THE OBJECT GAP-PSEUDORELATIVE GENERALIZATION

Previous literature contains two different points of view regarding the subjectobject asymmetry related to the DP head of pseudorelatives (PRs). Some authors claim that the DP head can only be interpreted as the subject of the embedded predicate (subject-gap PRs). Other authors point towards the possibility of finding other constituents (e.g. direct object) in head position (object-gap PR), too. In this paper I claim that there are certain languages that only allow the DP head to be the subject of the embedded predicate, that is, they only allow subject-gap PRs, whereas other languages allow both subject-gap and object-gap PRs. Thus, the aim of this paper is to present the object-gap PR generalization to account for the cross-linguistic availability of subject-gap and object-gap PRs: the availability of object-gap PRs is subject to object clitic doubling. The structure of this paper goes as follows. Section 1 introduces PRs. Section 2 presents data about subject-gap and object-gap PRs. Section 3 gives some remarks on object clitic doubling. Section 4 presents the object-gap PR generalization. Conclusions and further research issues are presented in section 5.


Introduction
Pseudo-relatives (PRs)1 are constructions formed by a DP (the head) plus an embedded clause headed by the complementizer que ʻthatʼ (1).Traditionally, the DP head is interpreted as the subject of the embedded predicate.
( On the other hand, Graffi (1980:133) points out that the insertion of an object clitic coindexed with the DP head would ameliorate the degree of acceptability of the object-gap PR, but still, the construction remains ungrammatical.( 9 [Graffi 1980: 133] Up to this point, it can be said that there are certain languages e.g.Spanish and Greek, that allow object-gap PRs if an object clitic is inserted within the PR predicate.The absence of the object clitic makes object-gap PRs ungrammatical.Besides, there are other languages such as Italian that do not allow object-gap PRs even in the presence of the object-clitic.Thus, what do Spanish and Greek have in common to favor the availability of object-gap PRs through the insertion of object clitics?And why do languages like Italian block the insertion of the object clitic making objectgap PRs unavailable?

