Possessives, from Franco-Provençal and Occitan Systems to Contact Dialects in Apulia and Calabria

: This article investigates the contact-induced reorganization of the possessive system in the Gallo-Romance dialects spoken from around the 12th century in the villages of Celle and Faeto in North Apulia and Guardia Piemontese in North-West Calabria. Gallo-Romance possessives exclude the article in the prenominal position, whereas in the Southern Italian dialects, possessives follow the noun preceded by the deﬁnite article. This original contrast is no longer visible in the varieties of Celle, Faeto and Guardia which changed the original prenominal position to the postnominal position combining with the article, except with kinship terms, preserving the original prenominal position. At the heart of contact phenomena, there are bilingualism and transfer mechanisms between the languages included in the complex knowledge of the speaker, suggesting a test bed for the treatment of language variation and parameterization. We propose an account of morpho-syntactic and interpretive properties of possessives, making use of the insights from the comparison of contact systems with prenominal (Franco-Provençal and Occitan varieties) and postnominal (Southern Italian dialects) possessives. The ﬁnal part examines the distribution of possessives, tracing it back to the deﬁniteness properties of DP and proposes a phasal treatment based on syntactic and interpretive constraints.


Possessives in Franco-Provençal and Occitan Varieties
This article deals with a contact phenomenon showing up in the Franco-Provençal spoken in North-Apulian villages of Celle and Faeto 1 and in the Occitan-type dialect spoken in the North-Calabrian dialect of Guardia Piemontese 2 . In both cases, they are settlements formed around the 12th-13th century by alloglot communities. As for Franco-Provençal, its place of origin is the South-Eastern French regions, where South-French varieties are spoken; the variety of Guardia has been connected with the Occitan spoken in South-Western Piedmontese valleys (Valle Pellice). We will investigate the syntactic reorganization of possessives in these dialects, which change from the prenominal nonarticulated form typical of Franco-Provençal and Occitan varieties into the postnominal articulated type corresponding to the possessive system of Southern Italian dialect 3 .
Our first step in the following sections is to illustrate the distribution of possessives in Gallo-Romance varieties in comparison with the Southern Italian varieties. Apulian Franco-Provençal (AFP) and Calabrian Occitan (ClO) follow the distribution of Southern Italian dialects except with singular kinship terms, the only contexts that preserve possessive-noun order, apparently in contrast with Southern Italian rules. The main differences are schematized in (1), where (+) refers to kinship terms contexts: (1) possessive prenominal postnominal D required D excluded Southern Italian dialects -+ + Franco-Provençal/Occitan + -+ Southern Italy Gallo-Romance (+) + + (+) Actually, also in the case of kinship terms, the Gallo-Romance dialects of Apulia and Calabria show a special treatment compared to that of Southern systems, as the latter too differentiate the possession of kinship referents by using the enclitic occurrence of possessives. The crucial point is that two languages, historically connected with two distinct Romance linguistic groups, Franco-Provençal and (Piedmontese) Occitan, in an ancient and prolonged contact relation with the Southern Italian dialects, undergo a strictly parallel grammatical evolution.
We note that even in Standard Italian, singular kinship terms combine with prenominal possessives excluding the article (cf. discussion in Section 5.1). A possible influence on the part of these structures cannot be totally excluded in theory. However, taking into account the history of Italian, which became the common-use language only in 20th hundred, we should conclude that the special behavior of kinship terms is a secondary and recent reconstruction compared to the overall change in the system of possessives. For this reason we think that the treatment of kinship terms still reflects an ancient reorganization.
After describing the Franco-Provençal and Occitan systems in (1) and the Southern Italian systems in (2), possessives of the Faeto and Celle dialects in comparison with the system of Guardia will be illustrated in (3). The theoretical and descriptive approach is presented in (4), where a representational lexicalist approach to the morpho-syntax of possessives is proposed, and a discussion of the contact mechanism is outlined in Section 6. In Section 7, the grammar of possessives is related to the morpho-syntactic parameterization and phasal array of sentence.
In Franco-Provençal varieties, possessives occur in the prenominal position and exclude the determiner both in DPs and in predicative contexts, apart from the pronominal use. This distribution is illustrated by the data of some Franco-Provençal varieties spoken in Aosta Valley, Sarre in (2), and in Western Piedmont, Coazze (Susa Valley) in (3). The data regarding Southern French varieties (Forez, from Escoffier 2002) presented in (4) attest the generalized occurrence of the prenominal possessive and the absence of the article. As shown by the data, at least in some dialects, the occurrence of the article inside DPs is not totally excluded, but it may variably occur as in Sarre, as illustrated in (2c) 4 . The examples in (a) refer to the combination possessive-common noun, while the data in (b) to kinship terms.
(2) a.  (Escoffier 2002, v. 6) (5a) for Sarre, (6a) for Coazze illustrate the predicative occurrence of simple possessives. The Sarre dialect utilizes the combination d@+personal pronoun, whereas the Coazze dialect presents the possessive form without the article. All of the varieties show the pronominal occurrence of the possessive preceded by the article, both in copular and in argumental positions as in (5b,c)-(6b,c); (7) shows pronominal forms of the varieties spoken in Forez (Escoffier 2002) where pronominal alternants are preceded by the article. 5 Sarre's data is based on the personal communication of Professor Stefania Roullet (Roullet 2020).
(5) a. l e d@ me/d@ te /d@ ly ClS is of me/of you /of his 'it is mine/yours/his' b. l e l-o m@-N /t@-N ClS is the-MSG 1PSG-MSG /2PSG-MSG 'it is mine/yours' c.
baLL@-me l-o m@-N /l-a mi-a give-me the-MSG 1PSG-MSG /the-FSG-FSG 'give me mine' Sarre (6) a. a l E l-u mE-l /l-I miø-i /l-a mje ClS is the-MSG 1PSG-MSG /the-MSG 1PSG-MPL /the-FSG 1PSG-FSG 'it is mine/they are mine' c.
lo mino the mine (Escoffier 2002, v. 6) In predicative contexts of the Coazze dialect in (6b), the combination D+possessive is introduced by the 3PSG of 'be' and an expletive subject clitic a l, occurring in impersonal and post-verbal subject constructions (cf. Manzini and Savoia 2005).
A similar distribution also characterizes Occitan dialects 6 , such as the Pomaretto dialect (Val Germanasca, Western Piedmont). (8a) illustrates possessives inside DPs and (8b) kinship terms. In predicative contexts, possessives may occur without the article, in (9a), or preceded by the article, in (9a'). (9b) exemplifies the pronominal reading. We recall that in the Occitan dialects, the final unstressed lexical -a changes to [O], according to the contextual conditions, possibly slightly differently from dialect to dialect.

Pomaretto
We note that predicative contexts where the possessive is preceded by the article, in (6b) for Coazze and (9a') for Pomaretto, display the partial agreement characterizing constructions with post-verbal 3rd-person subject, such as a j øt amny li mei'na 'The boys came, lit. ClS Loc has come the boys' (Coazze), where the verb is in the 3rd singular and the ClS is the expletive al. Analogously, in the Pomaretto dialect, the verb is in the 3rd singular and the ClS has the non-agreeing form l. This type of syntax with postverbal subjects is largely documented in the Northern Italian dialects (cf. Manzini and Savoia 2005, § 2.8). According to cf. Manzini and Savoia this special syntax is systematically associated with the focalized reading of the post-verbal subject; in other words, the partial agreement makes the structure operator-variable underlying focalized clauses manifest, in which the 3rd singular ClS/verbal agreement lexicalizes a partial specification, in turn being completed for number and gender by the postverbal subject. As known, 1st and 2nd subjects, inherently deictic, do not allow this syntactic organization. Coming back to the examples of possessive constructions of Coazze and Pomaretto, we conclude that when possessives in predicative contexts are preceded by the article, they behave as post-verbal focalized nominal elements, substantially as the syntactic subjects.
Summing up, the data presented above share two crucial properties: In DPs, possessives occur in the prenominal position, and do not require the article. With kinship terms, the definite article is excluded.

