Disjunctive clauses with o. o ‘either. or’ in Spanish and clausal cosubordination

1 Introduction

Two clause combining types are widely accepted in the Spanish grammatical tradition: coordination and subordination. However, a significant number of linked clauses are not adequately encompassed by these two types (see Conti 2012; Conti 2014; for an overview). ^[1] Examples of clause linkages that resist the traditional analysis include adjunct clauses, adversative clauses, comparative constructions and distributive clauses (see Section 2). This paper aims to shed light on this complex area, focusing in particular on examples like (1) ^[2]:

As this paper aims to demonstrate, two main features distinguish the type of clause shown in (1) from both coordinate and subordinate finite clauses: they are mutually dependent and they lack their own set of clausal operators. In view of these properties, o… o-clauses will be classified as examples of cosubordination, a third type of linkage that is widely accepted in cross-linguistic works but has yet to be applied to Spanish syntax. In fact, proposals offering alternatives to the “coordination-subordination” dichotomy in Spanish have failed to consider examples like (1), which thus far have been analyzed as coordinate clauses (Fukasawa 1985; RAE/ASALE 2009: 2445–2446). For instance, García-Berrio (1969–1970) considers comparative, consecutive and conditional clauses to be interdependent, i.e. mutually dependent, but does not address disjunctive constructions like (1), which are also reciprocally dependent. Similarly, Rojo (1978) (see also Narbona 1983; Narbona 1989; Blesa 1984; Moya 1996) argues that adverbial and adversative clauses are interordinated (semantically dependent but not integrated) but he does not take into account examples like (1).

This paper will use Conti’s tests (2014) to classify o… o-clauses and to determine the extent to which they are or are not coordinate. These tests are designed to recognize coordinate and subordinate clauses in Spanish and are based mainly on clause-level phenomena. Such phenomena are rarely employed to recognize linkage types in Spanish, despite the fact that several typologists have praised their usefulness in characterizing clause combinations cross-linguistically. In fact, the classificatory proposals for Spanish that have been put forward to date often hinge on the pre-established nature of the linking marker (e.g. coordinate vs. subordinate conjunctions). Only very occasionally do they rely on specific clause-phenomena (Conti 2014).

This paper will present some problematic linked clauses in Spanish before going on to describe subordinate and coordinate clauses according to Conti’s tests (2014). It will then illustrate the properties of o… o-clauses and demonstrate that they are cosubordinate before presenting the relevant conclusions and describing some future aims.

2 Complex sentences: classificatory problems in Spanish grammar

As noted above, adjunct clauses (2a)–(2b), comparative constructions (3), adversative clauses (4) and distributive clauses (5) have prompted several classificatory proposals in the Spanish literature dealing with the different clause-linkage types:

Most works consider adjunct clauses such as (2) to be subordinate (see, for instance, Alarcos 1994: 354–358 and Real Academia Española 2009: 3530), but they have been also classified as coordinate (Marcos-Marín 1980: 369–70 for conditionals) or as paratactic constructions (Jiménez-Juliá 1995: 25). Some scholars even argue that clauses such as (2) are neither subordinate nor coordinate but rather interdependent (García-Berrio 1969–1970: 225) or interordinated (Rojo 1978: 126–127; Narbona 1983; 1989; Blesa 1984: 41–42; Moya 1996: 38–40). In general terms, interdependency and interordination are intended to account for clauses that are mutually dependent (from either a semantic or a distributional perspective) but are not syntactically integrated into the main clause.

Comparative constructions like (3), which traditional grammars consider to be instances of clause subordination (RAE 1973: 543–546; Alarcos 1994: 340–347), have also been analyzed as coordinate structures (Sáez del Álamo 1999: 1144). In fact, comparative and coordinate constructions (in contrast to subordinate clauses) allow the ellipsis of non-focal constituents (e.g. Luis compró dos libros y Pedro tres ‘Luis bought two books and Pedro three’, Juan compró más libros que Pedro ‘Juan bought more books than Pedro [bought books]’).

The analysis of adversative clauses with pero ‘but’ (4) is also controversial. In contrast to academic grammars, which classify adversative constructions under coordination (Real Academia Española 1973: 510; Real Academia Española 2009: 2452), other scholars treat them as interordinated or interdependent, akin to adverbial clauses (Rojo 1978: 108; Narbona 1989: 111–112; Blesa 1984: 41–42; Moya 1996: 38–44). In the case of the exclusive adversative sino ‘but not’ (4b), only Echaide (1974–1975: 20) has noted its proximity to subordination, and even then only in passing.

