5.3.1. The usage of vós

The employment of vós has been scarce, as Table 7 depicts.

Although I have already made clear that the usage of vós does not depend upon any sociolinguistic feature (as determined by the statistical proof), I have intentionally included this information in Table 7 in order to demonstrate the generalised disappearance of vós in Madeira society. Scarcely 10% of informants have produced it within the expected contexts from where it may emerge.

The twelve occasions on which speakers produced it referred to the church, precisely one of the situations where it is expected to be attested, but its scarcity and its substitution by the pronoun vocês in most informants, regardless of their sociolinguistic profile, indicate that it is close to disappearance even in these contexts.

The gradual loss of vós even for certain stereotyped contexts resembles its Spanish counterpart in Hispanic America. The opposition vosotros-ustedes that dominates in Spain, the former being T plural and the latter V plural, stopped being valid in Hispanic America in the late 18^th and early 19^th centuries, when ustedes generalised as the single plural pronoun, regardless of the degree of politeness (Lara Bermejo 2022a). Nevertheless, vosotros was maintained as the selected pronoun within the church, army and parliament, exactly the same situations as for European Portuguese (Cunha & Cintra 1972). The usage of vosotros is widely attested in Hispanic America for these purposes, as Vázquez Laslop (2010) demonstrates for Mexico, but this same author has drawn attention to the fact that even these contexts are gradually dismissing vosotros as a formula and are on the verge of choosing ustedes plus 3pl. This is exactly what my corpus suggests regarding current Portuguese, at least in Madeira.

5.3.2. Address within the family

With respect to the family environment, Bazenga (2019) remarks upon the profound maintenance of non-T strategies from children towards their parents, with a persistence as regards the hierarchical view of family, as that which occurred in western societies prior to the French Revolution. According to this author, nearly three quarters of the people she surveyed preferred alternatives such as o senhor / a senhora, the noun pai and mãe followed by 3sg rather than 2sg, and the strategy of null subject plus 3sg. This tendency was, however, restrained to elderly speakers and to informants with a low educational background. My survey has further analysed this fact and in Table 6 I referred to the fact that the elderly are more inclined to do so, with this being less and less the case the younger the speaker is. Regardless, in this section I will deal with the strategies people from Madeira resort to for relatives in a higher position. In Tables 8, 9 and 10, I detail the number of occurrences of T, N and V strategies respectively to address parents.

Tables 8, 9 and 10 indicate firstly that T alternatives are based upon the pronoun tu and 2sg inflections. Nonetheless, non-T strategies exhibit further possibilities: the explicit usage of você, the employment of o senhor / a senhora, the status held within the family as the subject of the sentence (mãe, pai) as well as the null subject plus 3sg.

Firstly, the choice for the kinship status is always followed by the 3sg, revealing that it is not a T strategy. Secondly, the accusative inflected in 3sg is not reinforced by any stressed pronoun or form, which is why I have labelled it as N, as in the strategy of null subject plus 3sg, for it leaves the address opaque. Finally, the most diffused prepositional phrase is si, again increasing the opacity of the form of address, since it is useful for any resource that envisages the 3sg regardless of the subject form. In any case, as (22-24) demonstrate, at present 40% of Madeira speakers produce non-T strategies to address parents.

The resource of você has been employed by 4 elderly, 5 middle-aged and one young informant. O senhor / a senhora, on the contrary, has been used by one elderly, 2 middled-aged and two young informants. The noun phrase strategy is divided as follows: 9 elderly, 4 middle-aged, and 4 young people. Finally, the null subject has been preferred by 1 elderly, 4 middle-aged and 1 young. The overall results suggest the tendency within the family entity towards solidarity, but undergoing an intermediate phase earlier.

Apart from the fact that 40% of the sample used non-T strategies towards parents, the most diffused pronouns and alternatives are those associated with N, since o senhor / a senhora hardly appeared. In this sense, the family concept has passed from being considered hierarchical to being conceived slightly more solidary in these profiles, for they prefer N strategies such as você, the family status plus 3sg or null subject plus 3sg. It cannot be affirmed, however, that this slow tendency means solidarity, rather less hierarchy, since the abovementioned options are still deferential. Children gave V or N to their parents, but they were replied to with the T form, so there was no reciprocity in the exchange of the level of politeness. As a result, the transformation that the family has undergone in Madeira has entailed an in-between phase, before a final reversion to solidarity. Hence, the results found in this field study simply reveal the last relics of the family as a non-solidary entity, underlying this in-between stage before achieving this goal.