Possessives in Southern Italian Dialects
In the Southern Italian dialects, possessives are postnominal and require the article, both inside DPs, where the article precedes the noun, and in predicative or pronominal occurrences (Rohlfs [1949(Rohlfs [ ] 1968Ledgeway 2009 for Old Neapolitan). Additionally, some Southern Italian dialects obligatorily or optionally introduce the preposition d@ 'of' (a sort of linker) before the article both in predicative contexts and inside DPs . Here, Apulian varieties are exemplified for Gravina in (10) and Castelluccio Vallemaggiore in (11), the municipality bordering the territory of Celle and Faeto. Hence, it provides the contact linguistic context of the Franco-Provençal spoken in the two villages. (10a)-(11a) illustrate the postnominal position of possessives; the occurrence of the article before the possessive in predicative and pronominal contexts is shown in (10c)-(11c). With kinship terms, possessives occur as enclitics on the noun, excluding the article, in (10b)-(11b). Enclitics typically characterize singular referents. (20d) illustrates the structure where the introducer di precedes the postnominal possessive in indefinite DPs.
l-a kammis-a ma ( In the North Calabrian dialects (here Cervicati), possessives follow the noun and are preceded by the prepositional introducer di followed by the definite article, both inside DP and in predicative contexts, agreeing with the possessum (cf. Rohlfs [1949] 1968; Baldi and Savoia 2019, forthcoming). Within DPs, in (12a), the prepositional introducer followed by the definite article is usually lexicalized. This structure also characterizes contexts with an indefinite article in (12a'). (12b) illustrates the enclitic occurrence of the possessive with singular kinship terms, contrasting with plurals excluding the enclitic form, as in (12b'). In all cases, the article and the possessive agree with the head noun.

Cervicati
In Southern Italian varieties, the article is necessary for lexicalizing the interpretation of the possessum. Agreeing properties of the possessive element, although endowed with referential force, are doubled by the article in fixing the reference to the possessed argument of possession relation. This is confirmed by the fact that the definite article of the possessive also occurs within indefinite DPs in dialects where it is preceded by the preposition di as exemplified in (12a'); in other words, the definite article also occurs in DPs introduced by an indefinite quantifier and in a predicative structure. This suggests that possessive structures require possessive elements to combine with the independent lexicalization of nominal agreement features. Moreover, in many dialects, the prepositional introducer completes the possessive structure by lexicalizing the inclusion relation independently of the lexical content of the possessive. The result is that the structure introducer-definite articlepossessive lexicalizes part-whole relation and definiteness both by means of independent morphological elements and the possessive pronoun.
In Southern Italian varieties, possessives follow the noun preceded by the definite article; kinship terms require enclitic possessive forms. Possessives require the article in all contexts, predicative and pronominal. Many Southern Italian varieties also present a possessive structure ,where the possessive element is introduced by the preposition d@ 'of', both within DPs and in predicative contexts.

Possessives of the Celle and Faeto Dialects in Comparison with the Guardia System
The data from Celle (and Faeto) 7 in (14a) illustrate the possessives in the postnominal positions with common nouns (14b), the kinship terms with singular possessors and (14b') plural possessors, with 1st-, 2nd-, 3rd-person forms in the postnominal position, except the 2nd-person form of respect. In (14c), the predicative contexts are shown where the possessive is lexicalized as a pronominal element preceded by the article. 3PL possessives are lexicalized by the same element as the singular; alternatively, we find the postnominal form laur@ and the article. 1PL and 2PL possessives are postnominal not@/vot@, in turn combining with the article. The 2P is also lexicalized by the specialized alternant vutuN, vuta, vutO, vut@, which precedes the noun and excludes the article, as in (14b"); it occurs as a respective form regarding a singular interlocutor. 8 Prenominal suN can combine with postnominal not@/laur@, as in (14c), in kinship terms contexts. (14d) exemplifies the contexts where the noun is introduced by an indefinite element with a postnominal possessive. (14e) and (14e') illustrate the pronominal constructions D+possessive, in predicative and argumental contexts. In (14f), possessive constructions with partitive introducer of the type considered in (12) and (13)

Celle
An interesting fact is that the same redistribution of possessives also characterizes the Occitan dialect of Guardia, in (15), in turn in contact with the North-Calabrian dialects displaying a Southern Italian-type system. We find the postnominal position in DPs, with the definite determiner preceding the noun, in (15a), prenominal article-less possessives with kinship terms, in (15b). In predicative contexts in (15c), and in pronominal occurrences in (15d), possessives are combined with the article.

Guardia Piemontese
As shown by the data we have reviewed in (14), contact with the Apulian dialects has affected the possessive system of the Apulian Franco-Provençal; a similar redistribution is shown by the Guardia dialect in (15). In the Southern Italian systems, including the neighboring North Apulian and Calabrian dialects, possessives follow the noun, in turn preceded by the definite article, which precedes the possessive element also in predicative/pronominal occurrences. Enclitic 1st-/2nd-/3rd-person forms characterize kinship terms. By contrast, Franco-Provençal and Occitan place possessives before the noun exclud-ing definite article. In Apulian Franco-Provençal possessives have the same distribution of the Southern Italian dialects both inside the DP sand in predicative contexts. However, kinship terms preserve the prenominal possessive.
Summarizing, we get the distribution in the The comparison with the distribution of original varieties highlights some crucial differences: Contact has changed the distribution of possessives, favoring the postnominal position and the occurrence of the definite introducer in DPs and in predicative contexts Kinship terms retain the prenominal occurrence of singular possessors, selecting specialized forms, while plural possessors are expressed by postnominal forms, substantially, such as in the Southern Italian dialects. The structure introduced by the preposition d@ 'of' appears in DPs, in (14f) and (11d), (12) and (13). It is unknown to Gallo-Romance languages.

Possessives: A Morpho-Syntactic Approach
Addressing the analysis of possessives means, first of all, investigating the relationship between the possessum and the possessor in its morpho-syntactic manifestation. The crucial point is the nature of possession relation, that, as well known, entails a common set of properties underlying genitive, dative and locative Manzini et al. forthcoming;Savoia et al. 2020). By way of example, the oblique can encompass dative and genitive, as in some of Latin declensions, or in Albanian and in Romanian where the indirect case lexicalizes both the stative possession (genitive) and the eventive context possession (dative). This type of syncretism is illustrated by the Italo-Albanian examples in (17) from San Costantino Albanese (Basilicata, Italy), where Lkr (Linker) designates the prenominal article occurring in genitive contexts. Genitive in (17a) and dative in (17b) are characterized by the same oblique inflection, here -i-t.
j-a ðE burr-i-t to.him-it I.gave man-OBL-DEF 'I gave it to the man'

San Costantino
Moreover, in many languages, the same oblique morphology is selected for locative contexts. The key idea is that in all of these contexts, the conceptual property of 'inclusion' is involved, in the sense initially discussed in Manzini and Savoia (2011a, 2011b whereby all types of possession, including inalienable and psych state possession, fall under the same basic relation. This proposal resumes the analysis of possession in Belvin and Den Dikken (1997, p. 170), according to whom 'entities have various zones associated with them, such that an object or eventuality may be included in a zone associated with an entity without being physically contained in that entity . . . The type of zones which may be associated with an entity will vary with the entity'.
A complementary question concerns the relation between cases and prepositions. Specifically, Fillmore (1968) treats cases as the inflectional equivalent of prepositions, so that the elementary introducers such as Italian di/a or English of/to would be equivalent to genitive and dative, respectively.
According to Manzini et al. (2019, forthcoming), Savoia et al. (2020), an oblique case, on par with a preposition, is a predicate introducing a relation between the argument it selects and another argument. This means that we assign a relational content to cases, with the effect that the oblique case or prepositions are endowed with interpretable properties. This solution contrasts with the more traditional view that prepositions such as 'of' or 'to' are devoid of interpretive content, or, in minimalist terms, uninterpretable. Our conclusion is that oblique case and prepositions such as 'of' or 'to' can be thought of as elementary partwhole relators; in other words, a single property, namely inclusion/superset of, formalized as [⊆], is associated with the conceptual cluster underlying oblique and/or of/to-like prepositions. Let us consider the Italian examples in (18a,b).
il libro di Gianni the book of Gianni b.
Ho dato il libro a Gianni I.have given the book to Gianni Possession relation in (18a) can be understood as the lexicalization of an inclusion relation, in (19a). The idea is that the dative in (18b) is the possessor of the argument 'libro', essentially such as in the genitive, although, in this case, the predictive relation between possessor and the possessum is introduced by an agentive v, as in (19b) Dative may be interpreted as possession relation introduced by an agentive predicate, where, again, the argument surfacing ad the object of give is the external argument of the inclusion relation lexicalized by the dative preposition a. The semantic proximity between the Romance prepositions di and a is manifested by the possessive use of a (for French cf. Kayne 1977;Cardinaletti 1998), for instance in some Southern Italian dialects, where a can introduce the possessor, as in (20)

