As is well known, Spanish allows object drop when the antecedent is an indefinite bare noun (1), but not when it is a definite determiner phrase (DP) (2).Footnote ¹

Following Huang (Reference Huang1982) and Raposo (Reference Raposo1984), Campos (Reference Campos1986) proposes that in sentences like (1B) the argument position is occupied by a wh-trace of an operator OP that moves in the syntax.

As evidence for this analysis, Campos argues that object drop in Spanish is subject to the same locality constraints that hold for wh-movement. Concretely, he claims that object drop is island-sensitive. Campos offers the following examples and grammatical judgments:

Besides the theoretical validity of Campos’s analysis, the claim that object drop in (‘Standard’) Spanish exhibits island effects has survived unchallenged to the present dayFootnote ² and is largely assumed in the literature (see, for instance, Landau Reference Landau2010: 383; Gribanova Reference Gribanova2013: 110; Rothman & Iverson Reference Rothman and Iverson2013: 595; Armstrong Reference Armstrong2016: 13; Cyrino Reference Cyrino2019: 18, among many others). However, this characterization is not correct: none of the sentences in (4B)–(7B) is actually ungrammatical.

Campos’s examples are problematic for two different reasons. On the one hand, some of them contain dialogues that are pragmatically odd. This is the case of (5) and (7), which include a relative clause island and an adjunct islandFootnote ³ respectively. As can be observed in the following examples, given a proper discourse context, null objects can easily occur within these domains:

Furthermore, note that the answers in (5B) and (7B) are still anomalous even if they contain an overt accusative clitic referring to a definite antecedent, as in (10B) and (11B).

These cases severely weaken the claim that the oddness of (5) and (7) is related to object drop phenomenon or to syntactic constraints (i.e. islands). As said before, the unacceptability of these sentences and the ones above seems to be due to general discourse factors, namely, to the fact that none of these utterances constitutes a relevantFootnote ⁴ answer for the corresponding question. To illustrate, consider again the dialogue in (7) or its variant with a definite object in (11). In both cases, the assertion we were able to go into the cinema because we found (the) tickets clearly cannot be taken as a felicitous answer to the question did you find (the) tickets for the movie? Now, compare these dialogues to the one in (9a). In this case, this problem is avoided: the assertion do not bring bread because we have already bought some is a relevant answer to the question should I bring some bread? Thus, once pragmatic factors like relevance are properly controlled for, object drop can occur within an island domain without difficulty.

On the other hand, other examples proposed by Campos are not only grammatical, but also pragmatically acceptable. Consider first (4), which includes a complex NP island. According to my informants and to my own native judgment, the answer by speaker B is totally correct, especially if the main predicate existe el rumor de que ‘there exists the rumor that’ is interpreted parenthetically, i.e. if it functions as a kind of evidential which signals the source and reliability of the embedded claim (Simons Reference Simons2007). Once again, it is relatively easy to find similar and even more natural cases:

Finally, Campos’s example in (6), which includes a subject island, can be considered an analogous case: besides being grammatical, the answer in this dialogue is also felicitous. What makes this case a bit more tricky than the previous one is that here the assertion by speaker B requires a very particular interpretation, given that the anteposition of the clausal subject in Spanish typically triggers a contrastive topic reading.Footnote ⁵ Therefore, in order to make the utterance more natural, the example should contain a continuation that explicitly contrasts with the clausal subject, as in (14). Again, object drop can clearly occur within this strong island.

In sum, it can be concluded that object drop in Spanish is not island-sensitive. This means that Spanish behaves just like other well-studied languages, such as Greek, in which (indefinite) null objects can appear in island domains (Dimitriadis Reference Dimitriadis1994; Panagiotidis Reference Panagiotidis2002). From a theoretical point of view, this observation constitutes a non-trivial problem for an analysis of object drop in Spanish based on movement, and opens the possibility for a null pronoun approach (e.g. Giannakidou & Merchant Reference Giannakidou and Merchant1997) or a verb-stranding vP-ellipsis account (e.g. Merchant Reference Merchant, Merchant, Mikkelsen, Rudin and Sasaki2018).Footnote ⁶