The syntax of naming constructions in European Portuguese dialects : variation and change

This paper discusses the syntax of naming constructions with the verb chamar ‘to call’. We show that in some varieties of European Portuguese the verb chamar displays an alternation in the expression of the named entity, which is manifested by the presence/omission of the preposition a in sentences like Nós chamamos (a) isso cruzeta ‘lit. We call (to) this hanger’. We will look at this phenomenon from an areal, historical and theoretical perspective, trying to relate the intricate patterns observed. We will show that the alternation with the verb chamar is attested in the course of the diachrony of Portuguese, but it is synchronically confined to two restricted geographical areas: the archipelago of the Azores and the northern region of the Portuguese continental territory. As for the theoretical analysis, we claim that both the variants involve a small clause complement structure, with the named entity in the subject position. The a that precedes the subject of the small clause is taken to be a kind of ‘dummy’ preposition that showed up as a strategy to resolve syntactic ambiguity in sentences exhibiting the verb chamar with the personal pronoun se (which were ambiguous between an impersonal and an anticausative reading).


Introduction
In some non-standard varieties of European Portuguese, naming constructions with the verb chamar 'to call' display the alternation in (1): (1) a. Nós chamamos a isso cruzeta.
[Ø-DP DP structure] we call that cruzeta [hanger]   Sentences in (1) involve two post-verbal DPs corresponding to the named entity (isso 'that') and the name itself (cruzeta 'hanger'), the difference being that the named entity in (1a) is preceded by the preposition a.Some non--standard varieties of European Portuguese display both (1a) and (1b); however, the standard variety only displays (1a). 1 In this paper we scrutinize the alternation in naming constructions given in (1), focusing on the syntactic conditions that underlie the contrast between standard and non-standard varieties.We argue that both the variants in (1a) and (1b) involve a small clause complement structure with the named-entity in the subject position.We take the preposition a that precedes the subject in the variant (1a) to be a kind of a 'dummy' preposition that showed up at some point of the history of Portuguese as a strategy to deal with syntactic ambiguity in sentences involving the verb chamar associated with the personal pronoun se.Our claim is that the emergence of the preposition a in naming constructions led to the loss of the construction illustrated in (1b) in most varieties of Portuguese and concomitantly to the described synchronic variation.

A. The a/Ø-DP alternation
In some non-standard varieties of Portuguese the naming verb chamar displays an alternation pattern in the expression of the named entity.As illustrated in (2)-( 3), the named entity can be either a DP introduced by the preposition a2 or a plain DP: 3 (2) Chamam àquilo a giba.(PIC11) [named entity: a-DP] call.3PLto-that the giba [jib]   (3) Chamávamos isto uma dorna.(FIG02)  [named entity: Ø-DP] called.3PLthis a dorna [cask]   The a/Ø-DP alternation may coexist in the same variety, even in the speech of a single speaker, as illustrated by the following data: (4) Eu chamo aquilo varejar, sabe?

B. Word order
In the naming constructions with chamar the named entity and the name itself do not appear in a fixed word order.If both constituents are VP--internal, the named entity may precede or follow the name itself, as illustrated below: There are two factors that can explain reordering: (i) heaviness of the named entity and (ii) assignment of information focus to the named entity.If the named entity is heavier than the name itself, it may be postponed (see ( 14)-( 15)).( 14 Information structure also seems to play an important role in the relative order of the named entity and the name itself.It has been claimed that in some languages (Contemporary European Portuguese included) the assignment of narrow information focus drives the constituent expressing new information to the rightmost position of the sentence (cf.Costa 1998,  2004).This approach accounts for sentences exhibiting name itself-named entity order: in ( 16) and ( 17) the named entity surfaces in the rightmost position because it is the new information provided by the context.By contrast, if the name itself expresses new information, it stays in the rightmost position, as for instance in (2) and ( 3 We assume that the naming verb chamar in (19) can be derived from the one in (18) via the lexical operation of reduction (see Levin & Rappaport  1995), which affects the external theta-role and derives a one-place verb.Under this view, the verb chamar in ( 18) is a two-place verb, whereas the one in ( 19) is a one-place verb that selects a small clause complement (and involves DP-raising to the subject position).
Note that, abstracting away from the number of arguments involved, a similar derivation has been proposed for the causative alternation exemplified in ( 20)-( 21 Crucially, there are remarking similarities between the patterns in (18)--( 19) and ( 20)-( 21).Just like the unaccusative variant of afundar in (21), the one-place verb chamar in ( 19) is associated with the anticausative pronoun se. 5 Moreover, chamar-se patterns with afundar-se in not being able to co--occur with agent-oriented adverbs (see ( 22) and ( 23)) and purpose clauses (Manzini 1983) (see ( 24) and ( 25)): 5 The non-argumental status of se in ( 19) can be confirmed by the fact that it cannot be involved in clitic doubling constructions (Matos 2003)