Castelluccio
If we are on the right track, the similarity between partitive and genitive interpretation is reflected by the syncretism on di/of. Hence, we are induced to conclude that of /oblique case are the externalization of the elementary predicate connecting two arguments (possessor and possessum) in a part-whole relation, ad in (21): The lexical content in (21) allows us to account for the syncretism, very often encountered in languages between dative and genitive, whereby the same preposition or case expresses different types of part-whole relation.
If we think that possessive pronouns in turn involve the structural analysis in (21), we need to characterize the relation between syntax and the internal structure of nominal forms more precisely. The internal structure and, more generally, the inflectional properties of the noun can be related to categories such as gender, number and inflectional class. In the literature (Picallo 2008; Fassi Fehri 2015 on Arabic; Kramer 2015) at least two functional projections are required-roughly gender and number. Following Savoia 2014, 2018) and , the internal organization of the noun includes a category-less lexical root √ (Marantz 1997), that, in keeping with Higginbotham (1985), is understood as a predicate. The root merges with the inflectional elements endowed with interpretive content (gender, number, etc.), that restrict the properties associated to the argument x open at the predicate. Thus, the inflectional structure is based on the same computational mechanisms underlying syntax 9 . Inflected nouns are the result of a Merge operation that combines a lexical root with Class (gender feminine/masculine) and other classificatory properties, including number and case, that contribute to specifying the argument introduced by the lexical root. This model is illustrated in (7) for Albanian burrit 'of the man' in (1a), where number and case specifications are added to Class. According to Savoia 2011a, 2011b), the morpheme i, occurring as a masculine singular oblique and nominative, is tentatively characterized as a definiteness exponent lexicalizing the relation [⊆] between a singleton and a set including it, in the sense of the analysis of determiners in Chierchia (2010). In (22), the label OBL is used.
Languages 2021, 6, x FOR PEER REVIEW 14 of 32 We can bring back to this framework the analysis of possessive pronouns, which in many languages are expressed by means of genitival forms, as, for instance, English his/her(s). In other words, we may assume that possessives in turn implement the inclusion relation between the possessum, the head noun, and the possessor introduced by the possessive element. Based on this model, possessives lexicalize an oblique reading specified as the part-whole relation ⊆ (Baldi and Savoia 2019, forthcoming), as in the structure in (23).
The idea that possessives conceal a complex structure is proposed by Cardinaletti (1998), where possessives preceded by the article, in Italian and in French, are treated as elliptical forms including a weak possessive and an empty noun. Not substantially different is the analysis in Kayne (2018, § 3) whereby possessives include a silent personal pronoun, so that by way of an example, ho letto il suo libro 'I have read his/her book' corresponds to the representation in (24).
(24) ho letto il DP(POSS) suo libro The conceptualization of possession relation as being a relation introduced by an operator/predicate with two arguments is formalized in Cornilescu and Nicolae (2011) by assuming a prepositional small clause of which the possessum and the possessor are the arguments. The small clause is headed by the abstract predicate [BELONG], lexicalized by an overt preposition or the morphology of case. We ascribe the morpho-syntactic behavior of possessives to their interpretive nature as far as they lexicalize the inclusion relation between the possessum, the head noun, and the possessor lexicalized by the possessive element.
We can bring back to this framework the analysis of possessive pronouns, which in many languages are expressed by means of genitival forms, as, for instance, English his/her(s). In other words, we may assume that possessives in turn implement the inclusion relation between the possessum, the head noun, and the possessor introduced by the possessive element. Based on this model, possessives lexicalize an oblique reading specified as the part-whole relation ⊆ (Baldi and Savoia 2019, forthcoming), as in the structure in (23).
Languages 2021, 6, x FOR PEER REVIEW 14 of 32 We can bring back to this framework the analysis of possessive pronouns, which in many languages are expressed by means of genitival forms, as, for instance, English his/her(s). In other words, we may assume that possessives in turn implement the inclusion relation between the possessum, the head noun, and the possessor introduced by the possessive element. Based on this model, possessives lexicalize an oblique reading specified as the part-whole relation ⊆ (Baldi and Savoia 2019, forthcoming), as in the structure in (23).
The idea that possessives conceal a complex structure is proposed by Cardinaletti (1998), where possessives preceded by the article, in Italian and in French, are treated as elliptical forms including a weak possessive and an empty noun. Not substantially different is the analysis in Kayne (2018, § 3) whereby possessives include a silent personal pronoun, so that by way of an example, ho letto il suo libro 'I have read his/her book' corresponds to the representation in (24).
(24) ho letto il DP(POSS) suo libro The conceptualization of possession relation as being a relation introduced by an operator/predicate with two arguments is formalized in Cornilescu and Nicolae (2011) by assuming a prepositional small clause of which the possessum and the possessor are the arguments. The small clause is headed by the abstract predicate [BELONG], lexicalized by an overt preposition or the morphology of case. We ascribe the morpho-syntactic behavior of possessives to their interpretive nature as far as they lexicalize the inclusion relation between the possessum, the head noun, and the possessor lexicalized by the possessive element.
Returning briefly to the data in (12) and (13) relating to the Calabrian dialect of Cervicati, we note that the oblique form of 1st-and 2nd-person pronouns coincides with The idea that possessives conceal a complex structure is proposed by Cardinaletti (1998), where possessives preceded by the article, in Italian and in French, are treated as elliptical forms including a weak possessive and an empty noun. Not substantially different is the analysis in Kayne (2018, § 3) whereby possessives include a silent personal pronoun, so that by way of an example, ho letto il suo libro 'I have read his/her book' corresponds to the representation in (24).
(24) ho letto il DP(POSS) suo libro The conceptualization of possession relation as being a relation introduced by an operator/predicate with two arguments is formalized in Cornilescu and Nicolae (2011) by assuming a prepositional small clause of which the possessum and the possessor are the arguments. The small clause is headed by the abstract predicate [BELONG], lexicalized by an overt preposition or the morphology of case. We ascribe the morpho-syntactic behavior of possessives to their interpretive nature as far as they lexicalize the inclusion relation between the possessum, the head noun, and the possessor lexicalized by the possessive element.
Returning briefly to the data in (12) and (13) relating to the Calabrian dialect of Cervicati, we note that the oblique form of 1st-and 2nd-person pronouns coincides with the possessive elements, as exemplified in (25a)

Cervicati
Following the observation of an anonymous reviewer, these forms can be treated as evidence in favor of the interpretive coincidence of dative and genitive (cf. Franco et al. 2021). The perfect coincidence with the possessive elements (see (12) and (13)) supports an analysis of mi-a and ti-a as the morpho-phonological realization of the inclusion relation in (21). As tentatively suggested in (25b), the lexical base introduces the possessor, while the possessum is identified by the agreement inflection, in these dialects corresponding to the feminine/plural exponent -a (Manzini and Savoia 2017).
In the framework we adopt, all lexical material, including inflections, is associated with interpretable contents; this proposal is not substantially different from the conception of Agree in Chomsky (2001Chomsky ( , 2005 insofar as it expresses the identity between features under locality (Minimal Search) Savoia 2005, 2018). As a consequence, there is no Agree rule triggered by the need for a probe to interpret/value its features and, more basically, our model excludes uninterpretable features and probe goal-induced movement (see Chomsky et al. 2019). Agreement works by creating an identity relation between two or more referential feature sets lexicalizing the same argument Savoia et al. 2019) and all lexical material is interpreted at the Conceptual-Intentional (C-I) interface.

