Uncertainty as a disambiguating factor for conditional, temporal, and habitual constructions: evidence from Spanish

1 Introduction

Most studies on conditionality in Spanish have focused on the reality of the event as indicated by the verb morphology in both the protasis and the apodosis. Conditionals have been categorized in a bipartite manner – real and not real (e.g., Álvarez 2008; Pelly 1998) – or in a tripartite manner – real, potential, and not real (e.g., Lavandera 1984; Nowikow 1995). In Spanish, regardless of the type of categorization, these divisions often use the tense and mood of the verb to decide the (un)reality of the event (e.g., Rivas 1990; Vesterinen 2016). Hence, a protasis in the indicative would be regarded as real or possible, and a protasis in the subjunctive would be considered not real or counterfactual. This article is not so much focused on the differences among the various types of conditionals, as on the relationship between conditionality and other domains of meaning, the potential ambiguity of some constructions, and how the notion of uncertainty can help us disambiguate between these meanings.

Conditionality, as a semantic notion, shares many semantic and morphosyntactic properties with causal, temporal, and habitual clauses. Most conditional clauses share with causal clauses a cause-effect relation between the protasis – cause – and the apodosis – consequence (e.g., Deza Enríquez 1993; Rooij and Schulz 2019). For Spanish, Rodríguez Rosique (2008: 215) proposes a causality axis that includes causal, conditional, and concessive clauses where there is a blurred line between these semantic domains, making a clear-cut distinction among them difficult. The difference between causal and conditional clauses relies on the link between cause and effect (Deza Enríquez 1993; Comrie 1986). Causal clauses (e.g., estoy cansada porque corrí 10 km ‘I am tired because I ran 10 km’) have a positive relation where something happens because something else is done and the speaker commits to the truth of the two clauses (that is, the embedded clause is entailed). In conditional clauses (e.g., si me gano la lotería viajaré a Turquía ‘If I win the lottery, I will travel to Turkey’), on the other hand, the condition is that situation without which something would not be realized, and the speaker does not commit to the truth of either clause. These clauses are usually, but not exclusively, linked by a connective such as because, in the case of causal clauses, and if, in the case of conditional clauses.

As stated by Rodríguez Rosique (2008), conditionality and causality are intricately intertwined in Spanish. For instance, Schwenter (1999) analyzes si ‘if’-clauses that resemble causal structures, like that in (1) where si would be best translated with because, but only in its epistemic uses.^[1]

Moreover, some conditional constructions (CCs) usually have temporal and causal restrictions that must be met to have a felicitous interpretation. The conditional interpretation of (2) also has a causal connection between the antecedent and the consequent where getting good grades causes me to give you a car.

Equally, they are temporally linked as the gift comes after you get good grades. These causal and temporal restrictions are precisely what can allow some CCs to acquire these meanings.

Furthermore, CCs are intricately related to temporal and habitual clauses. In Spanish, some temporal connectives can have three main interpretations: temporal, conditional, or habitual (see Akatsuka 1986; König 1986; Liu 2019 for English and German examples). For instance, cuando in (3) can be interpreted as ‘when’ I have free time, ‘whenever’ I have free time, or ‘if’ I have free time.

At the very least, (3) is ambiguous between these three readings. The use of si no ‘if not’ in the following sentence, which indicates what would happen if the previous condition does not hold, can also help us make a case for a conditional reading. Some CCs like that in (4) can have a habitual iterative meaning that can be paraphrased with whenever as in whenever you smoke you cough (cf. Montolío Durán 1999: 3727).

Examples like (4) also demonstrate the tight causal-temporal relation of the clause where smoking causes the cough and coughing comes after smoking.

In this paper, I argue that the only way of disambiguating between these meanings is by incorporating uncertainty or the speaker’s lack of commitment to the truth of the antecedent into a definition of conditionality. This uncertainty feature is not only a trait of temporal connectives with conditional interpretations, but it is a characterizing factor of all conditional constructions. We will see, for instance, that some uses of the canonical conditional marker si ‘if’ are also ambiguous. I will also provide four tests that will help us discern between conditional, temporal, and habitual meanings.

