Cross-linguistically, in languages showing verb-subject agreement, full agreement between the preverbal subject and the verb is the norm. However, when the order is reversed, and the subject follows the verb, languages exhibit different agreement patterns; mostly either full (e.g., English in (1)) or defective agreement1 (e.g., French in (2)).
|(1)||English: full agreement with the postverbal subject|
|There have arrived three men.|
The observation of the importance of linear order for subject-verb agreement translates directly into the idea of agreement via Spec-Head configuration in the Principles and Parameters framework (Kayne 1989; Chomsky 1991). Since then, subject-verb agreement has often been treated as the reflex of a Spec-Head configuration. But this account fails to derive agreement with the postverbal subject in a straightforward manner: LF-movement of the postverbal subject is assumed in order to derive the necessary Spec-Head configuration for agreement between subject and verb (Chomsky 1995; Cardinaletti 1997). This account, however, has empirical as well as conceptual problems. Mentioning just one empirical problem, it can be shown that sentences with a postverbal subject do not receive the same interpretation as the equivalent with a preverbal subject, which would be expected under the LF-movement approach, as exemplified by Sabel (2000) for the quantifier many:
|(3)||There are not many men in the garden.|
|(4)||Many men are not in the garden.|
While for (3) only a narrow scope reading is possible where many does not take scope over negation, (4) allows both the narrow and the wide scope reading. Furthermore, it has been shown that agreement can obtain over a distance (see Section 4.1.1 for references).
Aside from the empirical problems, the concept of an LF-operation that triggers an overt morphological reflex seems problematic for the standard assumption of the (inversed) T-model (Chomsky 1981): morphology has to anticipate what happens at a later stage of the derivation, i.e. LF-movement of the subject DP to the specifier of the relevant projection for agreement. This mechanism can only be successful if one assumes a feature checking mechanism à la Government and Binding (GB) where already present features simply have to be checked. This and other considerations called for separating agreement from movement and Spec-Head which in Minimalism led to the concept of agreement as the reflex of a Probe-Goal relation as proposed by Chomsky (2001).
However, the most prominent account of subject-verb agreement with postverbal subjects in Romance is the Chain Analysis proposed by Rizzi (1982) and subsequent work: the preverbal subject position is occupied by a null pro which stands in a chain-like relationship with the postverbal subject via co-indexation. Through this chain relation pro and the subject share the same φ-features and nominative case, which is assigned to pro by T°. Being in a Spec-Head relation, pro and T° can check their φ-features.
However, this approach also raises several questions that – to my knowledge – have not been ultimately answered in the literature. First, how does the co-indexation between pro and the postverbal subject arise, i.e., are they the identical element and originate, putatively, within a Big-DP? If so, where is the Big-DP base-generated and which mechanism raises pro out of the Big-DP while the lexical DP remains VP-internal? The hypothesis that pro and the postverbal DP are two different nominal elements being co-indexed, gives rise to (at least) a Principle C violation: the postverbal DP is c-commanded by the co-indexed pronoun pro. Furthermore, the Chain Analysis predicts obligatory full subject-verb agreement in inversion of Null Subject languages (NSL). This prediction is not borne out in all NSL, as we will see in the following.
In this paper, I deal with the agreement properties in inversion of two North-Eastern Italian dialects: the varieties of Mestre (province of Venice) and Gazzolo d’Arcole (province of Verona). These dialects show a systematic alternation in past participle agreement (PPA) with postverbal subjects, as shown here for Mestre:
It is, thus, a case where a language does not choose the either-or option for verbal agreement in inversion, as it has been claimed for English and French, but where both agreement patterns are possible.
The relevant data of the two varieties will be presented, showing that the alternation is not an optional phenomenon and that it deserves a principled explanation in syntactic terms (Section 2). It will become clear that the information structural status of the DP and the pragmatic force of the sentence are crucial for agreement. More specifically, I will propose in Section 3, that the defective agreement pattern can only appear when the DP is discourse-new and the sentence represents a thetic statement, while full PPA is associated to categorical sentences. The agreement alternation found in inversion will be approached from the perspective of PPA in Romance. Therefore, in section 4, three accounts of PPA will be shortly discussed. Subsequently, a Minimalist syntactic account of the alternation will be presented. Section 5 offers the conclusion and outlook.
The dialects of Mestre and Gazzolo d’Arcole are closely related Veneto varieties but belong to different subgroups: the Mestre variety is Veneto lagunare, while Gazzolo belongs to Veneto occidentale (Zamboni 1974). The data presented in this section were gathered during the piloting phase for a related research project in North-Eastern Italian dialects. Here, I will report the judgements from two main informants, that were backed up by more native speakers regarding specific judgements.2 Before turning to the presentation of the gathered data, it is necessary to point out some of the similarities and differences of the two dialects.
The first similarity of the varieties in question is their use of subject clitics (SCL). The doubling of the subject DP by inflected clitics is a phenomenon particular to Northern Italian dialects, as exemplified in (7) for Florence (Brandi & Cordin 1989).
There is abundant literature on the nature and distribution of SCL, but due to space limitations, a very short remark about the relation between SCL and PPA will have to suffice for the present purpose. The dialects at issue differ in their systems regarding the occurrence of SCL. However, they behave similar when it comes to postverbal DPs: the SCL only appears when the subject represents a (right-dislocated) topic, as exemplified in (8). The presence of SCL correlates with obligatory full PPA, but, as can be seen in (9), this correlation is unilateral since full PPA can occur without SCL when the DP is not topicalized, as here exemplified for Gazzolo.
Thus, since full PPA can appear without SCL, I will not treat the occurrence of SCL as an ultimately decisive factor for the agreement alternation and, therefore, treat PPA separately from the occurrence of subject clitics.
Another similarity between the dialects is a syncretism of the 3rd person verbs: there is no morpho-phonological marking for number in the 3rd person, only on the subject clitic, if present.
In order to test verbal agreement with full DPs and 3rd person pronouns, we have to fall back on the morphological number and gender distinction on the participle in compound tenses and passive constructions.