The Cartographic Analysis: Surface Syntax Photographs Meaning
In syntactic literature, different orders between the same elements have been traditionally traced back to movement, generally triggered by a probe. So, possessives originate in a low position inside the NP from where they may move to higher positions. Consistently with the idea that prenominal and postnominal (or thematic) position of possessives can be 'transformationally related', as initially proposed in Kayne (1977), Cardinaletti (1998, p. 18) assumes that the same basic structure underlies pre-and postnominal possessives, as in (26).
If the base position of possessives is inside NP, as in (26), the postnominal order is derived by the movement of N to a functional head 'between N and D' (Cinque 1994). The prenominal order is obtained by moving in turn the possessive element to the inflectional domain of the DP, as in (26').
The idea of Cardinaletti (1998) is that the possessives in the prenominal and postnominal positions are not the same lexical elements, but correspond to two different alternants. More precisely, the postnominal position would host focalized forms endowed with complete referential features (agreement inflection), differently from prenominal possessives, lacking such interpretive properties. This distinction is treated in terms of the difference between deficient and strong elements (Cardinaletti and Starke 1994), separating strong and deficient pronouns. Deficient elements are devoid of the XP functional layer, including phi-features specification and case features. As a result, deficient elements must be licensed by a functional head. Extending this analysis to possessives, Cardinaletti (1998: 20) proposes that prenominal possessives are devoid of the functional layer so that they need to be licensed in a functional domain, namely the domain of the Agreement head (AGRS), of the DP.
This proposal would seem to fit in with the morphological differences between prenominal and postnominal possessives. Indeed, there are languages in which prenominal possessives have a reduced morphology, devoid of agreement inflection, while postnominal elements present it, such as in Spanish mi libro 'my book' vs. el libro mio 'the book (of) mine'. According to Cardinaletti, in Spanish, a prenominal possessive is a clitic, a reduced weak form, that can adjoin to D, excluding the determiner. In a language such as Italian, where there is no morphological difference between pre-and postnominal possessives, Cardinaletti (1998) concludes that the syntactic behavior of prenominal forms and the fact that they combine with definite articles suggest that they are deficient/weak forms. French prenominal possessives are in turn deficient clitic elements, showing the same type of the nominal inflection of clitic pronouns. Italian postnominal and predicative possessives would be strong, differently from French, devoid of strong possessives, i.e., of simple possessives in copular contexts. As for pronominal forms preceded by the article, the idea of the Author is that they, in Italian as well as in French, are elliptical forms including a weak possessive and an empty noun.
So, cartographic analyses assume a low structural position for possessives and derive their surface position by means of a movement rule induced by a licensing mechanism. A common point also concerns the defective nature of possessives inside the DP. However, it is evident that an analysis based on hypothetical categories, as an empty noun, or on the translation of C-I interpretive-level properties into the structural organization is very costly. The same is true for separating homophonous forms by assigning them different structural representations, as in the case of the pre-vs. postnominal occurrences of Italian possessives. Generally speaking, we may wonder on what parametric or principled basis languages should conceal by syncretism or other morphological means what they intend to mean.
What we intend is that a treatment such as that of Cardinaletti leads to costly and ad hoc proposals, such as, for instance, distinguishing two different lexical entries, weak, i.e., defective, and strong, for Italian possessives depending on their position. A solution that is adopted even in the presence of the same form and lacking clear morphological clues. In other words, the distinction between weak and strong possessives, in order to be maintained, requires ad hoc constraints insofar as morphological properties do not support it (see Section 6). Our idea is that possessives share a semantic core independently from their form and position, so that variation can be addressed in a simpler and more adequate way based on the mechanism of externalization.
Possessives introduce the participants in the speech act (1st/2nd person) or the definite reference to a discourse anaphoric argument (3rd person). In many languages, this interpretive content is able to exhaustively externalize the referential domain of DP. In other languages, the article is nevertheless necessary, such as in the Italian and Northern Italian dialects. This does not seem to be crucially connected to the position, prenominal or postnominal, of the possessive.

Lexicalist Framework
Descriptively, the following syntactic alternatives emerge, i.e., presence/absence of article, pre-/postnominal occurrence and, not in languages examined here, a possible morphological difference between complete/reduced inflection. Our idea is that the distribution of possessives may be explained as due to their inherent interpretive properties.
Indeed, these differences 10 involve the ability of possessives to encode the referential properties involved in the externalization of DPs and different scope domains. In fact, possessives introduce the deictic link to the participants in the speech act (1st/2nd person) or the usual definiteness specifications anaphoric to discourse in the 3rd person, so that, in many languages, this interpretive content is able to subsume D. In many languages, the article is nevertheless necessary, such as in the Italian and Northern Italian dialects, independently of the presence of possessives in the prenominal or postnominal positions.
Taking into account the representation in (21), we may suppose that the sequence D-possessive-agreement inflection externalizes the argumental structure of inclusion relation, whereby inflection, the clitic subject and the determiner identify the possessum, i.e., the external argument of the predicative relation with the possessor. As seen in Section 1, we assign interpretable content to the lexical material, including inflectional formatives. In languages such as Franco-Provençal and Occitan, the possessive element in the prenominal position satisfies the definiteness properties associated with the D head domain of DP, excluding the article, which in Romance varieties is generally the specialized exponent of the deictic/referential content of the noun . Thus, in those varieties, possessives lexicalize these properties on their own, as in (27). properties. Indeed, these differences 10 involve the ability of possessives to encode the referential properties involved in the externalization of DPs and different scope domains. In fact, possessives introduce the deictic link to the participants in the speech act (1st/2nd person) or the usual definiteness specifications anaphoric to discourse in the 3rd person, so that, in many languages, this interpretive content is able to subsume D. In many languages, the article is nevertheless necessary, such as in the Italian and Northern Italian dialects, independently of the presence of possessives in the prenominal or postnominal positions.
Taking into account the representation in (21), we may suppose that the sequence Dpossessive-agreement inflection externalizes the argumental structure of inclusion relation, whereby inflection, the clitic subject and the determiner identify the possessum, i.e., the external argument of the predicative relation with the possessor. As seen in Section 1, we assign interpretable content to the lexical material, including inflectional formatives. In languages such as Franco-Provençal and Occitan, the possessive element in the prenominal position satisfies the definiteness properties associated with the D head domain of DP, excluding the article, which in Romance varieties is generally the specialized exponent of the deictic/referential content of the noun ). Thus, in those varieties, possessives lexicalize these properties on their own, as in (27).

Coazze
In the Southern Italian dialects, possessives occur postnominally and combine with the prenominal article, as in (28) 10 Indeed, as an Italian native speaker, it seems to me that some evidence concerning the difference between pre-and post-nominal Italian possessives is very questionable, and possibly connectable to differences in the interpretive role of the possessive element, independently of structural aspects. In fact, as suggested by an anonymous reviewer, postnominal possessives in Standard Italian, such as in il libro mio 'my book, lit. the book of mine', can be associated with focus interpretation. As for 'loro', see .

