2.1. Surface DPCs

Romance dative structures with the thematic role of Recipient are traditionally identified as DPCs, since the Recipient is introduced by the functional, Case-assigning preposition a ‘to’ (e.g., Baker, 1988; Kayne, 1984; Vergnaud, 1974). However, the earlier work on datives has become somewhat overshadowed by proposals that analyze ditransitive constructions in Romance languages, including EP, as DOCs and applicative constructions (e.g., Spanish: Cuervo, 2003, 2010, 2020; Pineda, 2016, 2020; Portuguese: Torres Morais, 2006, 2012; Torres Morais & Lima Salles, 2010; Torres Morais & Berlinck, 2018, based on the applicative analysis (Pylkkänen, 2002, 2008) and on the idea that transfer-of-possession meaning is restricted to DOCs (e.g., Harley, 2002; Harley & Miyagawa, 2017).

The applicative analysis is drawn from Pylkkänen’s proposal of low applicative heads, and considers that datives in ditransitive structures bear no thematic relation with the verb, the transfer of possession relation being established directly between the Recipient and the Theme. Elsewhere, we have argued against the original applicative analysis of dative structures in EP, both on empirical and theoretical grounds. In particular, we considered that it originated from misinterpretations of both Bantu and Romance data (Duarte, Gonçalves & Hagemeijer, 2017). First, and differently from Romance, applicative extensions in Bantu are a morphological valency-increasing operation, which, when applied to transitive verbs, generally yield a DOC in which the primary object does not necessarily bear the semantic role of Recipient (cf. Larson, 2014). Second, and crucially, Romance languages exhibit true lexical ditransitive verbs, which subcategorize two internal arguments, a property that is actually also found in many Bantu languages, even though with a limited number of verbs. The search for universal properties of the computational system C_HL should not obscure these differences between Romance and Bantu grammars. Therefore, we assume that it is theoretically inadequate and empirically unmotivated to project an applicative head to derive arguments of the verb in Romance on the basis of the properties of applicatives in Bantu.¹

In EP, preposition a has traditionally been analyzed as a dative Case marker (Brito, 2009; Duarte, 1987; Magro, 2019; P. Gonçalves, 1991; R. Gonçalves, 2016; Torres Morais, 2006), which contrasts with the homophonous preposition a introducing the transitory Goals of motion verbs and alternates with para, which is typically used for permanent Goals. Evidence for the distinction between the functional preposition a and the homophonous directional preposition firstly comes from the fact that only the argument introduced by the former can be replaced by the dative clitic lhe(s) (see (1–2)), whereas the latter can only be replaced by the adverbial lá (cf. (3)).

The functional status of preposition a which occurs in ditransitive structures with the thematic role of Recipient finds additional support in its behaviour in coordination structures. As first noticed by Vergnaud (1974, p. 247) for French, functional a in EP does not license coordinated NPs, contrasting with the directional preposition a, as shown in (4–5).²

Finally, functional preposition a can occur in clitic doubling structures in EP, which, contrary to Spanish and Romanian, is only possible with pronouns and universal bare quantifiers todo(a,s), ‘all’, and ambos(as), ‘both’. This construction, labelled ‘pleonastic object’ by traditional grammarians, carries a contrastive focus interpretation.³

In section 5.2 it will be argued that the facts from Portuguese varieties can be accommodated under a modified Larsonian VP-shell (Larson, 1988, 2014) within a view of argument structure which results from the lexical properties of the verbs and the prepositions they combine with (Rappaport-Hovav & Levin, 2008). In particular, ditransitive structures which involve give-type verbs (core dative verbs in Rappaport-Hovav and Levin’s (2008) classification) and Recipients in EP will be analyzed as surface DPCs, identifying them as a-DP / K(ase)P structures. The head of the KP is a functional preposition which assigns Case to the DP (or a dative clitic); the Recipient is the complement of K. This analysis will also be extended to the dominant pattern found in AVPs, which is analogous to EP.