This in-between phase is likewise underpinned again by Spanish. García Godoy (2008, 2010) has documented the change in Spain that occurred in the early 19^th century, when upper-class and urban people started to address their parents through T. However, the usage of pronoun tú (T) for parents was not immediate and, before this happened, children underwent a stage in which they continued to employ V pronouns but with noun phrases that were perceived as T. Specifically, children ousted the terms padre (‘father’) and madre (‘mother’) by favouring the allocutives papá (‘dad’) and mamá (‘mom’). The emergence of these noun phrases did not imply the rise of tú, but there was a period of time in which T noun phrases coexisted with V pronouns for the same interlocutors. With time, children also deleted V pronouns and treated their parents by means of tú. This is not exclusive to the family, but Molina Martos (2021) states that T noun phrases are always the prelude to T pronouns. Her affirmations are based on the fact that there was also a generalised usage of tú in relationships of friendship or marriage in the late 19th century after the previous universalisation of T noun phrases for these contexts (Molina Martos 2020).

Unlike European Portuguese, Peninsular Spanish does not possess N pronouns or strategies. Whereas Spanish mixed T noun phrases with V pronouns in the intermediate stage before the total establishment of pragmatic solidarity in the abovementioned contexts, Madeira Portuguese has selected N forms and alternatives for this in-between phase towards pragmatic solidarity.

5.3.3. The usage of você and null subject + 3sg

One of the relevant factors to be assessed as regards the pronoun você is its pragmatic range in current Madeira. Based on my corpus, this resource has arisen on 15 (6.7%) occasions to address someone politely; on 83 (37%) occasions to address an interlocutor who is in a lower position; and on 126 (56.3%) occasions to treat an addressee who is in an equal position. In the former case, the informants selected it to address a manager (5 young, 6 middled-aged and 3 elderly), while in the second situation, the addresser was either a professor who spoke to a student or a manager who addressed a subordinate. In the latter case, the interlocutors covered clients, non-elderly strangers and neighbours.

The distribution of pronouns makes it necessary to assess whether it also depends on the diaphasic context or if it is only a matter of sociolinguistic factors. In Table 11, I detail the number of times every pronoun has arisen and in what context.

Table 11 illustrates that tu cannot be resorted to as a V strategy and neither can o senhor be employed as T. The former is useful in intimate contexts and, to a lesser extent, in neutral situations in which interlocutors share power. The latter is employed to give V to someone and it can also appear in N situations. As was stated by the literature regarding continental Portugal, você can be perceived as T, N and V. In Madeira, following the results of this field study, você as asymmetrical ascendant (V) is strictly limited to those informants that still use it for parents as well as, on very rare occasions, to address a manager. The most diffused pragmatic value of você in current Madeira lies on N and, in the case of T, if there is no reciprocity in the treatment. In other words: você is deferential when it is expected to serve for both interlocutors, and asymmetrical as it is also employed to address someone in a lower position. To sum up, tu is solidary, o senhor is both deferential and hierarchical to address someone in a higher position, and você is both deferential and hierarchical to address someone in a lower position. The difference in the employment of o senhor and você for deference is the degree of communicative distance, the former being more distant than the latter. As a matter of fact, the statistical results leave no doubt; based on the chi squared, the p-value is < 0.00001, significant at p < .05.

The apparent contradictory behaviour of você resembles the pragmatic path of singular vós in both Portuguese and Spanish as a V pronoun. The pragmatic origin of this form was V in the Middle Ages, but as V forms emerged by default, their courteous connotation stopped being perceived and by the late Middle Ages and the early Modern Era, speakers made up new polite strategies to replace the gradual loss of courtesy that vós was undergoing (Cintra 1972). Nevertheless, the pragmatic depreciation of vós entailed a period of time in which it could be considered in a contradictory way: polite, informal and even offensive. Vós became an N pronoun, for contexts in which a V strategy was too polite and where tu was not polite enough. Thus, European Portuguese exhibited at that time a tripartite system in singular, in which vós could be employed as the current você. The problem with N situations is the certainty about such a consideration on the part of the interlocutor. The boundary between a T and an N context can be narrow, as the frontier between N and V can also be. This is why tripartite paradigms in singular are rarely attested cross-linguistically (Helmbrecht 2005) and can be witnessed for example in Romanian.