The remainder of this paper is organized as follows. In Section 2, I make a case for the inclusion of uncertainty as a defining characteristic of CC. Section 3 discusses the structural similarities of conditional, temporal, and habitual constructions, and Section 4 provides a series of tests to differentiate between these three meanings. Finally, Section 5 summarizes the main points made in this paper.

2 Uncertainty

Following Mauri (2008) and Elliott (2000), the present work also takes uncertainty^[2] to be a heterogeneous semantic class that is marked differently in different languages (e.g., modal adverbs, interrogatives, question markers, etc.) but can also remain unspecified. If a construction is unspecified, then the default interpretation will be that of certainty and only the context can impose an uncertainty reading (Mauri 2008: 31). Attributing uncertainty to conditionals is not a new idea. Grice (1975 [1967]), for instance, indicates that by using if, the speaker is not certain that the antecedent is true: if they were, they would have used since. The claim is that conditional constructions always denote uncertainty, but not all constructions denoting uncertainty are conditional. For instance, the modal might in She might go to the park indicates that the speaker does not have enough evidence to state that she will go to the park.

Some conditional markers encode uncertainty in their semantics. When a speaker uses a verb such as suponer ‘to suppose’ or imaginar ‘to imagine’ they are signaling that they are not sure that what they are saying holds in the actual world. Suponer means to “to consider as truth or real something that is not or does not have to be” (RAE 2014; my translation). The verb itself invites us to create an alternative mental space in which something would hold, and it postulates this alternative as uncertain because we are only considering it is as true for the sake of making a point. As pointed out by a reviewer, in speech, it is not just the lexical content of the verbs themselves, but also pauses or the intonation given to verbs that can convey uncertainty or doubt on the part of the speaker (e.g., Dehé and Wichmann 2010).^[3] In (5), for example, we are considering the cost for each employee to be 25,000 euros, but we are not sure that this is actually the case.

Furthermore, if someone were to utter supongamos ‘let’s suppose’ and then have a long pause, we would understand just by hearing that word that they are about to say something that is not true in the actual world. The difference between verbs like imaginar and suponer and other conditional connectives is that they are realized as independent clauses, whereas connectives appear in dependent clauses in a subordinating relation with a main clause.

Conditional connectives like si ‘if’, on the other hand, do not convey uncertainty in their semantics, rather this meaning is pragmatically inferred through a generalized conversational implicature. Adding a level of uncertainty to CCs will allow us to distinguish constructions that are disguised as conditional, but that do not behave as such. The condition in (6), for instance, introduces an adversative relation in which si ‘if’ would be better translated as but (see Schwenter 1999; Castroviejo and Mayol 2019).

Examples (7) and (8) have been considered in previous research as echoic conditionals, where the consequent is entailed and the antecedent echoes information uttered in the previous discourse (e.g., Castroviejo and Mayol 2019). In this work, however, they are not considered conditional, as they behave more like causal clauses. As was indicated in the previous section, causal clauses entail their antecedent, and it is because examples (7) and (8) entail the antecedent that they cannot be classified as “true” conditionals: there is no doubt (no uncertainty) that they hold in the actual world. This can also be confirmed by the impossibility of translating example (8) with the equivalent if in English (*if I passed the exam …), which points to the fact that if and si do not show the same linguistic behavior in all their uses.

In these examples, the discourse context tells us that speaker A’s letter writing in (7) caused speaker B to write a letter too, and speaker B’s passing the exam in (8) causes her to celebrate.

I am aware that most studies on conditionality in Spanish treat factual protases as true conditionals (see Montolío Durán [1999] for an exception), primarily because they are introduced by si ‘if’. However, due to the diverse nature of conditional protases available to speakers, the only way to distinguish other types of CCs (e.g., those without a conditional connective) from different types of clauses (e.g., temporal) is by including a level of uncertainty. At a first glance, (9) presents two declarative clauses that introduce two events or facts. However, if we look at them more closely, depending on the context these juxtaposed clauses can yield other meanings like causality (9a), conditionality (9b), or habituality (9c).