2.2. Ambiguous input

Based on Rappaport-Hovav and Levin’s (2008) distinction between give-type verbs, referred to as core dative verbs, and throw and send-type verbs, labeled as non-core dative verbs, R. Gonçalves (2016) proposes that the argument structure of ditransitive verbs in EP is ambiguous, showing that core dative verbs, such as dar ‘give’, emprestar ‘lend’, oferecer ‘offer’ or mostrar ‘show’, which only have a caused possession meaning, exclusively select Recipients introduced by the functional preposition a (cf. (1) and (4, a)). On the other hand, non-core dative verbs, such as enviar ‘send’, atirar ‘throw’, levar ‘bring, take’ or trazer ‘bring, take’, which have both a caused possession and a caused motion meaning, can select [+ANIM] Recipients introduced by the functional preposition a (cf. (2) and (4.b)), as well as ‘true’ Goals introduced by the homophonous directional preposition a (cf. (3) and (5.b)), which is also selected by the directional motion verb ir ‘go’ (cf. (5.a)), and varies with para.

According to Rappaport-Hovav and Levin (2008), a test involving wh-questions shows that give-type verbs, unlike throw/send-type verbs, lack a Goal argument. Only the argument selected by non-core dative verbs can be questioned by both quem ‘whom’ and onde ‘where’ and subcategorize [±ANIM] arguments, as illustrated in examples (8–9) below (e.g., Brito, 2010; R. Gonçalves, 2016). With core dative verbs, the motion reading is unavailable, as illustrated by the ungrammaticality of (7b).

Furthermore, a can alternate with para with non-core dative verbs, irrespective of the direct object being [±ANIM].

In addition to the ambiguity in the argument structure of dative verbs, plain Beneficiaries (Van Valin & LaPolla, 1997)⁴ in EP can be introduced by both a and para and cliticize through dative lhe(s) or strong pronouns introduced by the preposition para (cf. (11)) (e.g., Brito, 2009).

Furthermore, contrary to what happens with Recipients, with plain Beneficiaries, the internal direct argument is typically an affected Theme, as with verbs like preparar, ‘prepare’, assar/cozer ‘bake’, ‘compor, ‘compose’, construir, ‘build’, desenhar, ‘draw’, fazer, ‘make’, etc.

From the perspective of an L2 acquirer/learner, the ambiguity caused by constructions involving non-core dative verbs and Beneficiaries may therefore be taken as evidence that datives in general can be introduced by a ‘to’ or para ‘toward’, leading to inter and intra-speaker variation.

In the next section, we will describe and discuss the properties of dative structures in each of the three AVPs, with a strong emphasis on variation.

3.2. Ditransitive structures in spoken urban corpora

4.1. Changana (Mozambique)

As in many Bantu languages (Hyman & Duranti, 1982), Recipients in Changana (S53) can occur with a small group of non-derived ditransitive verbs, i.e., verbs that select two internal arguments. With these verbs, Recipients occur typically adjacent to the verb (cf. (31.a)), can occupy the subject position of passive sentences (cf. (31.b)), and can be anaphorically recovered by an object marker (cf. (31.c)).

With these non-derived core dative verbs, the DOC construction in (31.a) above does not alternate with a DPC (introduced by ka in this language) (cf. (32)).

In addition to non-derived ditransitives, Changana also exhibits core dative verb roots, such as nyikel ‘hand(to)’, which result from a diachronic process of lexicalization of a verb root (nyik-) and the applicative extension -el.⁸ Differently from non-derived roots, these forms are able to license DOCs and DPCs introduced by preposition ka (cf. (33)).

Moreover, the Recipient may occur as a pronoun incorporated into the verb, with the same form of an object marker (cf. (34.a)), or as the object of prepositional ka (cf. (34.b)).

Finally, in both cases the Recipient can occupy the subject position in passive sentences.