The fear of committing a face-threatening act obliges the addresser to know some data about his/her interlocutor in order to satisfy the communicative act. As você can be problematic for the abovementioned reasons, speakers solve this problem by not explicitly producing a pronoun and resorting to a null subject plus 3sg or to a noun phrase. This alternative is crucial, since the explicit expression of a subject is sometimes necessary for topical continuity or for contrastive purposes. When the informants who took part in this survey needed to produce a subject in an N situation and did not want to use você, they asked me about the proper name of the addressee. Only when they were given the name could they go on with the discourse. In the cases in which no proper name was given, they could not help inventing it. Additionally, if they knew the job their addressee had, this problem was solved by referring to the professional status, but the knowledge of the proper name of the addressee has become essential in current European Portuguese to avoid impoliteness in certain situations in which the addresser does not consider você the right subject. This can even be valued in the distribution of the null subject + 3sg and the explicit usage of a subject in both N and V (Table 12).

Table 12 shows that, contrary to the statements in the literature regarding continental Portugal, Madeira speakers prefer the explicitness of a subject in the most problematic pragmatic context of all: N. However, V situations give the opposite picture though with a lesser gap between percentages. It is pertinent to recall that accounting for the null subject plus 2sg is useless, because any 2sg inflection with or without a subject implies the usage of tu. The data Table 12 provides is interesting, as it shows the need to make the degree of politeness clear through the treatment itself in spite of the fact that the 3sg alone avoids such a commitment. In this sense, Madeira seems to follow a different path from continental Portugal, as Bazenga (2019) claims.

The results I present regarding você contradict a number of statements that can be found repeatedly in the literature, but which are not supported by any empirical research. For instance, Bacelar de Nascimento, Duarte & Mendes (2018) state that in Portugal, você is employed by an elite who use it interchangeably with T as a way to differentiate themselves from other social classes. This is the first theory yet to beunderpinned by any empirical data and which demands specific research. Secondly, these authors also remark that Portugal is experiencing an increasing tendency towards the usage of você in younger generations and speakers with low educational backgrounds. My study does not support this hypothesis at all (it is important to highlight that these statements are not backed up by any empirical study) and, with regards to Madeira, você is widely attested irrespective of sociolinguistic variables, except for sex, though to a minimal degree. Thirdly, Bacelar de Nascimento, Duarte & Mendes (2018) resort to quite an extended cliché in Lusophone linguistics when they affirm that the supposed increase of você is a direct influence of TV series produced in Brazil (telenovelas) as well as of the mass migration on the part of Brazilian people to Portugal.

As regards the aforementioned cliché, apart from the lack of empirical evidence it emerges as the least likely hypothesis of all. In the first place, language contact and dialect contact (which more or less behave the same) exhibit a series of stages that depend on both linguistic and extra-linguistic features. While the former usually affect lexicon, to a lesser extent phonetics, very rarely morpho-syntax and virtually never pragmatics, the latter respond to the prestige and consideration of every language and dialect involved in the contact (Van Coetsem 1988, Heine & Kuteva 2005). Applied to European Portuguese and its contact with Brazilian Portuguese via television and migration, this means that the Portuguese would have adopted many lexical items from the Brazilian variety, had they incorporated você as it is used in Brazil. Moreover, the Portuguese would have started to produce certain phonetic contexts in a Brazilian way, had they incorporated você as it is used in Brazil. Additionally, the Portuguese would have developed morpho-syntactic features coming from the Brazilian variety, had they incorporated você as it is used in Brazil. To summarise: if Portugal had reinterpreted the pragmatic employment of você because of a direct influence from the Brazilians, then European Portuguese would have to exhibit an array of linguistic features, mainly in lexicon, less in phonetics and even less in syntax, influenced by the Brazilian variety. However, this has not occurred. The usage of você has nothing to do with the Brazilian reality.

