement
Independently of the social situations (that will not be analysed here), the levellings in Spanish and Portuguese (ustedes and vocês) are characterised by their exhibiting an identical syntactic behaviour. Let us begin with Spanish.
The use of ustedes has produced three geographical areas based on the frequency of the levelling and its grammatical behaviour (Map 2).
Map 2 shows that the centre-northern area of Córdoba and the most eastern area of Málaga are characterised by a low or null use of ustedes as the single plural address term. In other words, the areas with < 33%, either follow the standard pattern, or have a low proportion of speakers that choose the vernacular response. On the opposite side, we find the territory comprised of southern Seville and all of Cádiz, with > 66% of use of the levelling in ustedes. In these areas, the virtual totality of speakers selects a single pronoun and only a few of them tend to employ the normative pattern. In an intermediate position, there is the area with 33%-66% of use of the levelling. That is to say, this last area possesses the same proportion of speakers that choose the standard paradigm as the number of informants that maintain the vernacular model.
From the grammatical point of view, the three areas present different behaviours regarding the agreement between ustedes and the verb. While the area with < 33% is characterised by a strong disagreement (virtually all verbs adopt the 2PL) (3 -4), the area with > 66% has widely made regular the 3PL in the verb (5 -6). Again, the area with 33%-66% shows an intermediate behaviour, where the 3PL is more numerous than in the area with < 33%, but it is not yet consistent (7 -9).
(3) Ustedes, no tenéis nómina. You-3pl, no have-2pl.pres.ind salary 'You do not have any salary' (4) Ustedes, habéis desorganizado mi casa. You-3pl, have-2pl-pres.ind mess up-pcp my house 'You have messed up my house' In examples (3 -4), there are several occurrences in which the informants belonging to the area with < 33% have expressed a pause between the pronoun ustedes and the verb, which is inflected in 2PL; this is why the comma is written.
Within the area 33%-66%, we find a higher proportion of agreement in 3PL although it is still low. Furthermore, we attest the concatenation of the 2PL and 3PL inflections materialised in unstressed pronouns and verbal morphology (5 -6).

Clitics and possessives agreement
Let us begin with the analysis of the unstressed pronouns or clitics ( Figure 1). As can be seen in Figure 1, the clitic likelier to adopt the 3PL is the reflexive, followed by the accusative and, then, the dative. Based on these data, three phases can be identified in (12 -14). In a first stage, the reflexive adopts the 3PL (352 examples out of 490), even though object pronouns are still inflected in 2PL. In a later stage, the accusative takes the 3PL (74 out of 190 cases) whereas the dative prefers to keep the vosotros morphology. In an ulterior phase, the dative starts agreeing in 3PL (59 out of 349 cases), and this is why all unstressed pronouns end up receiving the agreement induced by ustedes. Let us analyse the behaviour of Portuguese to this respect ( Figure 2). In Figure 2, which encompasses the percentage of use of either the 2PL or the 3PL in clitics, we see that, again, the reflexive is widely inflected in 3PL (62 out of 65 instances), followed by the accusative (whose alternative in 2PL increases up to 35%: 42 out of 62 occurrences have adopted the 3PL) and, the dative, which presents a wide percentage of tokens in 2PL (15 - The agreement patterns throughout the clitics show that, once ustedes and vocês become full subjects (I will refer to this later), they start spreading their syntactic agreement (3PL) throughout all the elements that refer to them. This process is gradual and progressive.

Reflexive
Let us end by looking at the situation of the possessive. Figure 3 shows that the prepositional phrase de ustedes is hardly attested and it is outnumbered by the choice in 2PL, vuestro, or the standard third person form, su, with an occurrence of 20%.
It is precisely in the area where the 3PL has extended until the dative (area with >66%) where the possessive begins to agree in 3PL; hence, it is the last element in the chain to adapt to the new agreement (

Diachrony
If we synthesise the result presented above, we can observe a diachronic path both in Portuguese and Andalusian Spanish. Let us begin with the latter.
Map 2 shows the three different areas that have resulted from the agreement patterns: < 33%, 33%-66% and > 66%. Based on this classification, the area characterised by > 66% presents systematic agreement in 3PL with ustedes. In principle, this leads us to put forward that this zone was the one in which the levelling originated. As a matter of fact, if we compare the current results to the ones attested almost one century ago in the ALPI (Map 3), and already described and analysed in Lara (2012), we observe that the same area of > 66% presented the most evolved stage nearly one hundred years ago.