With respect to examples such as (35.b), Chimbutane (2002, p. 131) states that “the passivized constituent in this case is no longer a goal but rather a locative. That is, [ka kokwani] is not analyzed as the “intended recipient” of the letter but rather a “location” to which such a letter has been sent.”

In sum, Changana exhibits two structures to express Recipients with core dative verbs, a DOC and a DPC. The former is the exclusive structure with non-derived verbs, such as nyik ‘to give’ and komba ‘to show’, which are inherently semantic and syntactic ditransitives, whereas the latter is additionally available with verb roots which exhibit a lexicalized applicative extension, such as nyikel ‘hand to’.

4.2. Kimbundu (Angola)

Predicates with a transfer of possession feature in Kimbundu require locative marker ku to introduce the Recipient. This marker, which diachronically derives from noun class marker 17, behaves like a free morpheme in the contexts under analysis (cf. Diarra, 1990; Hagemeijer et al., 2019). Kimbundu also exhibits differential object marking based on animacy, since only Recipients are introduced by a locative marker and connective/genitive morpheme -a, as shown in (36.a). Similarly to Changana, pronominalized Recipients in Kimbundu occur as class-based prefixes attached to the verb root (cf. (36.b)).

Kimbundu lacks overt passive morphology (e.g., Chatelain 1888–89; Givón, 1994), but allows for impersonal structures with the interpretation of functional passives.⁹

DOCs are not allowed in Kimbundu, irrespective of the relative word order between objects.

In sum, ditransitive verbs in Kimbundu expressing Recipients only occur in the form of DPCs.

4.3. Forro

DOCs are a property of most Atlantic creole languages (e.g., Adone, 2004; Michaelis & Haspelmath, 2003; Bruyn, Muysken & Verrips, 1999), and are also the dominant structure to express [±ANIM] Recipients in Forro, the Portuguese-related creole historically in contact with STP (cf. (39)).

Recipients cliticize to the verb (cf. Hagemeijer, 2007).

Dative passives (indeed, syntactic passives in general) are not attested in Forro. To sum up, Forro does not exhibit dative alternation between DOC and DPC to express Recipients with core dative verbs.

In the next section, we will summarize the structures exhibited in the AVPs and their contact languages to assess the role of contact language with respect to the observed patterns.

5.1. The interplay between language contact and EP input

Studies on post-colonial varieties tend to emphasize the grammatical properties shared between the nativized/nativizing varieties and the contact languages. In sections 3 and 4 above, we presented and discussed the dative structures with core dative verbs observed in each of the urban African varieties considered here, as well as their primary contact languages, and the degree of regularity in the productions of the informants at individual and group level. Table 12 summarizes the data.

From a merely comparative perspective of the broad sentence patterns available, the table shows that DPCs and dative clitics are observed across the three varieties at stake. Moreover, MOP and STP have in common that Recipients are commonly expressed through DOCs and DPCs, whereas AP shows a clear preference for DPCs. STP and AP, on their turn, can be set apart from MOP due to the absence of dative passives.

At first sight, a rough comparison of the information contained in the table suggests that language contact is a determining factor in the expression of Recipients in these AVPs, since (i) both Changana and MOP exhibit two dative sentence patterns and dative passives, (ii) both STP and Forro exhibit DOCs, but no dative passives, and (iii) AP and Kimbundu only exhibit DPCs and lack dative passives. That the role of language contact cannot be simply downplayed also follows from the numbers regarding the predominant alternative patterns to standard DPCa: DOCs with full NPs in MOP and STP represent respectively 29,6% and 46,4% of the occurrences; DPCs introduced by em in AP correspond to 29,7% of the contexts. As shown in section 4, DOCs in the former two varieties are clearly based on the predominant pattern in the contact languages (Changana and Forro), whereas the DPC with locative preposition em in AP is arguably based on a similar syntactic pattern with KB locative marker ku. All in all, when crossing the different patterns available in the AVPs with the patterns in the main contact languages, structural convergence can therefore be observed to some extent.