Empirical evidence supports my proposal that the employment of você in European Portuguese is not affected by the Brazilian paradigm, at least regarding Madeira. Pragmatics is a characteristic that varies depending on a specific country, but not on the language. It is very rooted in the beliefs of a given society and, consequently, the pragmatic system of Portugal is different from that of Brazil, which is also different from that of Angola and so on. It is not Portuguese that is the factor that imposes a type of pragmatic system, rather, it is the country regardless of the language and languages spoken within. For example, politeness in Switzerland is not conceived the same way as in Germany, Italy and France, despite the fact that this country presents the three languages. Indeed, Switzerland applies a type of politeness that is useful for all of its official languages (Hickey & Stewart 2005).

The same happens in the Hispanic world. The pragmatic conception of politeness in Spain is not the same as that in Costa Rica, although both countries possess Spanish (Hummel, Kluge & Vázquez Laslop 2010, Hummel & Lopes 2020). Within Spain, however, speakers of Spanish, Galician, Catalan and Basque behave identically in this respect because the pragmatic system is constrained to what Spain understands, and not to what every language triggers. Furthermore, the shift in pragmatics, specifically in relation to politeness, follows the prestige (Lara Bermejo 2022a). Spain changed to pragmatic solidarity in the late 18^th century as a French influence because the prestige at that time was France (Lara Bermejo 2020b). But in the case of dialect contact, the change is usually made by those who are in a lower position. Sinner (2010) investigated the pragmatic behaviour of Argentinians who migrated to both Germany and Spain. While they did not develop any shift in Spanish if they lived in Germany, those who settled in Spain gradually took the Spanish pragmatic system and not the other way around. Lara Bermejo (2018b) studied the adoption of the politeness system from Ecuadorian people in Spain and concluded that they tended to adopt the Spanish system because it is the prestigious model in that country, but they did it in stages and very slowly, underlying again the reluctance towards pragmatic shifts even within the same language. Hence, it is very unlikely for Brazilian immigrants to have conditioned the Portuguese in Portugal. The statements by Bacelar de Nascimento, Duarte & Mendes (2018) do not have counterparts elsewhere.

5.3.4. Tendency towards solidarity

I affirmed above that present day Madeira is gradually accepting solidarity in the plural. The comparison of occurrences suggests this fact: vocês was produced 311 times whereas os senhores / as senhoras was expressed 29 times. The latter was resorted to when the addressees were elderly and in situations that implied distance, but not necessarily hierarchy, such as when dealing with customers. It is worth mentioning that the same speakers that talked to their parents by means of a V strategy in singular opted for vocês and not for os senhores when they addressed both parents. Likewise, most informants that resorted to o senhor or a senhora in singular preferred vocês for the same situations in plural. As that which could occur in singular with the choice of o senhor and você or null subject plus 3sg, the selection for either vocês or os senhores in certain contexts, mainly N, is subject to the degree of distance and deference that the addresser establishes with respect to his/her addressees. But, roughly speaking, vocês can be used for plural contexts where o senhor / a senhora would be the most adequate strategy in singular.

Solidarity usually starts in the plural rather than in the singular in languages that have pronominal divergences in this grammatical number for different diaphasic situations. For example, French or Italian do not exhibit different pronouns in plural to distinguish the degree of politeness, for vous can serve for both T and V and voi behaves alike. However, Peninsular Spanish opposes vosotros for T to ustedes for V and Lara Bermejo (2022b) has demonstrated that pragmatic solidarity in Spain began in the plural in the late 18^th century. From that moment, Spanish society gradually tended to accept more contexts via vosotros, but did not do the same thing in singular. It was only when this happened in plural that Spain further prompted pragmatic solidarity by also resorting to singular T pronouns in contexts where a V pronoun was previously mandatory. Moreover, as Siewierska (2004) argues, it is quite rare to attest languages that also establish pronominal differences in plural to express social deixis, and this is why most languages only exhibit an array of pronouns (two in French or Italian, up to four in Hispanic America) to mark diaphasic distinctions in singular. The plural lends itself to syncretism and, as a result, entails fewer pragmatic risks; therefore, the plural favours profound pragmatic changes in society, as occurs with the case of pragmatic solidarity.

5.4.1. Objects and oblique: stressed pronouns, dative and the prepositional phrase

As for the usage of stressed pronouns in unstressed positions, this has only happened with the pronoun você. Specifically, it has been produced 9 times as an accusative and, though there are no restrictions that apply to age or sex, 2 people have a high educational background whereas 7 informants have a lower educational level (25-26). Likewise, the employment of dative pronouns in accusative contexts has also arisen 4 times, all of them produced by young speakers with no educational background (27-29).