Coazze
In the Southern Italian dialects, possessives occur postnominally and combine with the prenominal article, as in (28) Languages 2021, 6, x FOR PEER REVIEW 18 of 32

Castelluccio
As well known in the literature, kinship nouns give rise to specialized constructions in many languages. An example is provided by the Southern Italian dialects, in which kinship terms select enclitic possessives excluding the definite article. In other varieties, kinship terms exclude the article when combining with possessives. This is, in fact, the case of the Gallo-Romance dialects that we are examining, as well as of Standard Italian, at least for a subset of these terms 11 . Syntactic solutions proposed in the generative literature connect the lack of an article to the high position inside DP, substantially in D (Longobardi 1995(Longobardi , 1996. This specifically applies to contexts where the kinship noun rules out or precedes the possessive, as in the case of the terms for mother/father in some varieties, or, more generally, in the case of enclisis. According to Longobardi (1996), cases Castelluccio 10 Indeed, as an Italian native speaker, it seems to me that some evidence concerning the difference between pre-and post-nominal Italian possessives is very questionable, and possibly connectable to differences in the interpretive role of the possessive element, independently of structural aspects. In fact, as suggested by an anonymous reviewer, postnominal possessives in Standard Italian, such as in il libro mio 'my book, lit. the book of mine', can be associated with focus interpretation. As for 'loro', see Manzini (2014).
As well known in the literature, kinship nouns give rise to specialized constructions in many languages. An example is provided by the Southern Italian dialects, in which kinship terms select enclitic possessives excluding the definite article. In other varieties, kinship terms exclude the article when combining with possessives. This is, in fact, the case of the Gallo-Romance dialects that we are examining, as well as of Standard Italian, at least for a subset of these terms 11 . Syntactic solutions proposed in the generative literature connect the lack of an article to the high position inside DP, substantially in D (Longobardi 1995(Longobardi , 1996. This specifically applies to contexts where the kinship noun rules out or precedes the possessive, as in the case of the terms for mother/father in some varieties, or, more generally, in the case of enclisis. According to Longobardi (1996), cases such as casa mia 'my house, lit. house my' or mamma mia 'my mother, lit. mummy' require that the noun moves to D, excluding the article. Cardinaletti (1998), rather, associates the possessive with D, whereas the movement of the kinship term to D is assumed in correspondence of enclisis.
In all these of proposals, this class of nouns favors a different distribution within the DP. Independently of the formal machinery, the crucial idea is that in these languages, 'the lexical content of a kinship noun is sufficient to specify the reference to an individual, hence subsuming Definiteness properties, which with the other classes of nouns are lexicalized by the article' (Manzini and Savoia 2005, p. 721). We may think of the kinship terms as being relational in nature (Croft 1991;Dobrovie-Sorin 2013;Giusti 2016) and implying an inherent internal argument (a sort of possessor). This lexical property accounts for the special syntactic status of kinship terms as capable of exhaustively lexicalizing the referential properties of the noun. In dialects such as the Cervicati dialect, the possessive clitic has a specialized inflectional structure, as in frat-i-tt-a 'your brother' in (29) (cf. (12b)), the possessive takes the ending -a. We can treat -a as an exponent associated with a rich set of referential properties, as discussed in Manzini et al. 2020), assigning the structure in (29) to the combination noun+clitic, where the undersigned y and x indicate the two arguments associated with the kinship term. In (29), the noun frat-i 'brother' is able to realize the definiteness properties of the DP by virtue of its inherent referential properties. This explains why enclisis is usually restricted to singular nouns, as far as plural forms are not able to satisfy the required definiteness and specificity properties. 12 As a result, the order of elements linearizes the inclusion predication, where frat-i is the external argument, 'a part of your zonal domain', and ta the internal argument.
Possessives introduce the participants in the speech act (1st/2nd person) or the definite reference to a discourse anaphoric argument (3rd person). In many languages, this interpretive content is able to exhaustively externalize the referential domain of DP. In other languages, the article is nevertheless necessary, such as in the Italian and Northern Italian dialects. This does not seem to be crucially connected to the position, prenominal or postnominal, of the possessive.

Contact Effects
The preservation of the alloglot language for several centuries in a situation of contact with different morpho-syntactic, phonological and lexical systems involves code switching and mixing processes and the production of mixed sentences and borrowings Cervicati In (29), the noun frat-i 'brother' is able to realize the definiteness properties of the DP by virtue of its inherent referential properties. This explains why enclisis is usually restricted to singular nouns, as far as plural forms are not able to satisfy the required definiteness and specificity properties. 12 As a result, the order of elements linearizes the inclusion predication, where frat-i is the external argument, 'a part of your zonal domain', and ta the internal argument. 11 In Standard Italian, the article is generally required with a subset of kinship terms, for instance figliuolo, figliuola 'son, daughter', nonno, nonna ' grandfather/ grandmother', mamma 'mum', babbo/ papà 'dad', that Serianni (2000) characterizes as affective forms. However, there is a lot of variation depending on the regional Italian. 12 An anonymous reviewer raises the question of why "i miei genitori' [my parents], which are highly definite and specific, do not pattern with singular nouns'. We can think that it is specifically the reference to a plurality of individuals, the collective-like nature of 'genitori', that blocks the precise reference implied by the special behaviuor of singular kinship terms. It is interesting to note that the forms for 'mother' and 'father' in many dialects and regional Italian varieties can occur without an article, 'mamma' in the sense of 'my/your mother', or with a postnominal possessive (Longobardi 1995(Longobardi , 1996. In the Celle dialect, as an example, mamm@ 'mummy' is sufficient, while 'my parents' includes the article and the postnominal possessive, lO ddZ@n@ttaw@ miNN@ 'the parents of mine'. Possessives introduce the participants in the speech act (1st/2nd person) or the definite reference to a discourse anaphoric argument (3rd person). In many languages, this interpretive content is able to exhaustively externalize the referential domain of DP. In other languages, the article is nevertheless necessary, such as in the Italian and Northern Italian dialects. This does not seem to be crucially connected to the position, prenominal or postnominal, of the possessive.

Contact Effects
The preservation of the alloglot language for several centuries in a situation of contact with different morpho-syntactic, phonological and lexical systems involves code switching and mixing processes and the production of mixed sentences and borrowings (Bakker and Muysken 1994). Needless to say, Apulian Franco-Provençal and Calabrian Occitan show a wide range of lexical bases of Apulian (Melillo 1959) or Calabrian origin together with cases of syntactic hybridization or reorganization. According to Thomason (2010), an intense contact situation can also induce changes in grammar components such as phonology, morphological paradigm and syntax, altering the typological organization of the language. In these cases, convergence may also make morpho-syntactic patterns of the languages in contact more similar, as discussed in (Matras 2010).
In the literature, the acquisition of loans is connected with functional generalizations, whereby the tendency to prefer nouns is related by Authors to the wider autonomy that nouns have in the discourse (Romaine 2000). On the contrary, verbs need to be integrated in the morpho-syntactic system of the host language. Another generalization concerns the fact that loan processes and interference would tend to spare the nuclear lexicon-nouns denoting body parts, numbers, personal pronouns, conjunctions, etc. (Romaine 2000, Muysken 2000)-revealing the crucial role played by genetically determined conceptual primitives. Lexical borrowing and contact exclude non-natural results but operate in compliance with the constraints inherent to language as a specialized system of knowledge and its interface systems.
There is a conceptual link between interaction phenomena in multilingual competence and transfer (Verschik 2017). Code switching and mixing, borrowing and re-coding of semantic or morpho-syntactic properties of a language stem from multilingual minds, as the effect of the intertwining of grammars in the competence of speakers. Contact morphosyntactic reorganization is induced by the transfer processes characterizing bi/multilingual knowledge: in these communities a certain number or majority of the speakers have been bilingual for many centuries, thus manifesting the usual outcomes of transfer and language mixing. In all cases, we expect that changing L2 during the acquisition process is not conceptually different from changing LX in social or cultural contact contexts (Foote 2009;Thomason 2001Thomason , 2010. In keeping with (Cook 2008(Cook , 2009, L2 acquisition has access to basic properties of language, in part leveraging the parameterization fixed in L1, which, therefore, influences L2. However, L2 does not fail to influence L1 in turn, confirming the hypothesis that the speaker has a single overall system (Cook 2009;Thomason 2010), where semantic-syntactic and phonological representations obey morphological, syntactic order and lexical restrictions imposed by structural parameters and semantic cognitive principles (Baldi 2019). In other words, parameterization is the result of linguistic and cognitive restrictions mapping syntactic and phonological information onto representations available for interpretive sensory-motor and conceptual-intentional systems indifferently regardless of whether the source is L1 or L2. We must admit that basic properties of computation (UG) (Chomsky 2015) are recoverable and available to the learner as part of her/his internal language faculty not so much in childhood as in the subsequent acquisition. The properties of lexical and functional elements working in the influencing language, the source language, SL, and in the influenced language, the receiving language, RL (Thomason 2010), concur in fixing syntax and interpretation of sentences. In conclusion, as regards the mind of multilinguals, we will identify the Italo-Romance dialects as the source language, and the Gallo-Romance dialects as the receiving language, regardless of their original acquisitional status.
In the present case studies, we can reasonably surmise that (a large part of) secondand subsequent-generation speakers grew up bilingual Franco-Provençal/Apulian, Occitan/Calabrian and that their overall linguistic system was subject to a partial reorganization due to transfer from Lx, the Italo-Romance dialect, to Ly, the alloglot variety.