The weak phonetic status and the lack of semantic transparency of preposition a as input for L2 acquisition, combined with the observed EP ambiguous input in the expression of [+ANIM] Goals with non-core dative verbs and Beneficiaries (a/para), arguably function as additional factors in the process of restructuring toward the contact languages.

The contact hypothesis has been extensively explored by P. Gonçalves (1991, 2002, 2004, 2010) with respect to MOP. She proposes that MOP speakers start out with the parameter settings of their L1 Bantu languages while acquiring Portuguese, and that ambiguous input leads to the retention of features that are non-convergent with EP. Analyzed from the perspective of Bantu languages parameters, P. Gonçalves further argues that the occurrence of the dative clitic may suggest to the learners/acquirers that preposition a is not required for Case assignment. Therefore, she claims that this preposition is absent from the initial acquisition stages and introduced in more advanced stages of acquisition, where it becomes part of the underlying Bantu Case-marking system. With respect to Recipients in MOP (Beneficiaries in P. Gonçalves’ terminology), it is therefore claimed that DOCs are cases of incorporation of preposition a (à la Baker, 1988). Moreover, she argues that, since Bantu nouns with the semantic feature [+human] occupy the most prominent syntactic positions (e.g., Hyman & Duranti, 1982), in final stages of acquisition, MOP speakers reanalyze the preposition a as what we would nowadays call a differential object marker, introducing nouns with the semantic feature [+human]. As a result, she claims, in MOP grammar, direct objects are introduced by a or realized as lhe(s) in intermediate and final stages of acquisition.¹¹

In fact, in the past, few Mozambicans had access to or spoke Portuguese (24,4% of L1+L2 Portuguese speakers in 1980, of which only 1,2% where L1 speakers), which renders the substrate hypothesis plausible, irrespective of the theoretical details of the treatment of Bantu languages and, by extension, of MOP. However, the numbers of L1 and L2 speakers, especially in urban environments, have been on the rise, with increasing bilingualism and proficiency in Portuguese over the last decades. The numbers are even more compelling in the case of urban AP and STP (cf. section 3.1), where widespread L1 transmission takes place. This means that the substrate hypothesis increasingly lacks explanatory power to account for the contemporary urban AVPs in question. The role of the substrate still lingers on, but is primarily the role of intensive, historical language contact.

Abstracting away from contact, our quantitative analysis of the data in section 3 shows that the dominant structure in the three AVPs is the standard DPC introduced by the functional preposition a. Although DOCs and other prepositional sentence patterns are also employed (DPCem/para), the dative clitic is systematically used in the three AVPs. On the one hand, the dative clitic is convergent both with the contact languages and EP, as all of them resort to clitic pronouns or affixes. On the other hand, its generalization, even among those informants who use DOCs and/or DPCs introduced by para or em, lends support to the hypothesis that em and para are being reinterpreted as functional prepositions, similarly to a. In other words, a significant number of speakers exhibit a grammar that is largely convergent with EP concerning the expression of Recipients.

But competition between grammars is also part of the equation. Indeed, Lightfoot’s (2006, p. 164) claim that “there are coexisting grammars within [these] speech communities and even within the brains of some individuals, who have multiple competencies” can be entirely applied to the AVPs. Individual speakers may exhibit different sentence patterns to express Recipients and different groups can be established based on the patterns used. It is, however, hard to find speakers whose grammar solely reflects the substrate patterns. The small group of STP speakers that exclusively exhibits DOCs (6/64) would be a case in point.