Some remarks about Clitic Doubling
According to Anagnostopoulou (2006: 520) clitic doubling is a construction in which a clitic co-occurs with a full DP in argument position forming a discontinuous constituent with it inside the same propositional structure (Kallulli & Tasmowski 2008:1) Bošković (2008) provides a list of languages that allow clitic doubling.The list includes the following: Albanian, Macedonian, Bulgarian, Greek, Somali, Spanish, French (some dialects), Catalan, Romanian, Hebrew, Arabic, Dutch.However, clitic doubling displays intriguing cross-linguistic variation which has been widely discussed in the literature.In particular, some languages such as Spanish or Romanian have clitic doubling of objects, but others such as French or Italian lack this type of construction (Anagnostopoulou 2006: 520) 3) -( 8), an important similarity between object-gap PRs and object-clitic doubling constructions arises: the same languages (French, Italian and Portuguese) that do not allow object-gap PRs do not allow object clitic doubling constructions, and those languages that allow object clitic doubling (Spanish or Greek) also allow objectgap PRs.
Since the presence/absence of the object clitic coindexed with the head of the PR appears to be the key for the availability of object-gap PRs, then I propose the objectgap PR generalization in ( 12): ( 12) Those languages where PRs and object clitic doubling are available allow object-gap PRs.Those languages where PRs are available, but lack object clitic doubling do not allow object-gap PRs.
In fact, once ( 12) is applied to the whole PR picture, it can be observed that PRs are not only restricted to the subject or the DO of the embedded predicate, but they are also available with the indirect object (IO) of the embedded predicate ( 13).( 13 The contrast between (13a) and (13b) shows that again, if the dative clitic le is inserted within the que-clause (13a), the construction results grammatical.However, if the dative clitic is not present within the que-clause, the construction results ungrammatical.In other words, PRs where the DP head is interpreted as the IO of the embedded predicate are also available (at least in Spanish) and are subject to the mandatory presence of the object clitic.  1 is the result of applying ( 12) to the cross-linguistic PR panorama.The first column in table 1 shows languages where (subject-gap) PRs are available 8 .The third and fourth columns show the correlation between the availability of object clitic doubling and object-gap PRs as explained in (12).
In the case of Spanish, Galician 9 , and Greek, object-gap PRs are available.These languages agree with (12) since all of them have object clitic doubling.
Although clitic doubling is available in Catalan, as well as strong pronoun doubling, clitic left dislocation and right dislocation (Bošković 2008;Tsakali & Anagnostopoulou 2008; Complements duplicats Fitxa 7674/3 2014; Gramàtica de la llengua catalana 2016), Catalan does not seem to agree with (12).However, it is interesting that a mandatory intonation break needs to be present in clitic doubling structures 10 as well as after the head of the PR 11 .Further research needs to be done to see if there is any connection between these two phenomena. 4See Kallulli & Tasmowski (2008), Hill & Tasmowski (2008) and Alboiu et al. (2013) for more information about Romanian object clitic doubling. 5The question mark points out that further research needs to be done to fully confirm the availability of object-gap PRs because parallel syntactic phenomena out of the scope of this paper interfere. 6The availability and display of object clitic doubling changes depending on the different varieties of Dutch.See van Craenenbroeck & van Koppen (2008) and Koopman & Sportiche (2014). 7See Runic (2014); Bošković (2008Bošković ( , 2016) ) for a discussion about Serbo-Croatian clitic doubling and argument ellipsis.See Mis̆eska (2008) for a detailed discussion about the variability of object clitic doubling in standard Serbian and in the South-Eastern Serbian varieties. 8Adapted from Grillo & Costa (2014: 133 Italian, French and Portuguese also agree with ( 12), in the sense that neither of them allows object clitic doubling and thus, object gap PRs are unavailable too.
Korean and Japanese lack clitics 12 , and thus, the prediction that follows from ( 12) is that they lack object-gap PRs.However, further research needs to be done to discard other syntactic strategies that might lead to object-gap PRs.
The cases of Dutch, Romanian and Serbo-Croatian are more complex because clitic doubling is subject to other syntactic factors such as argument ellipsis, article availability (Bošković 2008(Bošković , 2016;;Runic 2014), lack of clitic doubling for fully lexical DPs or the differences between subject/object clitic doubling (van Craenenbroeck & van Koppen 2008: 223-225).Additionally, clitic doubling structures behave differently depending on the varieties of Dutch or Serbo-Croatian (Mis̆eska 2008: 80-83).Thus, further research needs to be done to check whether the predictions following from (12) for Romanian, Serbo-Croatian (they should allow object-gap PRs), and Dutch (Dutch should not allow object-gap PRs) are fulfilled and if they are not, what grammatical issues interfere.

Conclusion and further research.
Previous literature on PRs presents a subject-object asymmetry regarding the syntactic nature of the DP in head position with respect to the embedded predicate.In this squib I present the object-gap PR generalization that relates the availability of object-gap PRs and object clitic doubling, namely, those languages that allow object clitic doubling, allow object-gap PRs, whereas those languages that lack object clitic doubling do not allow object-gap PRs.Further research includes extending and testing the cross-linguistic availability of indirect object-gap PRs or finding an explanation for the cross-linguistic availability of PRs.Besides, it would be important to see whether Japanese and Korean reach object-gap PRs using alternative syntactic strategies, if Catalan intonation breaks are related to object clitic doubling phenomena or if the predictions made by the object-gap PR generalization hold for Romanian, Serbo-Croatian and Dutch.
Nuria Aldama García Laboratorio de Lingüística Informática Facultad de Filosofía y Letras Universidad Autónoma de Madrid nuria.aldama@estudiante.uam.es PRs have been studied from very different angles since the late 19 th century Angelopoulos (2015) with the head of the PR is present, the structure is grammatical (7a).This is what I call from now on object-gap PR 3 .If the DO clitic is removed, as in (4b, 7b) the structure is again ungrammatical.María who was brought to the faculty by carʼAngelopoulos (2015)independently observes that DO-gap PRs are available in Greek.Greek DO-gap PRs are subject to the same condition as Spanish DO-gap PRs: the presence of the DO clitic coindexed with the head of the PR is mandatory. . .

Table 1 .
Relation between object-gap PRs and object-clitic doubling.
Anagnostopoulou (2006)oulou (2006)doubled objects have the same intonation and distribution of arguments while right-dislocated objects have the intonation and distribution of peripheral elements.11Thepause after the PR head turns the PR into an appositive relative clause.