However, the 3PL in the verb had not completely generalised and it started emerging every now and then. Nowadays, it exhibits complete agreement in 3PL. Likewise, the area classified as 33%-66%, which presents an intermediate percentage of 3PL agreement with ustedes coincides with the area in which the 3PL only arose in reflexives and stressed pronouns. Lastly, the zone < 33%, which nowadays presents a low percentage of agreement in 3PL exhibited no agreement in 3PL with any syntactic element almost one hundred years ago. In fact, it only construed the stressed pronoun in 3PL. Therefore, the grammatical situation has changed during at least the last 75 years. Again, it is the area >66% the most innovative one, by spreading the 3PL throughout more syntactic elements than the rest of geographical areas. Hence, it is in this zone where the levelling commenced and was later diffused toward its outlying areas until reaching almost all of western Andalusia. In sum, the area > 66% only agreed in 3PL the stressed pronoun and the reflexive in the first half of the 20th century and sometimes the verb, but now it induces 3PL to all verbs, apart from reflexives and stressed pronouns. The area 33%-66% could only agree in 3PL reflexives and stressed pronouns previously, but now it also starts doing so in verbs, though not systematically. Finally, the area < 33% only inflected in 3PL the stressed pronoun in the last century, but now it also induces it in reflexives and in an extremely low proportion in verbs.
In the case of European Portuguese, the data analysed in Lara (2012) regarding the extension of the levelling as well as the grammatical agreement shows that at least two stages can be observed (Map 4).
In the first stage, the 3PL agreed with the subject, the reflexive and the verb, while the second one had extended the 3PL also onto the accusative pronoun. Map 4 can be compared to the current linguistic situation illustrated in Map 5.
Map 5 shows that nowadays European Portuguese presents four stages. In phase 1, the pronoun vocês has displaced vós as an informal address and has generalised the 3PL in the verb and the reflexive (23 -24). The second stage implies the adoption of the 3PL in the accusative, despite the fact that the standard does not induce 3PL inflections in object clitics. This phase is attested in rural environments of south-eastern Alentejo and all Algarve. Immediately afterwards, the dative adopts the 3PL though its extension is even less than that of the accusative.
Lastly, the possessive starts being construed in 3PL although the geographical space of this phase is only documented in the most south-eastern part of Portugal, on the border with the Andalusian phenomenon of the levelling to ustedes. The single area that has remained isolated to the successive innovative waves of the vocês phenomenon has been the city of Lisbon, whose status of urban, cultural, social and political centre has caused it to stay faithful to the standard pattern that does not allow the 3PL to generalise further than in the reflexive and the verb.
As has been remarked for the Andalusian levelling, the most advanced areas in terms of the extension of the 3PL in the vocês phenomenon in the first half of the 20th century are now again the most innovative one, since the 3PL has even spread throughout all the syntactic elements that refer to vocês. We cannot know what agreement datives or possessives adopted in the last century, because there were no occurrences about these elements in the available corpora, but virtually all the area characterised by inducing 3PL in the accusative has remained in the same stage or has furthered by spreading the 3PL to the dative and, to a lesser extent, to the possessive. The areas that presented fewer cases of 3PL extension have either maintained the stage attested last century or adopted the 3PL in the accusative too. Lastly, there are new areas affected by the levelling; these only agree in 3PL the subject, the reflexive and the verb, as the first stage attested in the data from one hundred years ago. The only region that does not show any kind of change in terms of the extension of the 3PL is the city of Lisbon. As in the data analysed in Lara (2012), it only inflects the subject, the reflexive and the verb in 3PL, but prefers the 2PL for object pronouns and possessives. Though the singularity of Lisbon has been repeatedly referred to,1 such as in the case of the phonetic change of /e/ to /a/ before palatal consonant (Teyssier 1982), or the uvular pronunciation of /r/ (Barbosa 1983) or even the maintenance of the diphthong /ei/ despite the southern tendency of converting it into the monophthong /e/ (Cintra 1983), it is a priori more convenient to suggest that the conservative behaviour of Lisbon responds to its demography. As has been demonstrated by Wolfram & Schilling -Estes (2003) or Chambers & Trudgill (1980), urban centres gather the political, economic and social power of a given territory and this power regularly imposes the standard variety (Joseph 1987). The non-adoption of the 3PL in object pronouns and possessives in the levelling of vocês in Lisbon would only be an attitude inclined to the standard pattern (it must be remembered that the 2PL is the norm in these syntactic contexts) at the expense of vernacular innovations from more rural areas. Lisbon simply follows the standard model imposed by itself (Lara 2017).