To sum up, the dominant use of the DPC introduced by preposition a shows that acquisition of the AVPs privileges the standard, EP pattern. On the one hand, this is underscored by individual variation, which shows a predominance of groups whose production is restricted to DPCs (and dative clitics), even though some speakers exhibit variation between different prepositions, especially in AP, where DPCem is the main competitor of the canonical pattern. On the other hand, speakers who produce patterns that are distinct from EP generally also exhibit variation with DPCa. Thus, in a scenario of language transmission and acquisition, the balance appears to be tipped to favor input of the EP pattern. Moreover, the effect of schooling, which we did not discuss here, cannot, of course, be ignored with respect to variation and change. Exposure to the EP standard through education could be the basis of further stabilization of DPCa.

In the next section, we will try to formalize the variation observed in the expression of Recipients in AVPs by using syntactic and lexical microparameters.

5.2. An account of variation through microvariation theory

In the sections above, we have argued the syntactic variation observed in AVPs can be mostly reduced to two structures, a DPC and a DOC, which may or may not occur in each variety, depending on inter and intra-speaker variation. The dominant structure in the three AVPs, which converges with EP, is that of Recipients as extended DPs, since they are introduced by a functional, Case-assigning preposition. As we have shown, the co-occurrence of the dative clitic not only with DPCs introduced by a, but also with other prepositions, clearly points to the reanalysis of em and para as functional prepositions in the structures at stake. In other words, similarly to the morphologically syncretic form a in EP, which corresponds to a Case marker and to a full preposition introducing Goals, em and para have evolved to syncretic prepositional heads, corresponding both to Case markers and to full prepositions introducing Locatives and Goals. On the other hand, in DOCs, which correspond to the substrate-induced hypothesis of P. Gonçalves (1991, 2002, 2004, 2010) for MOP, and which were also shown to be widely available in STP, Recipients are plain DPs, which may pronominalize as strong pronouns in a limited number of cases (11,1%).

Within the Minimalist Program (Chomsky, 1995, and subsequent work), the locus of parametric variation, which concerns the way a speaker’s knowledge of language can differ from another’s, is assumed to be restricted to functional elements in the lexicon, since syntax is arguably uniform across languages and not subject to variation. The proposal that variation occurs in the lexicon, firstly presented by Borer (1984) and later adapted and integrated by Chomsky in the Minimalist Program, was coined by Baker (2008) as “The Borer-Chomsky Conjecture” (cf. 41)).

As a result, as stated by Baker (2008, p. 355), “the standard microparametric view is that the primitive, scientifically significant differences among languages are always relatively small scale differences”, which would be observable even among close related dialects (or varieties). This is exactly what we observed in the AVPs’ DPC structures described in previous sections, which act as a powerful testing ground for formal hypotheses.

In order to derive the ditransitive structures exhibited in EP and AVPs, we will not adopt an applicative analysis, as several crosslinguistic arguments have convincingly shown that neither a low nor a high applicative analysis for Romance is empirically motivated (e.g., Boneh & Nash, 2012; Duarte, Gonçalves & Hagemeijer, 2017; Larson, 2010, 2014; Miguel, Gonçalves & Duarte, 2011; R. Gonçalves, Duarte & Hagemeijer, 2016).¹²

Indeed, “applicative constructions are a means some languages have for structuring clauses which allow the coding of a thematically peripheral argument or adjunct as a core-object argument. Such constructions are signaled by overt verbal morphology” (Peterson, 2007, p. 1); however, Portuguese and Romance languages in general lack dedicated applicative morphology. Besides, whereas most ditransitive structures in Bantu involve monotransitive verbs which need an applicative extension to license the Recipient argument, ditransitive structures with give-type verbs in Portuguese and, in general, in Romance involve semantically ditransitive verbs (Brito, 2008, 2009, 2014; R. Gonçalves, 2016). Additionally, contrary to what happens in Bantu, the word order IO-DO has no bearing on grammatical relations, instead it expresses differences in information structure (for Portuguese: Cépeda & Cyrino, 2020; R. Gonçalves, 2016; R. Gonçalves, Duarte & Hagemeijer, 2016; for Spanish: Beavers & Nishida, 2010; Pineda, 2016).