Distribution of Possessive Morphology
In Apulian Franco-Provençal the morphology of possessives separates two different alternants, whereby the prenominal possessive form is different from the one occurring in the postnominal position as well as in the pronominal/predicative contexts Four distinct agreement inflections emerge: -uN MSG, -O MPL, -a FSG, -@ FPL, in (30a), coinciding with the paradigm of the definite article, illustrated in (30c). A distinct paradigm characterizes postnominal/pronominal alternants in (30b), not least based on a type of inflection different from that in the article and devoid of the contrast sing/pl in the masculine. A similar paradigm characterizes the Guardia Piemontese dialect in (31). All in all, what is evidenced is that prenominal and postnominal possessives show a comparable referential force, at least in terms of distinct inflections, disproving the hypothesis that the prenominal position could be associated with a reduced functional layer in the sense of Cardinaletti and Starke (1994), Cardinaletti (1998).
In the same way as the South French-type varieties, prenominal possessives take on a specialized morphology, illustrated in (30a), which is different from the one associated with postnominal or predicative/pronominal possessives, in (30b). It is interesting to compare the data in (30) and (31) with the distribution that we find in in (32), in Piedmontese Occitan (Pomaretto) in (33), and in the Calabrian dialect of Cervicati, in (34). The latter has a radically different distribution of agreement inflections, again, interestingly, regardless the position of possessives.

Cervicati
Let us consider the distribution in (30)-(31) and (32) in more detail. Franco-Provençal/ Occitan Systems and Apulian/Calabrian Gallo-Romance share similar distinctions in the paradigms, but with an important discrepancy. In fact, in Franco-Provençal/Occitan, the prenominal possessive occurs with all noun classes, while in Apulian/Calabrian Gallo-Romance, the prenominal occurrence is limited to kinship terms. It is in Apulian/Calabrian Gallo-Romance that the pronominal/predicative alternants of possessives have been extended to postnominal contexts, missing, on the contrary in the original varieties. In the Southern dialects, in (32), the only distinction is the difference between enclitic forms.
Summing up, we have: Possessives in the left position in the DP show an inflectional paradigm coinciding with the system of articles.
The prenominal position selects a more differentiated inflectional system: by way of example, in the Celle/Faeto dialect, in the postnominal position, the contrast between singular and plural masculine is neutralized and the same form -NN@ is introduced.

The Effect of Contact: Two Types of Order
In the Franco-Provençal of Celle and Faeto, both readings of possessives, with and without the article, occur. Contact with surrounding dialects has favored postnominal possessives; moreover, the definite article is required in predicative and pronominal contexts, such as the Southern Italian dialects. The prenominal and article-excluding occurrence is limited to the kinship nouns, as in (35), where the original syntax is retained, whereby the special meaning of these terms is inherently able to constrain the set of possible referents.

The Effect of Contact: Two Types of Order
In the Franco-Provençal of Celle and Faeto, both readings of possessives, with and without the article, occur. Contact with surrounding dialects has favored postnominal possessives; moreover, the definite article is required in predicative and pronominal contexts, such as the Southern Italian dialects. The prenominal and article-excluding occurrence is limited to the kinship nouns, as in (35), where the original syntax is retained, whereby the special meaning of these terms is inherently able to constrain the set of possible referents.

Celle
Interestingly, both the Franco-Provençal and Southern Italian dialects limit this type of lexicalization to singular possessors, otherwise selecting the usual postnominal occurrence. Only the 2nd plural person vutuŋ 'you' has both the postnominal and prenominal alternant, where the latter corresponds to the respective 2nd singular possessor form 13 . The prenominal occurrence satisfies definiteness requirements only in the case of singular referents; on the contrary, the referential properties of plural possessors are not sufficient to subsume definiteness properties of the noun, so that we find the postnominal possessives in conjunction with the prenominal article. The asymmetric behavior of plurals partly also concerns possessed kinship nouns, that, in turn, are excluded from enclitic constructs in Southern varieties, while in Apulian Franco-Provençal and Calabrian Occitan, they alternate with the postnominal occurrence. Substantially, the preferred condition for the possessive to satisfy the definiteness requirements of the DP is when single referents are involved: the co-presence of a 1 st /2 nd /3 rd referent and a singular kinship term gives rise to the definite reading.
The double lexical/inflectional mapping of the possessum's definiteness properties, i.e., the inflection of the possessive + the article such as in the Southern Italian dialects is applied to most of the lexicon as in (31), differently from the other Franco-Provençal varieties, where possessives are also the exponents of the deictic/referential properties associated with the left domain of DPs, as seen for Coazze in (27). 13 In any case, in the Franco-Provençal dialects, 1st/2nd plural person possessives with a morphology coinciding with that of singular persons is attested (Hasselrot 1938). The Celle dialect seems to preserve this morphology only in order to refer to a single recipient/possessor.

Celle
Interestingly, both the Franco-Provençal and Southern Italian dialects limit this type of lexicalization to singular possessors, otherwise selecting the usual postnominal occurrence. Only the 2nd plural person vutuN 'you' has both the postnominal and prenominal alternant, where the latter corresponds to the respective 2nd singular possessor form 13 . The prenominal occurrence satisfies definiteness requirements only in the case of singular referents; on the contrary, the referential properties of plural possessors are not sufficient to subsume definiteness properties of the noun, so that we find the postnominal possessives in conjunction with the prenominal article. The asymmetric behavior of plurals partly also concerns possessed kinship nouns, that, in turn, are excluded from enclitic constructs in Southern varieties, while in Apulian Franco-Provençal and Calabrian Occitan, they alternate with the postnominal occurrence. Substantially, the preferred condition for the possessive to satisfy the definiteness requirements of the DP is when single referents are involved: the co-presence of a 1st/2nd/3rd referent and a singular kinship term gives rise to the definite reading.
The double lexical/inflectional mapping of the possessum's definiteness properties, i.e., the inflection of the possessive + the article such as in the Southern Italian dialects is applied to most of the lexicon as in (31), differently from the other Franco-Provençal varieties, where possessives are also the exponents of the deictic/referential properties associated with the left domain of DPs, as seen for Coazze in (27).