Theoretical analysis
I have pointed out that Andalusian induced person mismatches between the subject and the verb. These person mismatches are not anomalous in Spanish nor are they in many other languages, as has been researched by Ackema & Neeleman (2013), Choi (2013) and Höhn (2016). As a matter of fact, Ordóñez & Treviño (1999) or Fábregas (2008) (for Spanish) and Papangeli (2000) (for Greek) have investigated the employment of a verbal agreement different to that of the subject, such as in (25)  Though a vast amount of literature has been devoted to the study of noun phrases (mainly in the plural), there have been a few studies on the lack of agreement between personal pronouns and verb tenses that depend on them. In fact, one of the most remarkable phenomena in Spanish regarding pronoun disagreement is voseo, whose agreement swings between that of vos and that of tú. Abadía de Quant (1992), Bertolotti & Coll (2003) and Fontanella de Weinberg (1979) argue that the use of vos starts in the stressed pronoun and, later, it induces its agreement gradually: firstly, in imperatives; secondly, in the present of the indicative; and, eventually, in the present of the subjunctive and the past simple. Currently, the rest of verbal tenses, as well as clitics and possessives, are built with tú and not vos morphology.
Bosque & Gutiérrez Rexach (2009) explain that person disagreement is the result of topicalisation, in which the element that is located in the left periphery is not really the subject that must agree with the verb, but the topic. Precisely, the characteristics of the latter are its position outside the clause, its autonomy and the obligation to be recovered anaphorically within the sentence; additionally, this anaphor is not forced to receive obligatorily the same syntactic features as those of the topic.
Topic constructions have led to deep linguistic changes in Spanish and in many other languages. Elvira (1993Elvira ( , 1996 and Fernández Ordóñez (2009) argue that the current word order in Spanish (SVO) is due to the frequency in the Middle Ages of placing the subject in a topical position; according to Adams (1987), old French behaved as a V2 language, but the tendency to place the subject in a topical position prompted the current order and the obligation of making it explicit. Italian exhibits nowadays three third person pronouns (lui, lei, loro) which, at one time, were oblique. Their frequent placing in a topical position triggered the displacement of the older normative subject pronouns (egli, essa, essi) and the imposition of the oblique pronoun as new subject third person pronouns (Rohlfs 1968;Ernst et al. 2008).
One of the particularities of Spanish is its pro-drop character; this is why the expression of the subject emerges in contrastive and disambiguating readings. Therefore, the above mentioned examples (3 -6) lack a syntactically 2PL pronoun (vosotros) because it is omitted. To this respect, RAE-ASALE (2009) affirms that the western part of Andalusia presents records in which two 2PL person pronouns are concatenated, as reproduced in (27). (27) Ustedes vosotros sois hermanos. You-3pl. you-2pl. be-2pl.pres.ind. siblings 'You you are siblings' Example (27) effectively shows that the expression of both pronouns within the same sentence exists and that vosotros is still present, at least, in a certain area of western Andalusia. Although my corpus does not include concurrent occurrences of both stressed pronouns, it has recorded the concatenation of these pronouns in other grammatical contexts (28 -30). As can be seen in (28 -30), the overt use of the two pronouns in the same sentence is vernacularly possible (32 examples were collected). In the first case, the informants produce the reflexive in 3PL and, then, in 2PL. Even the verbal inflection shows in its hybrid form the emergence of both agreements: -ro in 3PL and -is in 2PL (40 examples of this case). It is precisely in the area with 33%-66% where these tokens have been recorded: the area in which the use of ustedes is intermediate and the syntactic agreement in 3PL is higher than in the area with < 33%. However, it is not as consistent as it is in the area with > 66% (in fact, from 270 examples of ustedes + verb, 147 adopt 3PL, and 123, 2PL). These instances point out that the concatenation of the stressed pronouns (ustedes + vosotros) exists, but it is uncommon due to the pro-drop parameter of Spanish. The no need to express the subject (vosotros) leaves the topic (ustedes) as the single explicit pronoun. Ustedes is followed by a verb in 2PL, which really agrees with vosotros and not with ustedes.