The only way to discern between these three notions is through uncertainty. Of course, one needs more context to determine the speaker’s intentions.

In this section we saw that CCs introduce an antecedent that is always uncertain. Including uncertainty as an integral part of conditional meaning will help us identify CCs that arise from connectives that have other prototypical interpretations (e.g., temporal) or CCs without an overt connective. This topic is further explored in the following section.

3 Conditionality, causality, temporality, and habituality

Crosslinguistically, conditionals can be interpreted with a “temporal value and temporal conjunctions are often best interpreted conditionally” (Harris 1986: 429). Thompson and Longacre (1985: 193) argue that the distinction between when and if in English is “one of degree of expectability” (i.e., uncertainty). Wenn in non-past clauses in German can also yield either a temporal or a conditional meaning (König 1986), which underscores the links between conditionals and temporal markers. Another overlap between temporal, causal, and conditional clauses comes from Finnish, where even though the language possesses a conditional marker (jos) speakers also use kun ‘when’ to convey temporal, causal, and conditional meanings (Vilkuna 2014). This also holds true for the Spanish word cuando ‘when’. When interpreted as conditional, these connectives are always ambiguous, and only the discourse context will allow us to disambiguate. Another property of these connectives is that when they acquire a conditional interpretation, they often refer to generic truths.

Temporal conjunctions that denote a simultaneous event (as opposed to a successive one) are likely to acquire a conditional interpretation in which what is postulated in the consequent is processed in relation to the content of the antecedent (Harris 1986; König 1986; Kratzer 1995; Montolío Durán 1999; Vilkuna 2014). In fact, one of Traugott’s (1985) sources of conditional markers are lexical items with temporal meanings like so long as in English and siempre y cuando in Spanish. Thus, prototypical temporal connectives can have temporal, habitual, or conditional meanings, as we can see with cuando ‘when’ in (10).

Connectives that are more grammaticalized like siempre y cuando ‘as long as’ often select a verb in the subjunctive in Spanish, as in (11), which helps solidify the uncertainty reading of the clause. However, when the speaker uses a verb in the indicative, it yields a habitual reading, as in (12).

In example (12) the use of siempre y cuando with the indicative verb tengo ‘I have’ changes the meaning to its literal interpretation ‘always and when’ (that is, whenever). The same holds true for other temporal connectives. The RAE (2010) and Montolío Durán (1999) indicate that mientras ‘while’ must be followed by the subjunctive to get a conditional reading. Furthermore, besides the possibility of acquiring a conditional interpretation, some temporal connectives can also have a habitual reading if speakers are asserting a generalization, like the example in (13) which can also mean ‘whenever I have time’.

Similarly, the use of si in (14) can have a temporal, conditional, or habitual reading paraphrased as when we are late, if we are late, and whenever we are late, respectively (see also Montolío Durán 1999: 3727). Notice that this is not exclusive of connectives, as this ambiguity also arises in juxtaposed constructions that are not explicitly linked. The popular saying querer es poder, literally ‘to want is to be able to do’, can have multiple interpretations like temporal ‘when you want something …’, habitual ‘whenever you want something …’, or conditional ‘if you want something …’.

This section delved into the intricate relationship of conditionality, causality, temporality, and habituality. We saw that certain constructions can have multiple readings, that is, they are ambiguous. The following section proposes four tests that will helps us disambiguate between these meanings.

4 Disambiguating tests

As we saw in the previous sections, some constructions are potentially ambiguous between conditionality and other meanings, such as temporal or habitual readings. This section proposes four tests that can help us determine the possible interpretation(s) that these constructions can have. These tests will help us determine whether (a) the antecedent is uncertain (i.e., whether there is a conditional reading) or (b) the habitual or temporal value of the clauses.