The examples in (5) range from juxtaposition to coordination, depending on the treatment of the linking marker, which is interpreted as a lexical word in some works (Alarcos 1994: 317; Myre 1998: 31) and a coordinate conjunction in others (Real Academia Española 2009: 2414).

These classificatory problems beg some crucial questions regarding clause combining in Spanish. In particular, if a significant number of complex constructions cannot be easily classified as coordinate or subordinate, are coordination and subordination really sufficient to account for them? One possible solution, posited by several scholars (Haiman and Thompson 1984; Lehmann 1988; Van Gijn et al. 2011), is to consider coordination and subordination the poles of a scale. While some constructions match all of the properties attributed to these poles, others, which only partially satisfy them, lie somewhere in between. According to Van Gijn et al. (2011: 7), this is the case with adjunct adverbial clauses, cosubordinate clauses and paratactic subordinate clauses.

Moreover, cosubordination has been proposed as a third linkage type within functional-typological grammar by Van Valin (1984), Van Valin and LaPolla (1997), Van Valin (2005) and Hengeveld and Lachlan Mackenzie (2008), to account for units that are dependent but are not integrated. As Sections 5 and 6 show, this paper adopts this theoretical assumption to explain o… o clauses and other related constructions.

3 Subordinate and coordinate clauses in Spanish

4. The special case of o… o-clauses

5 An alternative analysis of o… o-clauses: Clausal cosubordination

According to several scholars, cosubordinate clauses are dependent, like subordinate clauses, but are not embedded, like coordinate clauses (see Van Valin 1984: 546; Foley and Van Valin 1984: 242; Van Valin and LaPolla 1997: 454; Van Valin 2005: 188). Furthermore, cosubordination is defined by operator dependence or conjunct illocutionary scope, as noted by Bickel (2010: 52). In particular, Van Valin (2005: 201) argues that cosubordinate clauses depend on the same matrix of clause operators, like evidentiality, status (epistemic modality, external negation), tense, and illocutionary force (Van Valin 2005: 8–11). Cosubordinate clauses also seem to denote an episode or a sequence of events showing some sort of continuity in time, space and participants (Hengeveld and Mackenzie 2008: 157).

Cosubordinate clauses are commonly illustrated with medial clauses from Papuan languages (51). ^[12] Medial clauses are dependent and host verbs that lack some of the verbal categories (mainly mood) but they are not embedded. They are part of a sequence of clauses that ends with the so-called final clause, which, in contrast to medial clauses, contains a fully inflected verb that may function as a sentence. Quite often, medial verbs host a marker indicating whether the subject is co-referential with the subject of the final clause (like the sequential conjunction -re in [51]):

Even though Indo-European languages do not attest the sort of structures illustrated in (51), the notion of cosubordination has progressively spread to some constructions from English, French, and Spanish. For instance, Van der Auwera (1997: f. 3) suggests that mutually dependent clauses in correlative structures like The harder you run, the sooner you will get home are cosubordinate. For Spanish, Mora (2006: 46–47) notes in passing that juxtaposed clauses are cosubordinate, arguing that they share tense, modality, and polarity. More recently, Conti (2016: 21–22) has claimed that adversative clauses with sino que (restrictive but) are also cosubordinate: they allow the ellipsis of non-focal elements and backward-control ellipsis but reject the imperative mood and cannot be asserted independently.

Van Valin (2005) takes a more detailed look at other constructions from English and French, which he includes within clausal cosubordination. In particular, he considers clauses like laughing loudly, in Pat ran down the hall laughing loudly, and bought some beer, in Leslie drove to the store and bought some beer, to be cosubordinate (Van Valin 2005: 198). Notice that these clauses are prosodically dependent, suppress their subject, and depend on the operators of the previous clause.

Unlike English, in which cosubordinate clauses show internal marking of both grammatical and distributional dependency (like the occurrence of the converb in the former example and the subject suppression in the latter), Spanish clauses in o… o-structures are not so easily classifiable as cosubordinate: firstly, o… o-clauses may function as sentences outside of the correlative structure; that is to say, if we attend only to their form (leaving aside correlative markers), the clauses illustrated in Section 4.2 do not have any markers of dependency that indicate that they are defective. In addition, it is hard to imagine that fully inflected verbs like those in (36)-(50) above depend on just one matrix of clausal operators as is expected for cosubordinate clauses. In short, it seems necessary to employ additional grammatical criteria in order to demonstrate that the clauses in o… o-structures are both desententialized (like subordinate clauses) and not embedded (or structurally parallel, like coordinate clauses).