The existence of two different forms that refer to the same entity within the same sentence has also been documented in Italian. Again, in the development of the pronouns lui, lei and loro at the expense of egli, essa, essi, it has been observed that, in a certain period of this development, both pronouns coexisted within the same phrase (31 -33) (Ernst et al. 2008 As can be seen, the use of lui begins in topical constructions and it is recovered by an anaphor (egli). With time, the topic starts being reinterpreted as the subject although it does not possess all the features that a topic has and it coexists in the same sentence with the old pronoun (that loses phonic weight) which the new one wants to oust. Lastly, lui ends up being imposed and displacing completely the old use, which disappears. This last phase is exactly what one can find in Andalusian, within the area with > 66%, where the use of ustedes is hegemonic and so is the agreement in 3PL, as can be observed in the examples (34 -37) (out of 151 cases, 134 adopted the 3PL in the verb and the rest, the 2PL).
(34) Ustedes, ¿no estarían cotilleando? You-3pl, no be-3pl.cond gossip-ger 'Wouldn't you be gossiping?' The occurrences show an automatic agreement in 3PL due to the subject status that ustedes possesses in this area, since it has stopped being a topic. Even in sentences where ustedes is made explicit as a topic (34 and 36), the 3PL arises because the subject is the same as the topic. It is in this area where vosotros does not exist anymore and ustedes has completely displaced the old 2PL informal pronoun. In Table 6, the development of ustedes and the agreement with the verb is synthesised. This conversion from topic into subject is widely documented cross-linguistically. Hopper & Traugott (2003) argue that subjects are basically reanalysed topics and the latter tend to become subjects because they are usually placed in positions prototypically held by them (the left periphery). Givón (1975Givón ( , 1990) is one of the best exponents of the change from topic into subject. According to the author, the development of a topic into a subject undergoes three different stages. In the first one (38), there is a topical construction, where the topic is inserted in the left periphery, followed by a comma that marks the prosodic pause with the rest of the sentence. In addition, the sentence contains an anaphor that refers to such topic and which really behaves as the true subject of the sentences.
The frequency of the construction (38) makes speakers reanalyse the element positioned on the left as the subject, since it occupies its prototypical position. Nevertheless, before being completely reinterpreted as the subject, the topic undergoes an intermediate phase in which it does not possess all the elements of a topic (the pause disappears and it is inserted inside the sentence), but it does not receive either the features of a subject (it still needs to be referred to by an anaphor), as shown in example (39).
(39) The man he came.
The last stage in this development is accomplished when the speakers completely reinterpret the old topic as the subject of the sentence, prompting the disappearance of the anaphor (40).
If this process is applied to the levelling of ustedes, the stage exemplified in (38) corresponds to the one documented in the area with < 33%, where ustedes acts as the topic and it is recovered by an anaphor (vosotros) which is silent (not produced phonetically) due to the pro-drop parameter in Spanish. Next, the stage illustrated in (39) is attested in the area with 33%-66%, since we observe the emergence of occurrences where the two forms are expressed (se os, intentarois): one of them refers to the topic and the other to the true subject. In this phase, the topic coexists with the still-subject, but it does not enjoy yet the status of the latter, as it needs to be anchored by an anaphor. The cases in which the form that refers to the true subject does not emerge are simply silent. Lastly, the stage reproduced in (40) is attested in the area with > 66%, where ustedes is definitely a subject and, therefore, it induces the verb to the 3PL. The apparent person mismatches, as has been explained, are due to nonexplicit elements. The existence of these phonetically covert elements has also been widely researched. According to Kayne (2003Kayne ( , 2005Kayne ( , 2007, French and Italian have constructions that contravene the norm or that show an apparent disagreement. If we take into account example (41) (43) and (44), third person stressed pronouns allow the non explicitness of the subject, unlike the rest of persons, where it is obligatory (45 and 46). The explanation for this, according to Kayne, is the presence of an element that is simply not expressed. As has been mentioned above, I have not found disagreements between vocês and the verb in European Portuguese, so I infer that vocês behaves as a true subject and not a topic.
Once the term of address has become the subject and does not work any longer as a topic, the 3PL starts spreading throughout the rest of syntactic elements that refer to ustedes or vocês. In the case of Andalusian Spanish, this extension follows a hierarchy that can be synthesised in (iii).
(iii) Subject > reflexive > verb > accusative > dative > possessive Based on this continuum, if the 3PL emerges in the dative, it also appears in the accusative, the verb, the reflexive and the subject. The spread of the 3PL runs the continuum from left to right through implicational phases.