Gazzolo exhibits a further restriction in the potentially tested verbs: most participles are truncated, with the exception of strong, irregular participles, restricting the number of testable verbs. Table 1 shows the truncated forms of the participle rivà (‘arrived’) and the visibly inflected forms of the strong participles morto (‘died’) and nato (‘born’).
|rivar (to arrive)||rivà||rivà||rivà||rivà|
|morir (to die)||morto||morta||morti||morte|
|nascer (to be born)||nato||nata||nati||nate|
The Mestre variety, on the other hand, exhibits truncation in the masculine singular of the participles, while feminine and plural forms show inflection, as Table 2 illustrates.
|rivar (to arrive)||rivà||rivada||rivai||rivae|
|venir (to come)||vegnù||vegnua||vegnui||vegnue|
I report on the data of the dialect of Mestre, since it offers the most complete picture due to the lack of truncation of feminine and plural participles, but the dialects at issue seem to function alike. Unsurprisingly, preverbal subjects obligatorily trigger full agreement with the verb:
The PPA alternation appears in unaccusative inversion (13) and in passive structures (14) only, where a subject can freely appear in postverbal position and the auxiliary essere enables a morphological reflex of inflection. Unergative and transitive verbs are excluded since they select the auxiliary avere which does not allow visible inflection, as exemplified for unergatives in (15).
The lack of gender and number agreement is ungrammatical in topicalization, left or right dislocation, which is probably due to the obligatory presence of subject clitics in topicalization.
The defective agreement pattern is also ungrammatical in contrastive focalization, wh-interrogatives and relative clauses.
The PPA alternation is thus restricted to unaccusative and passive VS configurations, but, as we shall see, in those two configurations it is restricted in a systematic way: the felicity or even grammaticality of defective and full agreement crucially depends on the pragmatic force of the sentence and the nature of the postverbal DP.
The PPA alternation can occur regardless of definiteness (see (21) and (22)), specificity4 (see (23) and (24)), and animacy (compare (21), (22) and (25)) of the DP. The two dialects show defectiveness regarding number and gender in PPA, as exemplified below.
Many Veneto varieties exhibit also a plural inflected form of the quantifier ‘nobody’ and ‘somebody’: nisunSG/nisuniPL; qualchedunSG/qualcheduniPL. These plural quantifiers are able to trigger plural agreement in other constructions, as can be seen in example (26) from Gazzolo where nisuni triggers the plural on the subject clitic:
But postverbal nisuni and qualcheduni obligatorily trigger defective agreement, while the fully inflected participle is ungrammatical:
The opposite is the case with referential null subjects: full agreement is obligatory, defective agreement is ungrammatical.
To summarize, the agreement alternation can appear with most types of DPs in unaccusative and passive inversion. There are, however, instances where either full or defective agreement is ungrammatical: the Veneto-typical plural quantifiers cannot appear with the full agreement pattern, while null subjects obligatorily trigger full agreement. Apart from these exceptions, the grammaticality of the two agreement patterns seems not to be at stake, as showed in examples (21)–(25). However, there are pragmatic factors that affect the felicitousness of the full or defective agreement pattern leading to the conclusion that the phenomenon is indeed a systematic alternation instead of mere optionality.
Full PPA is preferred when the subject DP is discourse-linked (d-linked). A d-linked DP is a referent drawn from the contextually determined set in the preceding discourse (Pesetsky 1987; Enc 1991), as in (30).
Preference for defective PPA, on the other hand, arises when the DP is discourse-new:
Existentials cannot be tested in order to verify the correlation of defective agreement with discourse-new DPs, because the dialects in question exhibit the already mentioned 3rd person syncretism and do not exhibit an expletive form that could provide diagnostics for agreement.
An additional pragmatic effect in which the defective PPA is preferred is unexpectedness:
On the other hand, expected circumstances are less likely to be expressed with defective PPA:
Summarizing, the agreement alternation in the investigated dialects cannot be satisfactorily explained by optionality. The pragmatic status of DP is crucial for verbal agreement: full PPA correlates with d-linked DPs, while defective PPA correlates with discourse-new DPs and unexpectedness. In the following section, I will show that defective and full agreement correlate with thetic and categorical sentences respectively.
The thetic/categorical distinction specifies the pragmatic force of an utterance as well as the pragmatic status of its constituents. Thetic sentences are logically unstructured, since they consist of entirely new information and are event- or entity-reporting (Sasse 1987; Lambrecht 1994). They do not obey a sentence partition into topic and comment or subject of predication and predication (Kuroda 1972; Ladusaw 1994).5 Categorical sentences, on the other hand, are bi-partite: the predicate comments on the subject of predication. In semantic terms, the grammatical subject of thetic sentences is interpreted inside the predicative nucleus, while the subject of categorical sentences is interpreted outside of the predicative nucleus, as indicated below (Ladusaw 1994; Bianchi & Chesi 2014).
According to Sasse (1987), the thetic/categorical distinction can only be explained if the interplay of semantics, pragmatics, syntax and information structure is taken into consideration. He claims that the crucial factor is the pragmatic dimension of expectation, i.e. the speaker’s assumption on the hearer’s expectation. Sasse (1987: 548) illustrates this using an example of English where subject-accented sentences license a thetic interpretation (Lambrecht 1994). While the sentence in (37) conveys that Truman’s death was expected, maybe due to a long illness, the death of Johnson in (38) came unexpected, as expressed by use of a subject-accented sentence:
Languages use different formal means to disambiguate between thetic and categorical sentences (Sasse 1987: 560): while English and German use subject-accented sentences to express theticity, Romance languages use word order, i.e. VS structures. Romance and especially Italian inversion has been extensively discussed in the literature. It is widely acknowledged that Italian unaccusative VS with indefinites can be interpreted in a broad focus reading (Antinucci & Cinque 1977; Belletti 1988; Pinto 1997), representing thetic sentences:
French thetic sentences can equally be expressed in inversion, but exhibiting defective verbal agreement (Sasse 1987: 560).