The resource of stressed pronouns in unstressed contexts, mainly direct objects, is widely documented in Brazil, but it has also been pointed out for Madeira by Bazenga, Andrade & da Silva Rodrigues (2016). However, in their study, this alternative is found in people with a low educational background. This is also the main sociolinguistic variable in my corpus, though it has also emerged in informants with the opposite profile. As for the dative in accusative environments, this lhe instead of the normative accusative pronouns has also been attested in continental Portugal as well as in Madeira with reference to a semantic third person (Segura da Cruz 1991, Bazenga & Rodrigues 2019), but in my corpus this has occurred with reference to você and not to any other 3sg strategy. This suggests that lhe could also appear for reasons of politeness, imitating the model in Brazil and in Spain. Regardless of the sociolinguistic factors that may prompt this shift, its mere existence obeys two specific purposes: the differentiation between a semantic and a syntactic third person, and the promotion of the human entity in the topical scale, conveyed by the most common grammatical case (the dative).

The former reason can also be found in Brazilian Portuguese and Peninsular Spanish, where the dative lhe and le, respectively, have established themselves as the unique unstressed pronouns in third person in order to distinguish politeness and a semantic third person. By doing so, both varieties resort to normative accusative pronouns where the entity is he or she, but change to dative in any unstressed context when the reference is an addressee who should be addressed politely. As a result, this variation in case responds to pragmatic motives (Fernández-Ordóñez 1999, De Souza & Lopes 2015, Gómez Seibane 2021).

The latter reason is a cross-linguistic tendency prompted by several semantic and discursive grounds. Givón (2001) pinpoints that the dative case usually conveys a human entity, embodied in a number of semantic roles: beneficiary, recipient, possessor or even dative, which the author defines as a conscious participant in the event, typically animate, but not the deliberate initiator. This last semantic role can materialise in different syntactic functions (30-32).

Instances (30-32) underlie that the element in italics is semantically a dative and, syntactically, it can be the subject, the direct object and the indirect object. In other words: a dative can be these three syntactic functions, but a patient or an agent cannot. Moreover, as the dative is usually a human entity and can emerge in all those syntactic positions, it is inclined to be conveyed by a specific marker, also because of discursive reasons. Again, according to Givón (2001), the human entity is the most salient feature in discourse, tending to be positioned at the beginning of the speech because of a topical hierarchy that promotes the human to the first place. Furthermore, the human entity also lends itself to being referred to in a specific way to follow its saliency: this means that, regardless of its syntactic function, it may be expressed coherently and differently with regards to non-human entities. As a result, humans may develop their own marker for any syntactic case in contrast to a different marker for other referents.

If we compare these cross-linguistic tendencies to the dialectic realities in Madeira or elsewhere in Brazil and Spain, it is possible to observe that all of them follow these assumptions. Firstly, the human entity is conveyed coherently with its own marker, in this case le in Spanish and lhe in Portuguese. Secondly, the human entity is inclined to be expressed differently with respect to another referent even for the same syntactic functions, which is why Spanish and some varieties of Portuguese have developed a clitic for polite third person and another one for a non-human or semantic third person (Fernández-Ordóñez 1999, Gómez Seibane 2021). Thirdly, the two of them have selected the dative case, which coincides with the semantic role of dative and which is nearly always a human entity (Givón 2001).

Apart from these particularities in the object position, Madeira Portuguese has syncretised the stressed pronoun in si (or consigo when the preposition is com ‘with’) within a prepositional phrase to refer to non-T forms: 176 times in contrast with 26 occurrences of stressed pronouns homophonous to the subject form. Nonetheless, the usage of si is only 3sg and cannot refer to a plural entity. The relevance of si is the fact that it is useful for você and o senhor / a senhora, again reflecting the problematic feature of singular pronouns that are not tu. By resorting to si, speakers avoid the explicit production of the address, which may only arise as the subject. This applies to noun phrases and null subject plus 3sg too. The most important pragmatic decision nowadays in the singular in syntactic terms is the production of the subject, while si has been applied in the prepositional phrase as a solution in this grammatical context.