For European Portuguese, the agreement also follows the same continuum, though in a slightly different way (iv).

(iv) Subject/reflexive/verb > accusative > dative > possessive
The continuum indicates that if the 3PL emerges in the accusative, it also appears in the elements on the left, but not yet on the right.
From a grammatical point of view, the main question now is why the extension of the 3PL follows this pattern and not a different one. According to Corbett (2006), the agreement depends on various factors: in the first place, on the opposition between controller and target. While the former is the element that induces the agreement, the latter is the element that receives it. So, if a controller induces two different agreements, the target will acquire one of the two based on two parameters. One of them refers to the position that both the controller and the target hold within the sentence. This means that the further away the controller and the target are from each other, the more independence the target has to select the agreement. Let us analyse the following example (47), taken from Corbett (2006).
(47) The committee has decided to pass the law but they have been discussing the whole night.
In (47), the controller (committee) induces singular agreement as the verb shows, despite the fact that it is semantically plural. Nonetheless, its reference is again recovered in the adversative sentence through a plural pronoun and a verb that agrees in the plural with this pronoun. The employment of they is the image of a higher independence, because the target is found far away from the controller; in fact, it is found in a different sentence. Thus, in this instance, its preference is the semantic agreement (plural) and not the syntactic choice (singular). The other element on which the adoption of the agreement also depends refers to the status that the target may have in a hierarchy, represented in (v).
(v) Personal pronoun > relative pronoun > predicate > attributive Based on the continuum, the further we move rightwards in the hierarchy, the likelier it will be for the target to choose syntactic agreement; while the further we move leftwards, the likelier it will be for the semantic agreement to emerge. Let us analyse example (48).
(48) Sus excelentísimas majestades están Poss.3pl. excellent.fem.pl. majesties be.3pl.pres.ind. muy satisfechos con la noticia. very satisfied.masc.pl. with the news 'Their majesties are very satisfied with the news' In (48) we observe that, although majestades is feminine, it refers to an inclusive masculine. The two adjectives that agree with majestades adopt both the masculine and the feminine, but its choice depends on the stage within the continuum in (v). Whereas excelentísimas behaves as the attributive, satisfechos belongs to the environment of the predicate and, therefore, based on the hierarchy, it is closer to the semantic agreement than the attributive. A proof of this alternating quality is found in the mass neuter.2 Fernández- Ordóñez (2006Ordóñez ( , 2007 explains that the part of the Iberian Peninsula where the mass neuter is reported presents an agreement extension that coincides with the hierarchy of Corbett, since the syntactic agreement emerges in the attributive and spreads gradually over to the pronoun. This continuum runs from the centre-east of Asturias (where the syntactic agreement is more rooted) up to the centre-west part of Castile, where the semantic agreement is higher.
Hence, in the case of ustedes and vocês, the verb and the reflexive adopt first the syntactic agreement, because their proximity and dependence with respect to the controller is quite close; only when the inductor pronoun behaves as a topic, does the agreement tend to be semantic, for the controller is no longer found in the same sentence. Object pronouns rely on a higher autonomy owing to the fact that they do not possess any controller that previously induces them an agreement, but they are the first reference to the entity within the sentence (except in topicalisation or double-clitic constructions). So, they take more time to access the 3PL, followed by the possessive, which also has enough autonomy (even deeper than for objects) so as to adopt an agreement not induced by any controller.
Despite the precious study by Corbett, his research is circumscribed to the syntactic and semantic difference mainly in gender and number, so the conflict of the grammatical person (2PL versus 3PL) is not resolved in his investigation. The work by Wechsler & Zlatic (2000, 2003 deals more in depth with the agreement conflicts and for this they distinguish two terms: index and concord. The index agreement is that which is established between subject and predicate and it is subjected to the features of person, number, gender (and sometimes, case); concord works with the agreement in the environment of a noun phrase and it usually responds to the features of case, gender and number.
However, as Wechsler & Hahm (2011)  Though all vous cases induce plural, part of the index behaviour chooses the singular as the only way to disambiguate the referent. In spite of the fact that ustedes and vocês are forms of address, the agreement they induce do seem to fulfil the principles of Wechsler & Zlatic (2000, 2003. Therefore, according to these authors, the reflexive and the verb must accept the features of agreement that the subject sends them, this is why ustedes and vocês, when reanalysed as subjects, have to induce the 3PL in both elements, as the index agreement establishes and this forces the subject to agree with the predicate. Otherwise, object clitics behave more independently. They are not usually governed by an entity that is previously expressed such as in double-clitic constructions. They nearly always present for the first time the reference in the sentence. As objects are not obliged to agree with anything else within the sentence, they are not obliged to adopt index agreement. The possessive however is usually built in the noun phrase, so it is the element with the least pressure to adopt the person marker (since its agreement is concord and not index).