Regarding the North-Eastern Italian dialects at issue, the sentences exhibiting defective agreement share some properties with thetic sentences. First, they represent all-new information: neither the DP nor the verbal complex is activated in the preceding discourse. The structure with defective agreement is construed as a felicitous answer to the question “What happened?”. They thereby describe an event rather than commenting on an established subject. Secondly, the defective agreement pattern conveys the effect of unexpectedness of an event, and is highly preferred over the full agreement when preceded or followed by an exclamation, as shown in (33) and (34). The sentences with full agreement, on the other hand, are bipartite in the sense that they comment on a subject that was already established in the discourse, reminiscent of categorical sentences.
Furthermore, as shown in (27) and (28), the Veneto-typical plural quantifiers nisuni (‘nobody’) and qualcheduni (‘somebody’) obligatorily appear with defective agreement. In the light of the thetic/categorical divide, the reason becomes clear: nisuni as well as qualcheduni are necessarily new and not d-linked. They therefore are restricted to thetic sentences, and cannot appear as the pragmatic subject in categorical sentences. A referential null subject, on the other hand, obligatorily triggers the full agreement pattern, as exemplified in (29). Following current literature, I assume that the referential null pro has to be topical in this context in order to be licensed (see e.g., Frascarelli 2018). Since thetic structures do not accept topical/given DPs, the defective agreement pattern is ungrammatical.
As many languages formally disambiguate between thetic and categorical sentences, it seems plausible that also the North-Eastern Italian dialects in question might do so. Contrary to Standard Italian, however, the opposition of preverbal and postverbal subjects is not ultimately telling, because of a language-specific restriction on the position of indefinite DPs, namely, that preverbal indefinite DPs are near ungrammatical:
This restriction bars the possibility of disambiguation of thetic and categorical structures by subject position. The North-Eastern Italian dialects in question seem to use the morphosyntactic means of agreement in order to disambiguate thetic from categorical structures: thetic sentences are marked by defective PPA in inversion (as in French), while categorical sentences exhibit full PPA.
There has been a lot of controversy on the pragmatic nature of the DP in categorical sentences, but scholars agree that it is topical. Sasse (1987: 568) further specifies the categorical DP as a discourse topic about which “information can be expected in the given situation”. Bianchi & Chesi (2014: 558) also hint at a topic-like interpretation: they state that presuppositionality of the subject “follows as a side effect of the thetic/categorical divide” and that the subject of thetic sentences “must lack any existential presupposition (and a fortiori D-linking)”. Thus, scholars dealing with the thetic/categorical divide agree that the notion of topic is a correct approximation of the pragmatic status of the categorical DP. Due to the ambiguous use of the notion topic in the literature, for the present purpose, I follow Bianchi & Chesi (2014) and shift to the notion of d-linking as an unmistakable sign of the topical status of elements, rendering the semantico-pragmatic relation between predication and postverbal argument more seizable.
As described above, thetic sentences do not dispose over a pragmatic subject by definition, i.e. a predication base over which the predication comments. Bentley & Cruschina (2018) use Bianchi’s (1993) Principle of Non-vacuous Predication (stating that every predication requires a preverbal subject) as a starting point in order to explain the possibility of Italian broad focus VS structures. According to them, a silent Subject of Predication (SoP) satisfies the need for a preverbal subject when the pragmatic subject does not raise to the preverbal subject position. They claim that there are different possibilities to retrieve a SoP: first, it can be lexically inherent to the verb (restricted to goal-entailing verbs of inherently directed motion, like arrive, come, enter, activity verbs and activity-based semelfactives, like phone, knock, call). Secondly, it can be situationally inferred by the context.
I propose extending the notion of SoP from the domain of pragmatics to the syntactic dimension: the SoP is actually a silent expletive-like element that is present in thetic sentences. Thus, in case of thetic sentences “an expletive formally complies with the requirements of Subj, thus conveying the interpretation that the event is not presented as being about a particular argument.” (Rizzi 2005: 213). My proposal is opposed to the traditional view of expletives as syntactic dummies that uniquely satisfy the need imposed by the EPP (Chomsky 1981, 2001). Rather, expletives are syntactic elements with pragmatic content.6 As a consequence – and in order not to violate the Inclusiveness Condition by inserting an expletive during the derivation – I propose that the pragmatically contentful expletive is a lexical item that is selected from the lexicon in order to form a thetic sentence. Evidence for this claim is abundant, in the form of overt expletives from different languages, including French il, English there and German es.7 These expletives clearly have differing syntactic properties (since some trigger agreement, while others do not), but are united in triggering a thetic interpretation.
Summarizing, there are two elements in the current literature: first, Bentley & Cruschina’s (2018) SoP, as a semantico-pragmatic element; and second, the traditional syntactic dummy that is inserted when a subject does not have the decisive properties in order to occupy the preverbal subject position (e.g. definiteness, specificity, topicality; see Svenonius 2002: 227). These two elements can be united in a pragmatically contentful expletive, what I will call ‘thetic expletive’ from now on, that is selected from the lexicon to form thetic sentences. Of course, as Bentley & Cruschina (2018) note, there are conditions that restrict the appearance of this element. While Bentley & Cruschina (2018) declare the lexical and contextual restrictions to be the source of the SoP, I propose that these restrictions are rather the consequences of the syntactic and semantico-pragmatic nature of the thetic expletive itself.
The distribution of the agreement patterns in the North-Eastern Italian dialects of Mestre and Gazzolo d’Arcole can be directly and easily explained by adopting an analysis in which a silent expletive is involved in the sentences that exhibit defective PPA. The presence of this thetic expletive triggers a morpho-syntactic reflex, i.e. defective agreement, as well as the interpretation of the sentence as a thetic one. Thus, the thetic expletive unifies the idea of a semantico-pragmatic SoP and of a syntactic dummy. Furthermore, the effect in the investigated dialects enables conclusions on its syntactic nature, suggesting that it carries a φ-feature set and, thus, is an adequate target for the T°-probe. In section 4.2.2, I will explain in depth how the agreement mechanism works and under which conditions the thetic expletive can be merged.