Geographical distribution and historical remarks
The dialectal data inspected in this paper are drawn from the Syntax-Oriented Corpus of Portuguese Dialects -CORDIAL-SIN. 6,7In this corpus the areal distribution of the a/Ø-DP alternation shows a clear asymmetry: all the EP varieties represented in the corpus display the a/Ø-DP structure, but only 11 of them exhibit the Ø-DP variant.In the light of the geographic and historical considerations presented above, it becomes clear that the question that needs to be answered is not why some Portuguese dialects display the a-Ø-DP alternation, but instead why some Portuguese dialects have lost it.9

Competing analyses
There are two competing analyses that can apparently account for the syntax of naming constructions: the ditransitive structure (34a) and the small clause complement structure (34b) (see Matushansky 2006).
(34) a. ditransitive structure (simplified) b. small clause complement structure (simplified) In Section 4.2.we will provide three arguments showing that naming constructions with chamar involve a small clause complement structure and not a ditransitive structure.For ease of exposition, in the examples provided in 4.2.we will only consider the Ø-DP alternant; note however that the same results are obtained for the variant with a.

Arguments in favor of a small clause complement structure
The constituent italicized in (35) closely resembles a predicate semantically, in that it indicates the property (in this case, the name) attributed to isso.Under this view, (35) can be conceived as involving a two--place verb chamar, which selects a subject and a small-clause complement.The small clause complement contains a subject, the named entity (isso), and a predicate, the name itself (cruzeta).
we call that cruzeta [hanger]   This analysis is supported by a number of syntactic arguments, which are listed in A. through C. below.

A. Passives
If the name itself and the named entity were both complements of the verb (in a ditransitive construction), then we would expect both to passivize.However, the contrast in (37)  The contrast in (37) is, however, predicted under a small clause complement analysis.Note that if a typical verb selecting a small clause complement is involved (as considerar 'to consider' in (38)), the same pattern is obtained: the subject of the small clause can passivize (see (39a)), but the predicate cannot (see (39b)). the bladder is considered that

B. Pronominalization
As already mentioned in Section 2, the named entity can be replaced by the accusative form of the personal pronoun (see ( 6) and ( 7) above The contrasting behavior of the named entity and the name itself is unexpected if both constituents were complements of the verb chamar in a ditransitive structure.This contrast receives, however, a straightforward explanation if we assume that the name itself is a predicate.In actual fact, if a typical verb with a small-clause complement is involved (such as considerar 'to consider'), the same contrast is obtained:

C. Definite article omission
In naming constructions if the name itself is a proper name, it licenses definite article omission, as illustrated in ( 42)-( 43 By contrast, the preproprial definite article is required in contexts where the proper name occurs in a complement position, as in ( 44).This suggests that the proper name in ( 42)-( 43) is a predicate and not the complement of the verb chamar.

The analysis
The empirical data presented in A-to-C above provide crucial evidence in favor of the hypothesis that the Ø-DP variant is derived from a structure like (34b).Under this analysis, the named entity is the subject of a small clause, but it does not receive case from within the clause in the absence of Agr.It is chamar that assigns accusative case to the named entity via Exceptional Case Marking (ECM). 12Note that, as already mentioned in fn.11, this mechanism derives without further ado the accusative form of the personal pronoun found, for instance, in (6)-(7) above.
Now the time has come to demonstrate how the small clause complement hypothesis can account for the a-DP variant given in ( 45) and ( 46 12 The term Exceptional Case Marking was coined by Chomsky (1981) to describe non-canonical structural case assignment to an embedded subject by a certain class of verbs.Much discussion over the past thirty years has been devoted to specifying the precise nature of the structural restrictions involved in this configuration.For more details on the governing/checking conditions on Exceptional Case Marking, see Pesetsky & Torrego (2011), Chomsky (2005) and Lasnik & Saito (1999).
The major challenge here is to explain the fact that a PP may occupy the subject position of a small clause. 13Building on the distinction between functional/non-thematic (or 'dummy') and lexical/thematic prepositions (see Rauh 1993; Tremblay 1996; Rooryck 1996, a.o.), we would like to suggest that the named entity in the a-DP structure is not a true PP, but actually a DP (bearing a theme theta-role) preceded by the 'dummy' preposition a. 14 There are four empirical facts pointing towards the 'dummy' status of a.
call However, the secondary predicates may be associated with internal arguments that, not being true PPs, superficially exhibit a prepositional form (in Duarte's (2003) terms) (see ( 55)).This clearly indicates that the status of the preposition determines the availability of secondary predication (and, consequently, the possibility of having a PP as the subject of a small clause).
the men like of-them blond Given this evidence, we propose, along with Brugè & Brugger (1996), that the direct object of the verb chamar is dominated by a functional projection whose head can be filled by a, as depicted in (56).When a is spelled-out, it assigns dative case to its complement.This explains the presence of the dative pronoun lhe in sentences like (46).( 56)