Irrespective of the type of agreement adopted by the syntactic elements that refer to ustedes and vocês (syntactic -semantic, index -concord), both are ruled by the same pattern in the grammatical cases (Table 7).  Table 7 shows that the first element to agree is the subject (and the prepositional phrase analogously, as it is the same form for both syntactic contexts) and those closely-related to the subject: reflexive and verb (in the case of Spanish, we observe that the reflexive agrees firstly and, later, with the verb; Portuguese has not offered cases of disagreements between subject, verb and reflexive). This means that the syntactic contexts typically held by the nominative or that refer to it are the first ones to adopt the 3PL. They are followed by the accusative, the dative and the genitive, whose marker is usually represented by the possessive. This order does not seem random, as can be observed in (vi).
(vi) Nominative > accusative > dative > ablative > genitive According to Pinkster (1985Pinkster ( , 1990, the usual order in Latin in case inflections followed a hierarchy that corresponded to the one reproduced in (vi). It is exactly the same by which the extension of the agreement is ruled (remember that the subject and the prepositional phrase are homophonous, so both adopt the 3PL at the same time; consequently, the ablative stage must be disregarded in this case). Moreover, such case hierarchy coincides with the hierarchy of syntactic contexts to which many languages obey. Let us analyse (vii).
(vii) Subject > direct object > indirect object > oblique The continuum reproduced in (vii) shows, according to Blake (2004), that most languages follow a non-marked order based on this hierarchy. Even the possibility to produce a passive sentence follows this pattern. While Spanish and Portuguese only have the possibility to passivise the direct object (53a -b), English does so with indirect objects too and, as a consequence, it implies that it produces a direct object passive (54a -b).
(53) a. El dinero te fue dado. The money dat.2pl be.3sg.pst. give.pcp 'The money was given to you' b. *Tú fuiste dado el dinero. You be.2sg.pst. give.pcp. the money *'You were given the money' (54) a. The money was given to you.
b. You were given the money.
This distinction on the basis of the case marker can also be attested in other cross-linguistic phenomena. Keenan & Comrie (1977) remark that the ability to relativise an element depends on its case function. So, there are languages capable of relativising only the subject, while others can do so with the subject and the direct object; others can do so with the subject, the direct object and the indirect object. However, no language is able to relativise the indirect object but not the direct object and the subject. Consequently, based on (vii), every language able to relativise an oblique object can do so with the elements on the left. But even the change of the argument structure responds to this criterion. Comrie (1976Comrie ( , 1989 argues that causativisation in Turkish follows this hierarchy in adding valency. For instance, if one more valency is added to an intransitive sentence, the former subject becomes the direct object, and the new valency becomes the subject; if one more is added, the direct object turns into the indirect object, the former subject becomes the direct object and the new valency is the new subject and so on (55 -56). As a result, if the analysis carried out by Blake (2004) or Keenan & Comrie (1977) in their investigations are applied to the extension of the 3PL, we notice that it is firstly attested in the subjects or the elements that depend on it or refer to it (reflexive and verb); later, it moves onto the direct object (a function prototypically held by the accusative), then onto the indirect object (the common case of datives) and in the last place, onto the possessive (usually the genitive). Even though Blake explains that the oblique case appears before the genitive in the hierarchy, ustedes and vocês have the same form for the subject and the oblique case, which is why once the 3PL is attested in the subject, the oblique case automatically adopts the 3PL.

Sprachbund and the Americas
The data analysed in the previous sections account for the fact that the development of the Portuguese and Spanish levellings have undergone an identical pathway regarding their grammatical evolution as well as they have definitely established themselves throughout the south-western part of the Iberian Peninsula. However, this is not the first time they end up developing a common strategy with regard to politeness.