In order to present the syntactic analysis of the PPA alternation, some theoretical considerations about Romance PPA are necessary. While previous analyses have considerable problems, a novel approach by Longenbaugh (2019) successfully accounts for the empirical data. Building on his analysis, the syntactic analysis of the agreement alternation in the investigated Italian dialects will be laid out in Section 4.2.
In Kayne’s (1989) influential account of Standard French and Italian, PPA requires movement of the DP, giving rise to Spec-Head agreement. Under this view, PPA obtains if the object moves through the specifier of a dedicated agreement head AgrO° to which the participle has moved. In this Spec-Head configuration PPA is licensed. Thus, Kayne’s account explains the observed correlation between movement and PPA in Romance, but crucially, it is not able to straightforwardly account for agreement in inversion, as in the Italian so-called Free inversion (Belletti 1988). In this case, LF movement has to be assumed to satisfy the Spec-Head condition of PPA. This solution, however, is problematic, since it predicts an identical interpretation for the corresponding SV and VS sentences, which is not borne out (see arguments in Section 1). Kayne’s account also empirically fails in cases where the object DP has moved but agreement does not obtain (Deprez 1998: 2):
Furthermore, many subsequent works on agreement phenomena have shown that agreement cannot rely on a Spec-Head configuration but that it can obtain at distance (see Chomsky 2001, among many others).
D’Alessandro & Roberts offer an account of PPA in the Minimalist framework using Probe-Goal agreement rather than Spec-Head. In their approach too, however, the morphological realization of PPA ultimately depends on movement of the agreement controller. They assume that PPA is the result of v° probing the internal argument (IA), but crucially, the morphological spell-out of the valued features will only take place if v° and the IA are sent to PF in one chunk, i.e. as the complement of the next higher phase head, i.e. C°. Their account, however, faces empirical as well as conceptual problems. To name one of the empirical problems, as Longenbaugh (2019: 57) points out, their account crucially relies on the assumption that Italian participles move to a relatively high position, at least to v°, granting that the participle and the IA are not contained in the same phase complement and, thus, are not sent to PF together. This is shown with the necessarily higher position of participles with respect to manner adverbs (Cinque 1999: 102):
In French, on the other hand, the participle does not raise over the manner adverb bien and may even appear below VP-level adverbs like presque and à peine. Assuming D’Alessandro & Robert’s account, this offers the precondition for the morphological realization of PPA. However, PPA does not obtain in those cases:
Apart from empirical problems, there is (at least) one problem from a conceptual point of view: D’Alessandro & Roberts predict that syntactic agreement with the IA always obtains but in reality we find that the conditions under which agreement is morphologically realized are highly restricted.8 This clearly constitutes a conceptual inferiority compared to analyses that only predict morphological agreement when syntactic agreement obtains.
Longenbaugh investigates the relationship between movement and agreement and, therefore, focuses on a domain where agreement has been claimed to be closely related to movement: PPA. His central claim is that the agreement asymmetries we observe in Italian and French are not directly related to movement but rather to case: agreement can only target goals with an unvalued case feature. Before explaining his account, it is useful to shortly introduce the framework he works in as well as some theoretical tools.
Longenbaugh’s syntactic account is based on four ingredients: feature-driven Merge, the Obligatory Operation Hypothesis (Preminger 2014), the condition of Feature Maximality and the Bobaljik-Moravcsik treatment of case.
Longenbaugh follows the assumption that certain syntactic heads are associated to a Merge operation (e.g., merger of the external argument in Spec,vP of unergative and transitive verbs). He encodes this mechanism with a special Merge feature on these heads (Longenbaugh 2019: 23). These Merge features can be satisfied either by external or internal Merge. Furthermore, he assumes that a syntactic operation must apply if it can apply. This is a widely accepted assumption in Minimalism, no matter if one assumes the Obligatory Operation Hypothesis by Preminger (ObOp, 2014) or the Uninterpretable Feature account of Chomsky (2001). The important difference between those two accounts is what happens when agreement fails: under the Chomskyan view, a derivation crashes if an agreement operation fails whereas in the ObOp frame (Preminger 2014), an unvalued feature may receive the default value without affecting the grammaticality of the sentence. If one head is associated with several requirements, there is no specific order in which those requirements have to be satisfied: they can apply in every possible order as long as the result is interpretable at the interfaces.
The relationship between movement and agreement is governed by an economy condition:
|(46)||Feature Maximality (Longenbaugh 2019: 10)|
|Given head H with features [Fi] … [Fn], if XP discharges [Fi], XP must also discharge each [Fj] that it is capable of.|
Feature Maximality gives the instruction to maximize the discharge of requirements in one established Probe-Goal relation. Thus, if a head enters into an agreement relation with another element and the probe also bears a Merge feature that the goal can in theory satisfy, it has to satisfy it. Conversely, then, if a certain syntactic element satisfies the Merge feature of a given head, it must also value the unvalued features on the given head; “agreement is therefore predicted to be associated with movement if and only if the triggering head has an independent, unrealized need to project a specifier.” (Longenbaugh 2019: 21).
The fourth and last necessary tool in his account is the Bobalijk-Moravcsik treatment of case (Moravscik 1974; Bobaljik 2008; Preminger 2014), which has two essential components. First, case is valued configurationally. Unvalued case features on DPs can be licensed either via lexical/oblique case or via dependent case. Certain syntactic heads have the ability to value the case feature of their specifier or complement, resulting in lexical/oblique case. For the other cases, case features on DPs are valued configurationally. In the presence of two DPs, the lower DP, i.e. the IA, obtains dependent case which is morphologically realized as accusative. Case valuation obtains as soon as the configurational requirement is met. The case feature of the higher DP remains unvalued which does not result in a derivational crash but in the spell-out as nominative.9
|(47)||Dependent Case valuation in transitive clauses: Merge EA, value case on IA|
|(Longenbaugh 2019: 28)|
Case Accessibility determines whether a DP is a possible goal for agreement (Bobaljik 2008; Preminger 2014). This is a parametrical option for languages: some languages only allow φ-agreement with DPs without Case value, while others also allow agreement with already case valued DPs.10
Having introduced these theoretical tools, we can now turn to Longenbaugh’s detailed account of French and Italian PPA in unaccusative/passive inversion.