Celle
In conclusion, in the Celle and Faeto dialects, contact has induced a syntactic split anchored to semantic properties of the lexical elements involved.
The overlapping of the postnominal occurrence and the constraint on kinship terms is interesting evidence of the nature of transfer, as a non-arbitrary phenomenon. The transfer from contact dialects and the consequent reorganization of Apulian Franco-Provençal reshaped the morpho-syntax of possessives based on conceptual and structural properties in the range of the basic properties of the language faculty: definiteness and referential properties, syntactic representation of the part-whole relation and order of possessives.
Here, we may wonder why the structure of Lx has the upper hand. We conjecture that Apulian Franco-Provençal/Calabrian Occitan possessives were interpreted as endowed with the same lexical qualities as the Italo-Romance ones. More precisely, the coincidence in pronominal and predicative contexts, where D+possessive occurred, provided positive evidence for the combination between D and the possessive. It is no accident that in Apulian Franco-Provençal and Calabrian Occitan, there is no recategorization in possessives, since the possessive form that occurs postnominally is not the prenominal form, that keeps preceding kinship terms, but the predicative/pronominal alternant. The form that occupies the postnominal context is the form already requiring the article, in (30) and (31), corresponding to the predicative/pronominal forms attested in dialects retaining the original distribution, as in (32b) and (33b). If we identify predicative/pronominal forms of possessives as satisfying the usual distributional and functional nominal requirements, we conclude that the Merge operation in (37a), combining D and the possessive, is applied to the same syntactic object (conventionally labeled 'noun') in (37b).
Speakers exploit their linguistic knowledge of already acquired languages as clues to possible language structures, as in first language acquisition (Cook 2008), i.e., in terms of experience and of principles not specific to the faculty of language (Chomsky 2005), such as efficient computation. All in all, linguistic knowledge works in shaping new languages, recalling aspects of constructionist-like hypotheses, in the sense that the mapping from interpretive properties onto linguistic forms may also have recourse to non-specific cognitive devices (Goldberg 2004). However, the basic properties of computation (Chomsky 2015) are recoverable and always available to the learner as part of her/his Celle In conclusion, in the Celle and Faeto dialects, contact has induced a syntactic split anchored to semantic properties of the lexical elements involved.
The overlapping of the postnominal occurrence and the constraint on kinship terms is interesting evidence of the nature of transfer, as a non-arbitrary phenomenon. The transfer from contact dialects and the consequent reorganization of Apulian Franco-Provençal reshaped the morpho-syntax of possessives based on conceptual and structural properties in the range of the basic properties of the language faculty: definiteness and referential properties, syntactic representation of the part-whole relation and order of possessives.
Here, we may wonder why the structure of Lx has the upper hand. We conjecture that Apulian Franco-Provençal/Calabrian Occitan possessives were interpreted as endowed with the same lexical qualities as the Italo-Romance ones. More precisely, the coincidence in pronominal and predicative contexts, where D+possessive occurred, provided positive evidence for the combination between D and the possessive. It is no accident that in Apulian Franco-Provençal and Calabrian Occitan, there is no re-categorization in possessives, since the possessive form that occurs postnominally is not the prenominal form, that keeps preceding kinship terms, but the predicative/pronominal alternant. The form that occupies the postnominal context is the form already requiring the article, in (30) and (31), corresponding to the predicative/pronominal forms attested in dialects retaining the original distribution, as in (32b) and (33b). If we identify predicative/pronominal forms of possessives as satisfying the usual distributional and functional nominal requirements, we conclude that the Merge operation in (37a), combining D and the possessive, is applied to the same syntactic object (conventionally labeled 'noun') in (37b). In conclusion, in the Celle and Faeto dialects, contact has induced a syntactic split anchored to semantic properties of the lexical elements involved.
The overlapping of the postnominal occurrence and the constraint on kinship terms is interesting evidence of the nature of transfer, as a non-arbitrary phenomenon. The transfer from contact dialects and the consequent reorganization of Apulian Franco-Provençal reshaped the morpho-syntax of possessives based on conceptual and structural properties in the range of the basic properties of the language faculty: definiteness and referential properties, syntactic representation of the part-whole relation and order of possessives.
Here, we may wonder why the structure of Lx has the upper hand. We conjecture that Apulian Franco-Provençal/Calabrian Occitan possessives were interpreted as endowed with the same lexical qualities as the Italo-Romance ones. More precisely, the coincidence in pronominal and predicative contexts, where D+possessive occurred, provided positive evidence for the combination between D and the possessive. It is no accident that in Apulian Franco-Provençal and Calabrian Occitan, there is no recategorization in possessives, since the possessive form that occurs postnominally is not the prenominal form, that keeps preceding kinship terms, but the predicative/pronominal alternant. The form that occupies the postnominal context is the form already requiring the article, in (30) and (31), corresponding to the predicative/pronominal forms attested in dialects retaining the original distribution, as in (32b) and (33b). If we identify predicative/pronominal forms of possessives as satisfying the usual distributional and functional nominal requirements, we conclude that the Merge operation in (37a), combining D and the possessive, is applied to the same syntactic object (conventionally labeled 'noun') in (37b).
Speakers exploit their linguistic knowledge of already acquired languages as clues to possible language structures, as in first language acquisition (Cook 2008), i.e., in terms of experience and of principles not specific to the faculty of language (Chomsky 2005), such as efficient computation. All in all, linguistic knowledge works in shaping new languages, recalling aspects of constructionist-like hypotheses, in the sense that the mapping from interpretive properties onto linguistic forms may also have recourse to non-specific cognitive devices (Goldberg 2004). However, the basic properties of computation (Chomsky 2015) are recoverable and always available to the learner as part of her/his Speakers exploit their linguistic knowledge of already acquired languages as clues to possible language structures, as in first language acquisition (Cook 2008), i.e., in terms of experience and of principles not specific to the faculty of language (Chomsky 2005), such as efficient computation. All in all, linguistic knowledge works in shaping new languages, recalling aspects of constructionist-like hypotheses, in the sense that the mapping from interpretive properties onto linguistic forms may also have recourse to non-specific cognitive devices (Goldberg 2004). However, the basic properties of computation (Chomsky 2015) are recoverable and always available to the learner as part of her/his internal language faculty not so much in childhood as in the subsequent acquisition. This perspective explains many of the phenomena of transfer and mixing languages.

Morphology in the DP Phase
The treatments in the literature imply a recognition of the syntactic nature of the alternations we have examined. We propose explaining the distribution of the inflection of possessives in terms of its role in the DP phase. Our idea is that the internal structure of the phase is universally defined, in that head and complement of the phase are independently individuated by the Phase Impenetrability Condition in (38) (Chomsky 2001).
The domain of H is not accessible to operations at ZP; only H and its edge are accessible to such operations Chomsky (2001, p. 14) Phases connect the distribution of the relevant features with externalization. Specifically, phases are able to define boundaries in the application of agreement processes, in so far as they correspond to different stages of access to lexicon.
We are assuming DP to be a phase; notwithstanding the original proposal by Chomsky (2001), there is evidence that DP is a phase, insofar as it is a cyclic node for Subjacency, namely the difficulty in extracting from it ). This conclusion has been recently supported in Chomsky (2019). Our core proposal is that the D/NP split in the morphology of possessives depends on the independently motivated split between head of a phase and complement of a phase. More precisely, we assume that the prenominal possessive is associated with (the edge of) D, while postnominal or pronominal/predicative occurrence is associated with it NP complement. So, far from manifesting differences between different agreement processes-or between morphology and syntax-the distribution of the alternants reflects phasal structure interacting with externalization.
Applying phasal externalization to the data, a tentative parametrization can be stated, whereby the asymmetry in the distributional patterns of possessive alternants reflects the phasal status of the host lexical items along the lines of (39) (cf. Manzini et al. 2020;Savoia et al. 2019, forthcoming). In the process of externalization, a given property P may be differently realized on the head of the phase vs. the complement of the phase. All logical possibilities are expected to be instantiated.
The parameter in (36) is conceived to apply in principle to any (functional) property/feature, or at least to those with referential import. The asymmetry between agreement properties of determiners-and nominal modifiers/adjectives-and nouns have been discussed in the literature. Indeed, different types of split show up Savoia et al. 2019, forthcoming). Costa and Figueiredo Silva (2002) describe Brazilian Portuguese varieties, in which plural inflection -s only occurs on the determiners or prenominal adjectives, as in O-s/est-es/algun-s/un-s livr-o muit-o bonit-o 'The/these/some book very nice'. They adopt a distinction between dissociated and singleton morphemes, in the spirit of the DM treatment of Embick and Noyer (2001), whereby the plural in Brazilian Portuguese corresponds to a specialized interpretable morpheme (singleton), which combines only with the "element anchoring the information concerning number", namely determiners. The distribution in which prenominal determiners and adjectives lack (a set of) agreement properties in some Catalan varieties is addressed by Bonet et al. (2015), where prenominal agreement is connected to a 'family of constraints' enforcing or not general agreement at PF; on the contrary, postnominal agreement is syntactic in nature and triggered by Spec-Head agreement. The hypothesis that different manifestations of agreement could be referred to different syntactic operations, is pursued by several Authors.
A mechanism based on the split between different types of features, specifically marked vs. unmarked, is proposed in Pomino (2012) in order to account for the lack of number inflection in determiners in some Rhaeto-Romance dialects. Other principles referred to in the literature imply the asymmetric properties of pre-or postnominal agreement (Rasom 2006 in a cartographic framework) and the special Concord status of noun-modifier agreement as opposed to Agree (Bonet et al. 2015). In these approaches, number is treated as substantially accessory with respect to person and other referential properties. We put forward a different idea, assuming that what we see are different types of plural inflection, endowed with different interpretive import, which are inserted according to the morpho-syntactic context.
It is tempting to bring the contrast between pre-and postnominal possessives to the generalization of Costa and Figueiredo Silva (2002), whereby the phase head D is the referential anchor of the DP; thus, we expect it to externalize definiteness and other referential properties. Possessives seem to question this generalization, given that, in many Gallo-Romance varieties, they can replace determiners subsuming the entire burden of reference. This possibility closely recalls the phenomena concerning the distribution of plural inflection within the DP in some Romance varieties discussed by Costa and Figueiredo Silva (2002) and for Italian and Rhaeto-Romance varieties by Manzini and Savoia (2005), Baldi and Franco (2018),  and . In both cases, we see that referential properties can assume different types of inflectional exponent and a different distribution between the head element and the noun.
In Friulian (Montereale) and Cadore Rhato-Romance, plural feminine determiners exclude the plural exponent -s otherwise occurring on the feminine nouns . On the contrary, determiners present a different type of inflection, that in Cadore varieties is nothing but the feminine singular -a, as evidenced by the comparison between (40a) and (40b). In the Montereale dialect, (41a,b), in turn, the inflection -s occurs only on the noun, while the determiner has the other plural inflection.