The emergence of various noun phrases in the late Middle Ages occurred contemporaneously and both languages evolved into the same system: vuestra merced/vossa mercê became the least marked polite strategy, vuestra majestad/ vossa majestade was employed to address the monarch, vuestra excelencia/ vossa excelência was reserved for gentry and clergy (Menon 2006;Menéndez Pidal 2005). In the case of the former (vuestra merced/vossa mercê), the two of them underwent the same grammaticalisation process. In fact, Lara (2012) shows occurrences of old stages in the grammaticalisation of vossa mercê to você, since the ALPI data provide evidences of vossemecê or vomecê. These two alternatives coincide with intermediate phases of the evolution from vuestra merced to usted in Spanish, such as in vuested or vuesasted (Menon 2006;Pla Cárceles 1923). From a grammatical point of view, the path has repeated itself on several occasions. I have put forward the topical character of ustedes one hundred years ago and its tendency toward becoming a subject. According to Hammermüller (2010), the imposition of vuestra merced over vós followed the same process. It was firstly expressed as a vocative and, therefore, out of the sentence. The verb was inflected in 2PL because its agreement was induced by the pronoun vós. As Spanish is pro-drop, the production of vós had to be low in comparison to the likelihood of expression of the vocative or topic (vuestra merced). Once the construction became more and more frequent, vuestra merced was reinterpreted as the subject and, as a consequence, the verb started to agree in 3SG. In fact, according to Menon (2006), Menéndez Pidal (2005 and Cano (2008), the first uses of vuestra merced/vossa mercê alternated with 2PL and 3SG agreements.
Nevertheless, the most remarkable feature to be analysed concerning the forms of address system in both languages takes place in the abovementioned Sprachbund. The levellings are restricted to the south-western part of the Iberian Peninsula (although in the case of Portugal it starts spreading northwards because it represents the standard). Even though the best known Sprachbünde are the ones attested in the Balkans and the South Asian area of linguistic convergence (and even a major one in Europe, Haspelmath 2001), the Iberian region in which these levellings are witnessed is characterised by sharing a common development irrespective of the historical period. For instance, the pre-Roman language Tartessian, spoken throughout the Iberian Peninsula before the conquest by the Roman Empire extended virtually throughout the most south-western part of the Sprachbund, that is, the area in which the grammatical agreement is complete. However, the period of time in which such a geographical zone shared more exchange was the time after the discovery of America and the division of the world between the crowns of Spain and Portugal.
During the following centuries until the independence of the Spanish and Portuguese colonies in America, the south-western region of the Iberian Peninsula was the area from which the diverse expeditions departed and from which trade was carried out with the American continent. And, as in European Portuguese and western Andalusia, we see the same levellings in the American varieties of these two languages. The influence between both sides of the Atlantic to this respect has not been studied in depth. I do not wish to discuss the Andalusian influence in the Latin American varieties, since it represents a controversial topic, but somehow they have conditioned each other even in post-colonial era. I have already commented that the elimination of vosotros in Andalusia began in the 1700's, when Spain was still an empire. By the 19th century, ustedes was generalised (Fernández 2012; García Godoy 2012). Likewise, vosotros is said to have completely disappeared from the American varieties in the late 19th century (Bertolotti 2015 and this volume). Portuguese exhibits the same pattern. Just like for Spanish, I will not discuss the influence of southern European Portuguese on the Brazilian variety, but the levelling in vocês, which also started in the 18th century (Cintra 1972;Faraco 1996), has spread throughout Brazil and has represented the only informal pronoun in the plural for more than two centuries (Menon 2006). Nonetheless, in the singular, it is possible to attest analogous developments on both sides of the Atlantic. Currently, the formal pronoun usted in Spanish is perceived as informal and even affectionate in certain American regions, especially in Central America and the Caribbean (Hummel et al. 2010). Its use is shared with traditional informal pronouns tú and vos, since the three of them can be resorted to in informal and intimacy contexts. García Godoy (2012) and Calderón Campos (2010) state that usted as informal or affectionate could be used in Andalusian Spanish during the late colonial period, that is, at the time in which Latin American started adopting it for informality too. Again, the levelling in singular usted at the expense of tú originates in south-western Peninsular Spanish and is later exported to the American varieties. This also seems to be the case in the plural.
The situation of Portuguese is better documented. In the plural, Brazilian Portuguese does not possess vós for informality as 2PL and neither does southern European Portuguese. Likewise, vocês is informal and os senhores is the most common formal strategy to address a group. In the singular, você has ousted tu in almost all of Brazil (Lopes & Cavalcante 2011) and in southern European Portuguese, você can be attested as an informal pronoun too (Lara & Guilherme 2015). Furthermore, similarities also arise in the 1PL: the Brazilian spread of a gente 'the people' instead of the traditional nós 'we' for the 1PL is spatially attested in southern Portugal, but not in the north (Lara & Díez del Corral 2015). Notice that all these phenomena originate in the Iberian Sprachbund and then start being witnessed in the American varieties, both in Portuguese and Spanish.