Longenbaugh assumes, along with more literature (Belletti 2017; D’Alessandro & Roberts 2008), that PPA is triggered by the head that introduces the external argument, v°. In Longenbaugh’s account, unaccusatives and passives bear a vP (Legate 2003; Sauerland 2003) and v° is associated with a φ-probe [φ: __ ] and a [D] (Merge) feature, just like transitive v°.11 The [D] feature on v° can either be satisfied by internal or external Merge, with the constraint on unaccusative v° that it cannot externally Merge semantically contentful DPs, i.e. it blocks the insertion in Spec,vP of arguments that are not already assigned a theta-role.12 Longenbaugh follows Richards & Biberauer (2005) and Deal (2009) in assuming that expletives are Merged in Spec,vP.
Since v° has two requirements (discharge [φ: __ ] and discharge [D]), there are two derivational options that can apply as the ordering of the operations is not fixed: first discharge [φ: __ ] and then discharge [D], or first discharge [D] and then discharge [φ: __ ]. The ordering of these derivational options crucially affects the position of the subject as well as the success of the agreement operation due to Case Accessibility, as will be sketched out below.
The first possible ordering of derivational processes after the Merge of v° (first φ, then [D]) results in the PPA with the IA and its subsequent obligatory movement to Spec,vP and Spec,TP, thus, a SV structure:
|(48)||[TP [Trois sauterelles]i sont [vP ti [ mortes ti ]]]|
If, on the other hand, the [D] feature is satisfied first, an expletive will be merged in Spec,vP. The French expletive il is pronominal in nature and bears a case feature, thus, it is a case-competitor. Therefore, the necessary syntactic configuration for dependent case valuation of the IA is met, triggering the immediate valuation of the case feature of the lower DP. The IA is case-valued and therefore inaccessible for the φ-probe of v°. The unvalued φ-features on v° are morphologically realized with the default value [M.SG].
The Merge of a case-feature-bearing and, thus, case-competing expletive triggers dependent case valuation, making the agreement operation impossible and resulting in the default valuation of the φ-features on v°.
Standard Italian differs from French in two regards: there is no overt expletive in preverbal position, and PPA is obligatory with postverbal subjects.
Longenbaugh (2019) bases his analysis of PPA in Italian unaccusative inversion on two assumptions: firstly, Standard Italian disposes over a null expletive (Rizzi 1982; Sheehan 2006) that can satisfy the [D] feature on v° and T°; and secondly, the Italian null expletive does not carry a case feature, i.e. is not a case-competitor (as the English there expletive) contrary to the French expletive il (Rizzi 1982; Cardinaletti 1997).
Applying Longenbaugh’s system, the first derivational option (discharge φ and then [D]) results in a SV structure, as it is the case in French. But for the present purpose the second derivational option resulting in a VS structure is of bigger importance.
If after the Merge of v° the [D] feature is satisfied first, again, an expletive is merged in Spec,vP. The crucial difference to the French example is that the expletive is not a case-competitor since it does not carry a case feature. Therefore, the necessary syntactic configuration for dependent case valuation is not met: the IA is not valued with dependent case and remains accessible to the v°-probe. The φ-features on v° get valued by the IA.
Longenbaugh, thus, explains the different PPA properties in French and Italian inversion with the presence of two different kinds of expletives. While the French expletive il is a case-competitor and therefore introduces dependent case valuation of the IA, the Italian null expletive pro is not a case-competitor, i.e. no case valuation takes place. The French case-valued DP is not accessible to the φ-probe, while the Italian postverbal DP is: PPA can only obtain in the Italian case.
Concluding, the approach by Longenbaugh (2019) can account more properly for the empirical facts of Romance PPA, including optionality of PPA in French in wh-interrogatives. Furthermore, Longenbaugh’s system gives a possible new perspective to tackle the syntactic mechanisms of PPA alternation in the North-Eastern Italian dialects of Mestre and Gazzolo, as reported in section 2.
In this section, I develop a syntactic analysis of PPA alternation in the dialects at issue on the basis of Longenbaugh (2019). Note that for the moment, I will ignore the differing semantico-pragmatic factors for the appearance of the alternation. In section 4.2.3, the syntactic and the pragmatic analysis will be tied together.
Along the lines of Longenbaugh (2019), I analyse the defective agreement pattern as default agreement, i.e. the morphological repair of a failed φ-agreement operation. Unvalued features do not necessarily lead to a derivational crash but are spelled out with the language-specific default morphology: nominative for unvalued case features, 3rd person masculine singular for unvalued φ-features on the verb. As we will see, the existence of two different kinds of expletives is the key to understand the syntactic and pragmatic effects of the agreement alternation in unaccusative inversion. The North-Eastern Italian dialects at issue exhibit two kinds of expletive-like elements: first, an expletive of the kind of Standard Italian expletive pro, which is not a case-competitor and does not block agreement with the IA; and secondly, an expletive, which is a case-competitor and blocks agreement with the IA, as the French il. The two different expletives are linked to different pragmatic interpretations, namely the thetic and categorical interpretation.
The analysis of the full agreement pattern is based on Standard Italian (Section 18.104.22.168), including a non-case competing null pro. As sketched out before, Longenbaugh (2019) bases his analysis of PPA in Italian unaccusative inversion on two assumptions: first, the existence of a null expletive in Standard Italian; and, second, the assumption, that the Italian null expletive does not carry a case feature. Assuming a non-fixed order of derivational processes for one syntactic head, there are two derivational options: discharge [D] first or discharge [φ: __ ] first. As the second option triggers the obligatory movement of the IA to Spec,vP and subsequently to Spec,TP, resulting in an SV structure, as already seen for the Standard French and Standard Italian data, I will focus on the first option.