5.4.2.Extension of the 3pl anchoring vocês

In the previous paragraphs, I highlighted the fact that, even though vocês is syntactically 3pl, European Portuguese is characterised by exhibiting agreement mismatches between the pronoun, verb and reflexive (3pl) and possessives and objects (2pl). This does not happen in Brazilian Portuguese where vocês systematically induces 3pl, but Portugal also presents different extension of the 3pl, depending on the geographical area. In the case of Madeira, in Table 13 I depict the distribution of agreement referring to vocês.

Table 13 pinpoints that any syntactic element anchoring vocês is coherently inflected in 3pl as long as it is the subject, the reflexive and the verb. The rest of the elements swing between the 2pl and the 3pl, but not to the same degree. Whereas the accusative gives more occurrences of 3pl (though the 2pl is the major one), there is more reluctance towards the dative as regards 3pl and, finally, the possessive hardly inflects in 3pl. It must be stated that the prepositional phrase has also given vocês as the preferred stressed pronoun, except when it is preceded by the preposition com (‘with’). In this case, the usage of the 3pl is 86% (19 times), and the strategy convosco¸ syntactically 2pl, appears 3 times (14%).

The picture provided with by the data clearly establishes that Madeira follows the same pattern as continental Portugal, since the 3pl undergoes a specific hierarchy: subject > verb / reflexive > accusative > dative > possessive. The behaviour of the accusative allows for analysing the path transited by the 3pl, since it starts emerging in this case when the pronoun is enclitic (10 times in 3pl versus 15 times 2pl) while there is a leaning towards proclisis in 2pl (10 times 3pl versus 74 times 2pl) (33-34).

The rise of the 3pl in enclitic accusatives in my corpus has been mainly produced after an infinitive. The concatenation of the sound /r/ plus the sound /v/ might have been a phonetic context that has triggered the conversion of the 2pl into a 3pl. This hypothesis is supported by an analogous phenomenon witnessed in Spanish in the 16^th century. According to Calderón Campos (2010), the current 2pl unstressed pronoun is os and not vos as it was in the Middle Ages, because the pronunciation of the sequence /r/+/v/, such as the one in Portuguese, produced the gradual loss of /v/ in these contexts. Once vos turned itself into os whenever it followed an /r/, speakers ended up generalising os in every situation regardless of the phonological constraints. But the final shift of vos into os commenced in these specific phonological sequences. The same seems to apply to the establishment of the 3pl in the accusative in Portuguese.

However, this shift is firstly attested in the accusative, since the same speakers that preferred 3pl in the accusative after /r/ did not produce a 3pl in the dative after /r/. Therefore, the spread of the 3pl must necessarily undergo the abovementioned hierarchy. This continuum is not arbitrary either, but it responds to the grammatical behaviour that many phenomena follow when they are subject to grammatical case (Blake 2004). For instance, causativisation, passivisation and relativisation are also constrained by this hierarchy (Comrie 1976, Keenan & Comrie 1977). English can make a passive out of an indirect object (dative) because it can do so out of a direct object (accusative). The languages capable of relativising a genitive (whose) can do so because they are able to relativise the subject, the direct object and the indirect object. Hence, the extension of the 3pl in vocês obeys this same feature and can even have a counterpart in Andalusian Spanish, where the spread of the 3pl referring to the pronoun ustedes runs across this same hierarchy at the expense of the 2pl that refers to the eliminated pronoun vosotros (Lara Bermejo 2018a, 2020a).

5.4.3. The loss of vós

The loss of vós as 2pl has already been discussed: the scarcity of speakers who have expressed it in expected contexts is proof of this fact. But this tendency can even be underpinned by the agreement pattern referring to the stressed form, for not all informants expressed a verb in 2pl right after the subject, but two of them preferred a 3pl agreement, suggesting again that vocês has virtually established itself as the plural pronoun in any T and N situation (35-36). It is pertinent to recall that only ten people produced vós, so 20% them are already spreading the 3pl even with this referent.

The instances pinpoint the grammatical instability of vós, as a proportion of the few speakers who resorted to it were not coherent in agreeing on all the elements in 2pl. While (35) already enables the verb in 3pl right after the treatment, (36) shows that, despite the first verb inflecting in 2pl, the second prefers the 3pl though its referent is the same one. This apparent contradiction, according to Corbett (2006), responds to a change in progress that opposes two different agreements. The extension of one of them at the expense of the other reveals the deletion of the latter. However, this process entails not only gradualness, but the coexistence of the new agreement with the one which is being ousted. This is exactly what (35-36) suggest.