Arguments like these have been adduced by some authors to propose a distinction between thematic applicative heads (Bantu-type applicatives) and expletive (or raising) applicatives, present in the structure exclusively to Case-license some argument of the verb (e.g., Georgala, 2011; Georgala, Paul & Whitman, 2008) – in the case at stake, the dative argument. Recent accounts of ditransitive structures with core-dative verbs in Romanian make use of this second type of applicative head whilst sustaining that both the Theme and the Recipient are θ-marked by the verb (Cornilescu, 2020; Cornilescu, Dinu & Tigau, 2017).

The analysis we present here shares with the abovementioned accounts of Romanian dative structures the empirically motivated idea that Romance languages (and their varieties) exhibit true ditransitive verbs, the transfer of the Theme to the possession of the Recipient being a result of verb semantics. However, it departs from them in assuming an alternative source for Case-licensing of the dative argument: the functional preposition a ‘to’, which we consider a Case marker, following a long tradition of analyses of Portuguese dative ditransitive structures. This assumption allows us to dispense with the consideration of an expletive applicative node, postulated exclusively for Case-licensing reasons.

We will therefore preserve Larson’s (2014) view of little v as a semantically empty head with EPP and Case features. Also in the spirit of his 2014 account of dative structures, we will put the burden of the extended V projection derivation on both Case and θ-Theory. Indeed, by considering vP an extended projection of V, its structure depends on the selection properties of V. Differently from Larson, however, we will maintain the θ-Criterion, thus rejecting the idea that θ-roles are formal features in need of valuation during the derivation, against Hornstein (1999).

In our account of EP dative structures, the Theme argument is merged in the complement position of the ditransitive verb (see Aoun & Li, 1989; Brito, 2014, 2015, a.o.). Regarding the Recipient argument, the preposition a ‘to’ is a Case marker devoid of any descriptive content, as referred in section 2. Thus, if prepositional Case markers are instances of a K(ase) head (Fukui & Speas, 1986) in these structures, EP, AVPs and other Romance languages will merge a K functional head with the following feature: [uCase: Dat].¹³ The Recipient DP is merged as the complement of K, with (among others) the feature [uCase:?], and the KP projection is merged in Spec, VP. The derivation up to VP of a sentence like A Maria deu um livro ao João ‘Maria gave a book to João’, will therefore have the representation in (42):

At the stage of the derivation shown in (42), the Theme argument of the ditransitive V is not Case-licensed. Thus, v, the locus of accusative Case, probes the Theme DP for Case valuation. The crossing of the Theme DP over the Recipient KP shown in (43) does not cause any violation, assuming, on the one hand, that the movement takes place inside the same phase (the vP phase) and, on the other hand, that the Recipient KP does not intervene because it is no longer an active goal (see Pineda (2016) and references therein).

The structure of the VP presented in (42), reminiscent of the one proposed by Brito’s (2014, 2015) for the V-IO-OD word order in EP, is in line with the current mainstream (applicative) analysis in merging the Recipient DP in a position that c-commands the direct object, without the cost of the projection of an extra applicative head being required to derive core dative structures. The Theme-Recipient word order obtained through the derivation (43) is the unmarked one. As pointed out by several authors, the non-canonical order Recipient-Theme is derived for defocusing or processing related reasons (Beavers & Nishida, 2010; Pineda 2016; R. Gonçalves, 2016; R. Gonçalves, Duarte & Hagemeijer, 2016; Duarte, Gonçalves & Hagemeijer, 2017).

As often argued in the literature, Romance a-DP give-type structures exhibit symmetric c-command with respect to binding and scope (e.g., Giorgi & Longobardi, 1991; Harley, 2002; Pineda, 2016, 2020). Our analysis accounts for the absence of asymmetry effects, assuming that the whole KP is what counts for c-command. In fact, the symmetric relation arises precisely from the co-occurrence of two separate instances of asymmetric c-command: before and after the DO moves up to Spec, vp, for Case reasons.