The first test is used by Foley and Van Valin (1984) to distinguish subordinate clauses from coordinate clauses by testing the illocutionary force of a clause with a question: if the question can be answered, then we are faced with a coordinate relationship, if not, we have a subordinate clause. In (15a) the question addressing the content in the second clause can be positively answered, whereas in (15b) there is not enough evidence to answer yes.

This test can also allow us to identify possible conditional readings. For example, if we can answer (16b) and (17b) positively, then we are faced with a temporal or a habitual reading.

If we cannot answer it positively then we get, pragmatically, a conditional reading where we do not have enough evidence to commit to the truth of the antecedent. Thus, if I cannot identify specific instances where I have time in (16) and where you pay for dinner in (17), then we have two CCs. For instance, in (17) we can imagine a context where we are deciding on plans, the budget, and who is going to cover the costs of our plan. We then decide that we are going to split the costs and the condition under which I would pay for the movies is if you pay for dinner. At the moment of the utterance, I can ask Do you pay for dinner?, and this cannot be answered positively, because I do not know if you will agree.

The second and probably most direct test for determining the uncertainty of some constructions is asking Are you sure that X? Going back to example (16), I can ask you Are you sure that you have time to go out with your friends? If the answer is anything other than a conclusive yes, then we have a CC.

The third test seeks to differentiate between (habitual) conditional and temporal/habitual readings, and it is a little more difficult to administer as one must imagine a broader context by asking How many times has that happened? Imagine that you and your sister are at a party and your friends ask you to stay a little longer, and you respond with (18) or (19). Then your friends ask you: How many times has that happened?

If you can answer with any numerical value or with always or whenever we are late, then you are faced with a temporal or habitual meaning. However, if you cannot think of any instances where this has actually happened or if you do not have enough evidence to back your claim, then we are faced with a CC. It is important to mention that it is also possible for these examples to have a habitual-conditional reading, that is, we are scolded whenever we are late, and we do not know if we are going to be late on this occasion. That is, the habitual and the conditional readings are not mutually exclusive.

The aim of the fourth test targets the different readings that some temporal connectives can have. This test is concerned with the possibility of adding certain temporal adverbs to some of these connectives. For instance, although cuando is considered a temporal connective, when it is combined with a verb in the present tense, it is impossible to add a temporal adverb like mañana ‘tomorrow’, like in (20a); the same is true for other temporal adverbs like temprano ‘early’ or hoy ‘today’ (hoy ‘today’ is only acceptable with the generic reading of ‘nowadays’). Thus, the use of the present tense with the connective cuando can only have a reading other than temporal, in this case, conditional or habitual. To get a temporal interpretation, the verb in the protasis must be changed to the subjunctive, as in (20b), and the verb in the apodosis to a future form.

The addition of these temporal adverbs in Spanish is not felicitous (see Hoff [2019] for an exception in Argentinian Spanish). As such, the use of the present tense with cuando in both the protasis and the apodosis results in either a conditional or a habitual reading. In order to differentiate between these two we must go back to the first test and ask the question Do I have time? Another adverb that we can use to differentiate between a habitual and a conditional meaning is siempre ‘always’. If you can add it then you will have a habitual interpretation meaning ‘whenever’, if you cannot then you will have a conditional reading.

This section briefly presented four tests to determine the uncertainty value of CC. Since these CCs are ambiguous, the only way to distinguish them from other meanings is through the inclusion of uncertainty as a general property of conditionals. As with the other connectives, the argument is that the uncertainty value arises through a pragmatic inference.

5 Summary

In this paper, I have made the case for the inclusion of uncertainty as a characterizing trait of CCs. I have argued that factual protases, because they are not uncertain, are not “true” conditionals. The need to include uncertainty as a feature of CCs arises from the ambiguity that exists between conditionality and other notions. Although this paper has focused on the ambiguity between conditionality, temporality, and habituality, conditionality is also ambiguous with other domains of meaning like causality and adversativity. As such, the tests presented in Section 4 could also serve to disambiguate conditionality from these latter notions.