After the Merge of v°, the [D] feature is satisfied by the Merge of the Standard-Italian expletive pro. Since pro does not carry a case feature, it is not a case-competitor. Therefore, independent case valuation of the IA does not take place and the IA is still a licit goal for the φ-probe on v°; φ-agreement takes place, while the case feature of the IA remains unvalued resulting in a VS structure in which the participle fully agrees with the postverbal subjects. The postverbal subject is spelled out with nominative case.
|(54)||Insertion of non-case-competing pro, φ-agreement, default case for IA:|
|[TP proi [TP ze [AspP rivadaj [vP proi [vP tj [VP tj [na fia]NOM ]]]]]]|
The analysis of the default agreement pattern is based on French (Section 22.214.171.124), including a case-competing null expletive. The Standard French expletive is pronominal in nature. Thus, it carries φ-features as well as a case feature, qualifying as a case-competitor. Again, as the second derivational option is negligible for the present purpose, I focus on the first derivational option, the immediate satisfaction of [D] after the Merge of v°. In order to satisfy [D], an expletive is inserted in Spec,vP. The peculiarity of the dialects at issue is that they do not only dispose over an expletive pro of the Standard-Italian type, but also over a null expletive of the French type which is pronominal in nature and carries φ-features and a case feature. Due to the Merge of the case-competing expletive in Spec,vP, the necessary syntactic configuration for dependent case valuation is met and the IA is assigned dependent case.13 On the basis of Case Accessibility, the IA is no longer accessible to the φ-probe on v°. The agreement operation of the φ-probe fails and the unvalued φ-features on v° are morphologically realized, according to the language-specific default value, as 3rd person masculine singular.
|(55)||Insertion of case-competing expl, dependent case for IA, no φ-agreement:|
|[TP expli [TP ze [AspP rivàj [vP <expli> [vP tj [VP tj [na fia]ACC ]]]]]]|
Based on Longenbaugh’s (2019) account of French and Italian PPA, the North-Eastern Italian agreement alternation is analysed as the reflex of two different kinds of null expletives:
The case-competing expletive, (i), is pronominal in nature and carries φ-features and an unvalued case feature. Merge of the expletive in Spec,vP yields the necessary syntactic configuration to trigger dependent case valuation of the case feature of the IA, rendering the IA inaccessible to the φ-probe. The other expletive, (ii), does not carry a case feature and is not a case-competitor. Therefore, the case feature of the IA remains unvalued and is spelled out as nominative.
Based on these considerations, the following correlations arise: case-competing expletives appear with accusative-marked IAs and the participle exhibits default agreement. Non case-competing expletives, on the other hand, appear with nominative-marked IAs and the participle overtly agrees with the IA. Now, two questions arise:
Indeed, some Norwegian dialects dispose over two expletives: the pronominal third singular neuter expletive det and another expletive of the locative proform der, cognate to English there (Afarli 2009: 171). The following contrast regarding PPA appears in all unaccusative/passive contexts:
The expletive det correlates with (neuter) default agreement, while the expletive der correlates with PPA with the postverbal subject. Since Modern Norwegian does not mark morphological case, it is not clear which case the postverbal subject bears. Case is morphologically encoded on pronouns, but they are prohibited from expletive constructions due to the Definiteness Effect which is active in Norwegian: definite DPs cannot appear in the postverbal position in expletive constructions.
We can observe similar effects with the German (pronominal) expletive es and the (locative) expletive da:
While the pronominal expletive es correlates with an accusative marked IA and default agreement on the verb,14 the locative expletive da appears with a nominative-marked IA and the verb agrees with the IA. I chose the example with a plural DP, in order to highlight the agreement facts, i.e. the opposition between the plural full and the singular default agreement on the verb. The morphological case marking becomes more telling, if the DP is singular:
These data confirm the predictions coming from the present analysis: the different expletives show the predicted correlation. One kind of expletive (the pronominal es) appears with an accusative-marked postverbal argument and default agreement, while the other expletive (the distal locative da) appears with a nominative-marked postverbal argument and full agreement.
The same correlations show up in a null-subject language like Greek. The opposition between two kinds of expletives that we see overtly marked in Mainland Scandinavian and German, is still there, even in absence of an overt realization for them. But the predicted correlation of accusative case with default agreement and nominative case with full agreement is clearly visible in the Greek example:
These data support the proposal put forward for the dialects at issue. They show that Longenbaugh’s (2019) prediction regarding the nature of the expletive, case and agreement holds in these languages. Since the dialects of Mestre and Gazzolo d’Arcole are null subject languages, the assumed existence of two different kinds of null expletives is not far-fetched.15 As indicated in (62), this predicts that the postverbal subject is marked for nominative in a. and for accusative in b.
Since neither the variety of Mestre nor of Gazzolo is subject to the Definiteness Effect we could theoretically test whether this prediction is born out using pronouns. Unfortunately, the dialects show a syncretism in tonic 3rd person nominative and accusative pronouns, here shown for Gazzolo:
The next question to tackle is whether the appearance of the expletives is restricted to certain lexical and/or pragmatic contexts, as the descriptive data of the dialects of Mestre and Gazzolo d’Arcole implies.
Having put in place the pragmatic and the syntactic analysis, I will now shortly tie together these two ends. As presented in Section 3.1, languages use different formal means in order to disambiguate thetic and categorical sentences. While in French the case-competing expletive il is used, Standard Italian marks thetic sentences with unaccusative verbs by exibiting an inverted order of the constituents, i.e. VS. In the latter case, a syntactic dummy expletive is inserted in the structure: the non-case-competing pro.
