Author:

Abstract:

In classical Spanish, que ‘that’-deletion, e.g. (1), was quite frequent with a wide range of predicates, whereas in contemporary Spanish, que can only be omitted in subjunctive complement clauses mainly following volitional verbs. (1) Por ende vos rogamos [Ø] le dedes entera fee y creencia. (anon. 1497) ‘Thus we beg you [that] you give him full trust and belief’ Pountain (2015) suggests that que-deletion is a Latin calque typical of formal genres in the 16th century, which slowly disappears in the 17th century, and that the type of matrix verb governing que has an important influence on the alternation. Blas Arroyo & Miralles (2015, 2016) indicate that omission is especially common in lower prestige authors. This paper proposes a different parameter to be the main predictor of the alternation, namely the use of the subject in the subordinate clause (Spanish is a pro-drop language) and whether or not the subject is co-referential with the indirect object of the main clause (e.g. co-referentiality of vos ‘you’ and the subject of dar in (1)). In line with Givón’s (2001) complementation scale, we claim that co-referentiality between subject and indirect object leads to increased syntactic integration, which then allows for the deletion of the complementizer. The parameter of co-referentiality is related to the covert expression of the subordinate subject, as with ditransitive rogar ‘request’ in (1). The Esbozo grammar (RAE 1973) mentions that deletion is possible with ditransitive verbs exclusively when the subordinate clause has no explicit preverbal subject. Blas Arroyo and Miralles (2015) confirm this tendency for the non-ditransitive verb creer (‘to believe’), emphasizing the adjacency of the main and the subordinate predicates as a basis for omitting the complementizer. However, it has been shown that with many other verbs deletion was favoured by an overt subject, which is then claimed to mark the boundaries between the clauses, making the presence of the conjunction redundant (see also Pountain 2015:79). Hence, both the absence and presence of an overt subject seem to play a role in que-deletion depending on whether it concerns a ditransitive construction or not. We conduct a logistic regression analysis of some 3000 tokens from the historical CODEA+-corpus (GITHE, 2015) in order to predict the relationship between the dependent binary variable (que vs ø) and multiple independent variables (presence and position of overt subjects, presence and type of objects of the main verb --clitic pronoun, lexical expression, stressed pronoun, doubling-- that co-refer with the subject of the subordinate verb), so as to account for the origins, spread and decline of que-deletion. Our results indicate that que-omission is highly favoured by the proximity of the predicates, as well as by the features of a prototypical manipulation schema (Givón 2001): subjunctive mood in the complement clause, agent of the complement clause encoded as indirect object of the main verb and implicit in the complement clause, coreferentiality of indirect object and complement subject. Thus, the paper highlights that an explanatory analysis of que-deletion needs to go beyond the parameter of the lexical semantics of the matrix verb (volition, fear), analyzing also the prominence of its objects.