The similarities in phonetics, morpho-syntax and the forms of address systems in American and the south-western region of the Peninsula are attested in the areas where the trade ports were established. The geographical closeness in the case of western Andalusia and southern Portugal as well as their common historical development led to a shared local paradigm that can be still attested. As has been pointed out, the own development with regard to the forms of address of the south-western region of the Iberian Peninsula and its extension throughout Latin America can only be justified because of the intense exchange during centuries. The relations among Seville, Cádiz, Lisbon and Algarve with the American colonies produced the levellings attested up to now, in comparison to other major ports elsewhere in the Peninsula, which maintained other linguistic features. If only the ports had been the reason, we would expect Porto or Bilbao to have developed similar features. But Porto and Bilbao received commodities from elsewhere and did not foster much exchange with the American colonies (O'Flanagan 2008). Their independence made the opposition between vós -vocês and vosotros -ustedes survive; on the contrary, the interdependence of the south-western ports among each other (Pike 1972) and with their American counterparts provoked a shared local development which can be observed in the lexical and phonetic features commented above, but mainly in the forms of address systems and their evolution until the present day.
In terms of pragmatics, the levellings studied in the south-western region go in line with the diachronic evolution of forms of address systems in Spanish and Portuguese on both sides of the Atlantic. As Molina Martos (this volume) puts forward, the increase of informal tú in the late 19th-century Spain was triggered by upper-class members and it was later spread by lower classes too. Similarly, the use of former polite você in Portugal as informal is related to the upper class and was later adopted by the rest of the social spectrum. And the same applies for the plural. Fernández (2012) confirms that the diffusion of informal vosotros to contexts where ustedes was the norm in the 18th century is also a change from above. It is exactly what Faraco (1996) remarks about vocês in Portugal at the same period of time.

Conclusion
My fieldwork has allowed for the detailed analysis of the parallel linguistic levellings towards ustedes, in Spanish, and vocês, in Portuguese, as well as a comparison of results with those from the first half of the 20th century. Based on these data, the two phenomena are characterised by an analogous behaviour. Nearly one century ago, the levellings were spread throughout western Andalusia and the Portuguese districts of Faro, Lisbon, Setúbal, Beja, Santarém, Évora, Portalegre, Coimbra and Leiria. Nowadays, the Spanish case is attested in the same Andalusian area (although it is declining) and the Portuguese case has also established itself in the districts of Aveiro, Viseu and part of Castelo Branco.
From a grammatical point of view, ustedes and vocês are topics reanalysed as subjects and this is why there are apparent disagreements. This transformation allows for two different agreement patterns to emerge every now and then, as has been observed cross-linguistically. As soon as ustedes and vocês are reinterpreted as subjects, the 3PL spreads hierarchically throughout the rest of elements in this order: the reflexive and the verb are the first one to adopt it (in the case of Spanish), followed by the direct object, the indirect object and the possessive. For Corbett (2006), the adoption of either agreement depends, among other things, on the independence that the target has with respect to its controller. In our case, the verb and the reflexive depend directly on the subject, so their independence from their controller is very limited and, therefore, they are more inclined to the 3PL. Objects and possessives, on the other hand, do not depend on an inducing element, but they are, as a whole, the first reference given of an entity, and this grants them enough autonomy not to agree with ustedes or vocês automatically. According to Wechsler & Zlatic (2003), the verb and the reflexive adopt the 3PL earlier because they follow index agreement and they have to receive the features of person, gender and number of the subject; objects do not depend on the subject or any other inductor, so they are not obliged to be built under the same parameters as the verb and the reflexive, while the possessive responds to the features of case, gender and number, since it is usually inserted in noun phrases and, as a consequence, it is characterised by concord agreement. Finally, we notice that the extension coincides with the studies carried out by Blake (2004) regarding case-marking and syntactic contexts, as many linguistic phenomena obey the hierarchy exemplified throughout the chapter: from relativisation to passivisation or the unmarked word order.
Finally, the utility of the type of research conducted here brings together different data and analytical approaches to understand how shared address patterns in an understudied Sprachbund came to be.