The two investigated North-Eastern Italian dialects cannot use the strategy of the Standard Italian type marking categorical sentences by preverbal subjects since a restriction forbids preverbal indefinites. A French-type case-competing thetic expletive compensates this lack of disambiguation by word order. In the preceding section, the different syntactic mechanisms of the insertion of either the thetic or the syntactic expletive pro were explained in detail. Due to a mechanism of case-valuation and the premise of Case Accessibility, the insertion of the different expletives triggers the appearance of the full or defective PPA pattern respectively.
In the present paper, an agreement alternation in unaccusative inversion of two North-Eastern Italian dialects was presented. It was shown that the alternation is not subject to optionality and crucially depends on the pragmatic force of the sentence as well as the pragmatic status of the postverbal DP. In my pragmatic analysis, I proposed that the sentences exhibiting defective agreement are thetic sentences in the sense of Sasse (1987) and Lambrecht (1994). The interpretation of the thetic sentences was connected to the presence of a thetic expletive that is of a syntactic as well as pragmatic nature. Based on Longenbaugh’s (2019) novel approach to PPA, I developed my syntactic account which is based on the existence of two different kinds of expletives in the dialects at issue: a purely syntactic dummy expletive, as the Standard-Italian pro, and a thetic expletive which is a lexical item triggering the thetic interpretation of the sentence. The differing syntactic nature of the two expletives ultimately triggers a morpho-syntactic effect, namely the PPA alternation in the investigated dialects. Ultimately, these dialects show that pragmatics enters into narrow syntax without assuming additional (information-structural) projections in the clausal spine.
1Throughout this paper, I will use the term defective agreement for 3rd person agreement mismatches; it regards the assumed agreement between the postverbal subject and the verb, i.e., the participle. For the moment, it is left open whether the verbal agreement is partial (i.e., only person), no agreement at all with the postverbal subject, or agreement with another possibly covert element.
2I thank an anonymous reviewer for pointing out the potential importance of inter-speaker variation. However, at this point, further specifications regarding inter-speaker variation cannot be made due to the limited number of speakers that have been systematically interviewed during the piloting phase.
3The unfelicitousness of the sentence is due to the unnatural postverbal position of the subject of unergatives.
4I use the term of specificity in the intuitive way denoting a DP that is “intend[ed] to refer to a particular referent, the referent ‘the speaker has in mind’.” von Heusinger (2019: 145).
5As Sasse (1987) and Lambrecht (1994) note, existential sentences are a subtype of thetic sentences. Furthermore, I thank an anonymous reviewer for pointing out that this analysis of thetic sentences, and more particularly of existentials, is opposed to other accounts; most prominently by Moro (1997), who assumes a bi-partition of existentials consisting of a small clause and the element there which he analyses as predicate and not an expletive. Moro (1997) analysis cannot capture the here presented data since he posits the parametrical setting of the pro-drop parameter as reason for different agreement properties (on the basis of Italian and English copular sentences). Since the dialects in question exhibit pro-drop in all clause types and freely allow inversion, I conclude that they dispose over a positive setting of the pro-drop parameter. Thus, Moro’s (1997) account falsely predict that they only allow the full agreement pattern.
7The German Vorfeld-es, located in Spec,CP, can also appear with unergative and transitive verbs but does not trigger verbal agreement, while the “real” expletive es is located in Spec,TP (as shown below by using a subordinate clause) and does trigger agreement. This implies that they might be two different beasts in the end and shouldn’t be confounded.
8There are indeed languages that exhibit PPA with in-situ IAs. For ease of exposition and space limitations, I will omit these data here. The reader is referred to Longenbaugh (2019).
9In this theory, case valuation is not decisive for nominal licensing. A DP can remain unvalued for case in narrow syntax. Then, default case is assigned, i.e., nominative in nominative-accusative languages.
10Following Longenbaugh, this essentially is the difference between movement-based PPA languages and in-situ PPA languages.
12This is the translation of Burzio’s (1986) generalization (θAgent ↔ ACC) to modern terms. The deficient unaccusative/passive v° does not assign a theta-role to its specifier, such that no (external) argument can be introduced in this position. Only expletives can be introduced in this position since they do not require a theta-role. This might lead to the conclusion that there is no [D] feature on unaccusative v°. However, Longenbaugh (2019: 41) argues that it also carries a [D] feature (like transitive v°) presenting an argument for Spec,vP as an intermediate landing site for A-movement in Romance.
13One of the anonymous reviewers underlines the relevance of Belletti’s (1988) account where the postverbal indefinite DP receives partitive case, which is opposed to my account in which the DP bears nominative or accusative case. I decided not to consider Belletti’s account here, because it is not relevant for the presented data: we observe the PPA alternation with indefinite and definite (and specific) postverbal DPs which poses a problem for Belletti’s account, because definite DPs are by definition incompatible with partitive case.
14One of the anonymous reviewers points out that es gibt could be a grammaticalised form. I exclude this possibility on the basis of other instances of es occurring with singular verbal agreement:
15See also Roberts & Holmberg (2010: 22) that state if a language has free subject inversion, it also allows for expletive null subjects.
ACC = accusative, AspP = Aspect Phrase, EXPL = expletive, F = feminine, M = masculine, N = neuter, NOM = nominative, PL = plural, PPA = past participle agreement, RFL = reflexive, SCL = subject clitic, SG = singular, PART = past participle
The author has no competing interests to declare.
Belletti, Adriana. 2017. (Past) Participle agreement. In Martin Everaert & Henk C. van Riemsdijk (eds.), The Wiley Blackwell Companion to Syntax, Second Edition 3. 1–29. Hoboken, NJ, USA: John Wiley & Sons, Inc. DOI: https://doi.org/10.1002/9781118358733.wbsyncom081
Bentley, Delia & Silvio Cruschina. 2018. The silent argument of broad focus: typology and predictions. Glossa: a journal of general linguistics 3(1). 1–37. DOI: https://doi.org/10.5334/gjgl.677
Bianchi, Valentina & Cristiano Chesi. 2014. Subject islands, reconstruction, and the flow of the computation. Linguistic Inquiry 45(4). 525–569. DOI: https://doi.org/10.1162/LING_a_00166
Bobaljik, Jonathan. 2008. Where’s phi? Agreement as a post-syntactic operation. In Daniel Harbour, David Adger & Susana Béjar (eds.), Phi theory: Phi-features across modules and interfaces (Oxford studies in theoretical linguistics 16), 295–328. Oxford: Oxford University Press.