Accounting for synchronic variation
As we have seen in Section 3., EP dialects diverge with respect to the availability of Ø-DP structure with the verb chamar.The majority of EP varieties display only the a-DP structure, the alternation between a/Ø-DP structures being restrained to a limited group of Azorean and Northern varieties.In this section we will attempt to explain the synchronic variation found in contemporary EP as a consequence of a change that took place in the history of Portuguese.Moreover, we will suggest that the se construction gives rise to a structural ambiguity that has a twofold effect: (i) the loss of Ø--DP structure and (ii) the emergence of the preposition a.
In tandem with the Ø-DP structure, documented since the earlier Portuguese texts, in the course of the 14th century the verb chamar also starts to be associated with the passive pronoun se: Assuming, along with Martins (2003, 2005), that passive se is lost in Portuguese after the 15th century, we claim that sentences with chamar like the one in (61) became syntactically ambiguous between an impersonal construction (with a nominative se) and an unaccusative construction (with an anticausative se) (cf.Section 2, point C).
(61) Chama-se esse lugar Ribeira de Alfageja.call.3SG-se[CL] that place Ribeira de Alfageja [proper name] The sentence in (61) can be interpreted as involving: A. The two-place verb chamar in a Ø-DP construction, with the impersonal se corresponding to the external argument of the verb and the DP esse lugar being the subject of a small clause; B. The one-place verb chamar in an anticausative construction, taking the DP esse lugar as its subject.
Somewhat tentatively we would like to suggest that the appearance of the preposition a in naming constructions is a strategy to solve the ambiguity described above. 16In contrast to (61), if the preposition is introduced, the pronoun se in (62) can only be analyzed as a nominative se. 1716 According to the Dictionary of Medieval Portuguese Verbs (DVPM − Dicionário de Verbos do Português Medieval), the Ø-DP variant is attested since the earlier Portuguese documents, whereas the a-DP variant is not attested until the 14th century. 17We are aware that the proposal put forth in section 5 is not fully compatible with the chronology proposed by Martins (2003, 2005) for the emergence of the impersonal se in the history of Portuguese.We claim that in the 14th century a--insertion takes place in order to avoid ambiguity between impersonal and anticausative use of the pronoun se in naming constructions.However, according to Martins (2003, 2005), impersonal se constructions only emerge in the 16th century.This is an open problem that requires further research.Nevertheless, one hypothesis that is worth exploring is that impersonal se may have arisen before the 16th century.Evidence for this comes from earlier occurrences of non-agreeing se constructions, as in (i)-(iv), taken from Fiéis (2003: 431-432 As a consequence, the utterances manifesting Ø-DP decreased significantly in the primary linguistic data.Because the available triggering experiences changed in a critical way, children converged on a new grammar, which does not display Ø-DP.As such, in these dialects the two-place verb chamar started to be expressed only by the a-DP variant. O navio afundou-se.(Duarte2003: 306)  [unaccusative variant] the ship sank-se[CL] The 11 locations are confined to two restricted geographical areas: the archipelago of the Azores and the northern region of the continental territory. 8Map 1 shows the geographical distribution of the a/Ø-DP naming constructions within the CORDIAL-SIN corpus.Map I. Distribution of a/Ø-DP naming constructions with chamar in CORDIAL-SINIn the varieties that register the a/Ø-DP alternation, the relative frequency of the two structures is not uniform.The values are summarized in TableI.Map II represents the incidence of the Ø-DP variant in the relevant dialects.
As artes e os livros [name of the blog] one of-the melhores blogues portugueses -pelo menos, eu considero- ) Evaristo [proper name] him:cl.dat a arcaz [piece of furniture]

Table I .
a/Ø-DP naming constructions in CORDIAL-SIN -relative frequency Interestingly, if we look at the diachrony of Portuguese, we also observe cases of a/Ø-DP alternation in the expression of the named entity, as illustrated in (26)-(29): Map II.Incidence of the Ø-DP structure indicates that only the name itself can passivize.
).11By contrast, the name itself can only be spelled-out as a pronominal predicate, like assim (see (40b)).No other pronoun (in particular the accusative personal pronoun o in (40c)) can express this constituent.
.3PL to those cakes esses de Peniche [name of a cake] b.Esses bolos são chamados esses de Peniche.
the defense se[CL]spend the gold and the silver (iii) nom he razom que se tenha ceumes nem duvyda(CIPM, 1433/1438?)non is reason that se[CL]have jealousy nor doubt (iv) ese anno ou annos que se asy perder (CIPM, 1450) that year or years that se[CL] as-such loose (62) Chama-se a esse lugar Ribeira de Alfageja.call.3SG-se[CL]A that place Ribeira de Alfageja [proper name] upon