Montereale
As shown by the preceding examples, the clearly plural morphologies -s, -e and (-)i occur complementarily and in combination, excluding a different syntactic status, although an interpretive difference can be assumed. However, we may expect some type of morphosyntactic split (Manzini and Savoia 2005), whereby definiteness and deictic elements are endowed with specialized morphology, given the role that they play in the identification of arguments. Generally, the occurrence of specialized plural elements is associated with the head of the DP phase, i.e., determiners and possibly other nominal modifiers. In the varieties in (37)-(40), plurality is expressed differently on the D head and noun. We can hypothesize that the inflection associated with D has the strongest semantic properties necessary to fix the reference to the argument.
Such an asymmetry also characterizes possessives that show a specialized form entirely coinciding with the expression of the possessor. We think that a necessary condition of Differential Marking of Possessives is the presence of a specialized position on the tree, namely [⊆] for inclusion relation. As a consequence, D vs. NP distribution of different possessive morphologies reflects the fact that the NP complement of the D phase is shipped to externalization separately from the D head and its edge. So the left-right asymmetry is actually a head/edge of phase vs. complement of phase asymmetry. In the case of possessives, we may conclude that there is an externalization parameter as in (44). (44) Referential properties of DP: deicticness (1st/2nd), definiteness, anaphoricity, Externalization parameter: (i) uniformly on D and complement/possessive (ii) on the possessive So, we have the possible occurrences in (45) If we consider the conclusion of Chomsky et al. (2019) and Chomsky (2019), that surface order is a product of the externalization process, we are justified in linking the different parameterizations to an interpretive effect at the conceptual-intentional C-I interface level. In other words, the morphological/lexical mapping of the definiteness properties of DP seems to suggest that the prenominal position linearizes the scope of the definiteness element, here the possessive, differently from the postnominal occurrence, highlighting the relation between N and its referential domain, here between the possessum and the possessor.
As it appears, we find all possible combinations and orders, suggesting that the relation between the possessive and the determiner is reduced to the contrast between languages that treat possessives as sufficient to lexicalize definiteness properties and languages that require determiners to lexicalize it. Not surprisingly, the continuum between the Occitan and Piedmontese dialects gives rise to hybrid systems, such as that of the Franco-Provençal of Cantoira (Val Grande di Lanzo), where possessives are preceded by the article, in (46a), except with kinship terms, substantially, such as in other Northern Italian dialects. The same forms are found in all other cases, in predicative contexts with and without the article, in (46'a) and (46b), and with pronominal occurrence in (46'c).

Cantoira
The data of Cantoira again show that there is no clear link between the form of the possessive and its position or its availability to combine with the determiner. In this dialect, the inflectional properties of all forms can be licensed in any context, satisfying all relevant interpretive requirements. Thus, e.g., miN occurs in combination with the article within DPs and in pronominal contexts, but it can meet by itself the referential requirements of kinship terms and of predicative reading.
Let us now re-consider the paradigms in (30)-(33). First, we see that there is no clear contrast in the completeness and richness of agreement features between pre-and postnominal possessives in Franco-Provençal and Occitan varieties. Complementarily, there are dialects, such as the Cervicati dialect, in (34), that uniquely present indeclinable postnominal and pronominal forms. At least on the descriptive level, nothing supports the idea that the prenominal position is associated with weak elements, whereas the postnominal position is reserved to strong, fully fledged elements.
On the contrary, we must think that full inflectional forms may equally occur in any position. The constraints on the distribution possibly reflect the parameter in (44), in turn based on the semantic properties of the lexical items. So, there are languages that have specialized forms available for the referential properties of the head of DP, such as Franco-Provençal and Occitan. In this case, we may say that forms such as muN, tuN, etc., in (30)-(33) include the specialized content for D, exactly like the specialized plurals in (40)-(43). In other words, it is not the completeness of phi-features that is at stake, but their properties. We can hypothesize that the ability of possessives to define a specific subclass of arguments is preserved in the ability of these forms to fully satisfy the head requirements of DP. Additionally, the other possessives may be fully inflected, such as in the case of postnominal possessives in our dialects, on par with Italian and Southern Italian dialects; in fact, the usual inflection of nouns emerges in these contexts. Finally, kinship terms bring other principles into play, to effect that the semantic properties of the noun fix, in turn, the referential properties of the argument. Not surprisingly, we find the entire D domain replaced by the combination noun+possessive, where the possessive element can precede (as in Calabrian Occitan/Apulian Franco-Provençal, but also in Standard Italian mio padre 'my father'), or follow (as in enclisis in the Southern Italian Dialects, in Romanian, and in Standard Italian mamma mia 'my mather', casa mia 'my house').

A Brief Conclusions
The analysis of possessives we have proposed is based on the inclusion relation ⊆ between the possessum and the possessor, understood as the interpretive property underlying the possessive structure. This relation is morpho-syntactically realized in more ways in the Romance linguistic domain; in Gallo-Romance varieties and Italo-Romance varieties, many differences emerge. In Gallo-Romance, possessives exclude definite determiners and are prenominal, whereas in Italo-Romance, possessives follow the noun preceded by the definite determiner. In the alloglot Gallo-Romance varieties of Apulia and Calabria studied in this work, the contact with the surrounding Southern Italian dialects favored a system including both pre-and postnominal possessives. Contact phenomena can be seen as the result of situations of bi/multilingualism and the main cause of linguistic variation.
Actually, the distribution of possessives in the Apulian Franco-Provençal and Calabrian Occitan not only reflects the contact systems but introduces a new rule, whereby the original prenominal possessive is preserved with kinship terms. This split is absent in the original systems and is implemented differently from the enclisis attested in Southern varieties. We can conclude that the transfer from contact dialects and the reorganization of the system of Celle reshape the morpho-syntax of possessives relying on conceptual properties and structural principles in the range of the basic properties of the language faculty. As for the morpho-syntactic analysis of possessives and their distributional and formal properties, we have argued that the different lexical properties of nouns and the referential force of possessives are involved. In order to define this relation, we have proposed a parameterization concerning the degree of uniformity of the externalization of referential properties in the DP phase.