As shown above, o… o-clauses cannot express a command via the imperative mood and cannot be asserted independently. Thus, in contrast to coordinate clauses and like subordinate clauses, they are “defective” as regards illocutionary force. In addition, the fact that in most examples o… o-clauses share tense, mood, and illocutionary force may be evidence of their dependency on the same matrix of operators. However, it should be noted that clausal operator sharing does not always occur in o… o-structures. In particular, the occurrence of different tenses in examples like (52) suggests that the clauses within the structure have their own set of operators:

In example (52) it is not possible to interpret both of the clauses as being dependent on the same matrix of operators, which suggests that (a) the units in the correlative structure are sentences and (b) the relationship between these sentences is not cosubordinate. ^[13]

6 Possible future additions to this study: Other cases of cosubordination in Spanish, with a particular focus on correlative constructions

Similarly to o… o-clauses, other correlative structures, traditionally classified as coordinate or juxtaposed, could be treated as instances of clausal cosubordination. In particular, other clauses with disjunctive or copulative correlative markers like ni… ni ‘neither… nor’, bien… bien ‘either… or’, ya… ya lit. ‘whether… or’, etc. (see Ni estudia ni trabaja ‘(S)he neither studies nor works’), as well as clauses that are part of lexical correlations (as in Él se lo guisa y él se lo come lit. ‘He cooks it and he eats it’ or Uno reía, otro lloraba ‘One was laughing, the other was crying’) are also symmetrical and desententialized: they are dependent on each other, cannot be asserted and tend to share tense, mood, and illocutionary force.

In addition, y-clauses in backward-control deletion structures like El escritor redactaba ___i y revisaba las obras_i (‘The writer wrote and reviewed his plays’), which are traditionally treated as cases of coordination, can also be classified in a different manner. Even though these clauses are not desententialized (in fact, they allow the imperative mood; Limpia ___i y recoge tu cuarto_i ‘Clean and tidy up your room’), backward deletion gives rise to a sequence of mutually dependent clauses. The former, which would hold the deleted complement, is dependent on the joint clause (see the incompleteness of *El escritor revisaba ‘The writer reviewed’). The latter, like other joint clauses, is also dependent from a distributional point of view (it cannot occur by itself, as seen in *Y revisaba las obras ‘And he reviewed his plays’) and cannot be anteposed, *Y revisaba las obras, el escritor redactaba ‘And he reviewed the plays, the writer wrote’). Furthermore, unlike coordinate clauses these structures allow questioning: El escritor redactaba ___i y revisaba las obras_i, ¿Qué_i redactaba y revisaba el escritor ___i? (‘The writer wrote and reviewed his plays’, ‘What did he write and review?’).

In short, it seems that, in Spanish, the correlative constructions which have thus far been classified under coordination or juxtaposition display features of grammatical dependence that evince their hybrid nature. If we also take into account Spanish adversative clauses with sino que and restrictive aunque, which Conti (2016) treats as cosubordinate, then cosubordination would seem to be an interclausal relationship of enormous significance in Spanish grammar. Before this can be affirmed, however, each of the constructions highlighted must be studied in detail as many of the classification problems that still persist in grammatical studies of complex constructions in Spanish result from the use of few and often skewed examples that have not been obtained from samples of real language. Due to space constraints and for the purpose of explanatory expediency, detailed study of these constructions is beyond the scope of this paper.

Finally, it should be noted that the relevance of cosubordination in explaining and describing numerous complex Spanish constructions cannot be reduced solely to constructions originally considered coordinate or juxtaposed. Rather, it can be extended to other examples, such as the relationship between non-finite forms and main predicates in Vi salir a María – in which the infinitive is not a complement – and Bajaba las escaleras llorando, in which the gerund does not express manner.

7 Conclusions

O… o-clauses have been analyzed in the literature to date as coordinate structures, in accordance with the treatment of o… o as coordinating conjunctions. However, most instances of o… o-clauses behave differently from both coordinate and subordinate clauses. In contrast to coordinate clauses, o… o-clauses show traces of desententialization: they are mutually dependent, reject the imperative mood and cannot be asserted. In contrast to Spanish subordinate clauses, o… o-clauses share illocutionary force, mood and tense. This means that, on the one hand, no clause in the structure can be interpreted as head; and on the other, all clauses are co-indexed and depend on the same matrix of clausal operators as is the case in clausal cosubordination. In summary, the fact that o… o-clauses are both desententialized and structurally parallel undoubtedly points to their special status within clause combining in Spanish.

These clauses, whose behavior is quite systematic and predictable, cannot be accounted for by traditional clause combining types. This paper has attempted to show that o… o-clauses (and probably other clauses with disjunctive or copulative correlative markers like ni… ni ‘neither… nor’) are most often cosubordinate, which in turn entails implicit acceptance of cosubordination as a relevant clause combining type in Spanish syntax.