Finally, the AVPs’ data with prepositions para and em showing clitic counterparts instead of PPs with strong pronouns suggest that such prepositions evolved to syncretic heads, corresponding both to Case markers and to full-fledged Locative and Goal prepositions. Therefore, in ditransitive structures, the head of the KP can be spelled out as a (MOP, STP, AP), para (AP, STP) or em (AP) rendering the microvariation exhibited in AVPs an effect of how externalization operates (Chomsky 2010).

With respect to dative clitic structures, which were shown to be dominant in AVPs, EP does not allow clitic doubling of the type found in Spanish or Romanian, as referred above in section 2.1., hence dative clitics are not expletives. We will therefore adopt the D-hypothesis (see Corver & Delfitto, 1993; Torrego, 1988; Uriagereka, 1995, a.o.), and merge the clitic in Spec, VP, with the relevant feature: [uCase: Dat], a formal translation of the idea that clitics bear inherent Case.¹⁴ D selects for its pro complement, with the relevant feature [uCase: ?], and Case-licenses it, through Agree.

The difference between the a DP structure and the dative clitic structure thus reduces to the difference between the label of the argument merged in Spec, VP – KP in the former, DP in the latter – and to movement of the clitic D head for independent reasons related to its structural, φ-feature (or other type of) deficiency.

At this point, we have presented and discussed ditransitive structures in AVPs which are convergent (despite the spell out differences) with EP, and we have argued that the differences between the latter and the former reduce to the phonological form of the K head.

However, according to data showed in section 3, some speakers of MOP and of STP, which are historically in contact with languages with DOCs, display competition between a grammar converging toward EP and a grammar showing more convergence with the properties of the contact languages.

It was shown that in Forro, the creole language in contact with STP, DOCs are the only sentence pattern available for dative structures. However, only the small group of STP speakers in Group IV (6/64, cf. Table 12) consistently uses a sentence pattern convergent with the one found in Forro. This also encompasses the absence of dative passives. Like Forro, Changana, the contact language of MOP, is a language which exhibits Case syncretism, i.e no morphological difference between accusative and dative is found.

Along the lines of Kayne (1984), it has been suggested that in languages with Case syncretism, the functional head responsible for Case valuation can check more than one Case (for different technical implementations of this idea see, for instance, Baker, 1988; Hiraiwa, 2001; Manzini & Franco, 2016). At the same time, Changana is an asymmetrical language in the sense of Bresnan and Moshi (1990), as shown in Chimbutane (2002), which means that two objects can be Case-licensed by [+V] functional heads in the vP phase, although only the one higher in the Thematic Hierarchy may surface as the primary object and raise further to subject position in passive sentences.

In the light of (42) above, the derivation of structures with a true ditransitive verb would therefore start with a VP-shell where the two internal arguments are merged as DPs, with the Recipient in Spec, VP. V Case-licenses and θ-marks the Theme argument. Next, the complex V-v raises to the upper v head, where it probes the closest DP for Case-licensing, that is, the Recipient DP (MLC, Attract Closest), as shown in (45), a simplified representation of the maximal vP of a sequence like ensinar os outros a Bíblia ‘teach the others the Bible’.¹⁵

In this position, the Recipient argument becomes accessible to further A-movement, rendering the occurrence of dative passives possible. If we are on the right track by claiming that DOCs are the result of language contact in L2 settings of the acquisition of Portuguese, the availability or absence of dative passives in Changana/MOP and Forro/STP, respectively, confirms this hypothesis.

In sum, the dominant DPC sentence pattern AVPs share with EP to express core dative structures corresponds to the same syntactic structure, showing microvariation. However, the substrate-induced DOCs displayed in MOP and STP diverge in a more significant way both from AP and from EP; what is at stake here is what Biberauer and Roberts (2015) would consider a mesoparameter related to structural Case licensing by extended [V] heads.