Brandi, Luciana & Patrizia Cordin. 1989. Two Italian dialects and the Null Subject Parameter. In Osvaldo A. Jaeggli & Kenneth J. Safir (eds.), The null subject parameter (Studies in natural language and linguistic theory 15), 111–142. Dordrecht: Kluwer Academic publ. DOI: https://doi.org/10.1007/978-94-009-2540-3_4
Burzio, Luigi. 1986. Italian syntax: A Government-Binding approach. Dordrecht: Kluwer. DOI: https://doi.org/10.1007/978-94-009-4522-7
D’Alessandro, Roberta & Ian Roberts. 2008. Movement and agreement in Italian past participles and defective phases. Linguistic Inquiry 39(3). 477–491. DOI: https://doi.org/10.1162/ling.2008.39.3.477
Deal, Amy R. 2009. The origin and content of expletives: evidence from “selection”. Syntax 12(4). 285–323. DOI: https://doi.org/10.1111/j.1467-9612.2009.00127.x
Deprez, Viviane. 1998. Semantic effects of agreement: the case of French past participle agreement. Probus 10(1). 1–66. DOI: https://doi.org/10.1515/prbs.19126.96.36.199
É. Kiss, Katalin. 2002. The EPP in a topic-prominent language. In Peter Svenonius (ed.), Subjects, expletives, and the EPP (Oxford Studies in Comparative Syntax), 107–124. Oxford: Oxford University Press.
Frascarelli, Mara. 2018. The interpretation of pro in consistent and partial null-subject languages: A comparative interface analysis. In Federica Cognola & Jan Casalicchio (eds.), Null subjects in Generative Grammar: A synchronic and diachronic perspective, 211–239. Oxford: Oxford University Press.
Kayne, Richard S. 1989. Null subjects and climbing. In O. Jaeggli & K. J. Safir (eds.), The Null Subject. Dordrecht: Kluwer Academic Publisher. DOI: https://doi.org/10.1007/978-94-009-2540-3_8
Ladusaw, William. 1994. Thetic and categorical, stage and individual, weak and strong. In Mandy Harvey & Lynn Santelmann (eds.), Proceedings of the 4th Semantics and Linguistic Theory Conference (SALT), 220–229. DOI: https://doi.org/10.3765/salt.v4i0.2463
Lambrecht, Knud. 1994. Information structure and sentence form: Topic, focus, and the mental representations of discourse. Cambrige: Cambridge University Press. DOI: https://doi.org/10.1017/CBO9780511620607
Legate, Julie A. 2003. Some interface properties of the phase. Linguistic Inquiry 34(3). 506–515. DOI: https://doi.org/10.1162/ling.2003.34.3.506
Manzini, Maria R. & Leonardo M. Savoia. 2007. A unification of morphology and syntax: Investigations into Romance and Albanian dialects (Routledge Leading Linguists). London: Routledge. DOI: https://doi.org/10.4324/9780203968154
Moro, Andrea. 1997. The raising of predicates: Predicative noun phrases and the theory of clause structure. Cambridge Studies in Linguistics 80. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/CBO9780511519956
Preminger, Omer. 2014. Agreement and its failures (Linguistic Inquiry Monograph Sixty-Eight). Cambridge, MA: The MIT Press. DOI: https://doi.org/10.7551/mitpress/9780262027403.001.0001
Richards, Marc & Theresa Biberauer. 2005. Explaining Expl. In Marcel den Dikken & Christina Tortora (eds.), The function of function words and functional categories (Linguistik Aktuell 78), 115–153. Amsterdam, Philadelphia: Benjamins. DOI: https://doi.org/10.1075/la.78.06ric
Rizzi, Luigi. 1982. Issues in Italian syntax (Studies in Generative Grammar 11). Dordrecht, Holland: Foris Publ. DOI: https://doi.org/10.1515/9783110883718
Rizzi, Luigi. 2005. On some properties of subjects and topics. In Laura Brugè, Giuliana Giusti, Nicola Munaro & Walter Schweikert (eds.), Contributions to the 30th Incontro di Grammatica Generativa, 203–224. Venice: Cafoscarina.
Roberts, Ian & Anders Holmberg. 2010. Introduction: parameters in Minimalist theory. In Theresa Biberauer, Anders Holmberg & Ian Roberts (eds.), Parametric variation: Null subjects in Minimalist theory. Cambridge: Cambridge University Press.
Sasse, Hans-Jürgen. 1987. The thetic/categorical distinction revisited. Linguistics 25. 511–580. DOI: https://doi.org/10.1515/ling.19188.8.131.521
Sauerland, Uli. 2003. Intermediate adjunction with A-Movement. Linguistic Inquiry 34(2). 308–313. DOI: https://doi.org/10.1162/ling.2003.34.2.308
Silva-Villar, Luis. 2010. Subject positions and the roles of CP. In Armin Schwegler, Bernard Tranel & Myriam Uribe-Etxebarria (eds.), Romance linguistics: Theoretical perspectives: selected papers from the 27th Linguistic Symposium on Romance Languages (LSRL XXVII). Philadelphia: J. Benjamins.
Svenonius, Peter. 2002. Subject positions and the placement of adverbials. In Peter Svenonius (ed.), Subjects, expletives, and the EPP (Oxford Studies in Comparative Syntax), 199–240. Oxford: Oxford University Press.
von Heusinger, Klaus. 2019. Indefiniteness and specificity. In Jeanette Gundel, Barbara Abbott & Klaus von Heusinger (eds.), The Oxford Handbook of Reference, 145–167. Oxford University Press. DOI: https://doi.org/10.1093/oxfordhb/9780199687305.013.9