This paper is about a particular use of the German modal sollte (‘should’) in the antecendent of conditionals as illustrated in (1)–(3). We call this use the “deliberative” use of sollte. We argue that on its deliberative use sollte doesn’t behave as the weak necessity modal it is standardly assumed to be. The distributional facts suggest that the use conditions of sollte-antecendents are closely related to the use conditions of conditional antecendents with the complementizer falls (‘in case’). Following a recent proposal by Hinterwimmer for falls, we propose that sollte in the antecendent of a conditional introduces a use condition that takes the truth of the antecendent proposition to be a truly open possibility against a given conversational background.
We would like to thank the audiences at the Semantics and Pragmatics Workshop at Mie University, the Semantics Research Group Meeting at Keio University, the Oberseminar English Linguistics at Göttingen University and, in particular, Elin McCready, Shinya Okano, Jan Köpping, Osamu Sawada, Joe Tabolt, Hubert Truckenbrodt, Thomas Weskott and Ede Zimmermann for helpful discussions and comments. This work has been funded by the 2018 JSPS Summer Program.
The topic of this paper is the use of the German modal sollte in the antecendent of conditionals on a reading that we call “deliberative”. This use is illustrated in (1)–(3).Footnote 1Footnote 2
What is interesting about these examples is that sollte doesn’t seem to contribute any additional modal meaning to the antecendents – or at least it doesn’t seem to contribute its usual interpretation as a (deonticFootnote 3 or epistemic) weak necessity modal. The plot of the paper is as follows: First, we introduce some background on the modal sollte and its interpretations as deontic and epistemic weak necessity modal. Second, we argue that the use under discussion is neither a deontic use, nor a “conventional” epistemic use. We show that the use conditions of sollte on the relevant reading are in most respects identical to the use conditions of conditionals with the complementizer falls (‘in case’) on [2]’s account: the contribution of should to the antecendents of a conditional is a presupposition that restricts its use to a conversational background in which the antecedent proposition is a truly open possibility. In the last two sections, we discuss occurences of deliberative sollte in relative clauses and the relation of deliberative sollte to its epistemic use.
2 Some Background on the German Modal sollte
sollte is an inflected form of the modal sollen. Its closest counterpart in English is should. Morphologically, the form sollte can either be the past tense form of sollen or its past subjunctive form (German: “Konjunktiv Präteritum” or “Konjunktiv II”). Usually it is assumed that the form of sollte in the use under discussion is its subjunctive form, see [4]. The subjunctive modal sollte is considered to be a weak necessity modal, see [4] for a detailed discussion and empirical tests that support this assessment. Accordingly, [4] propose the following classification of the German modal forms muss and sollte in agreement with the corresponding classification of the English modals must and should.
Like other modal verbs in German (and English), sollte is polyfunctional, i.e., it can be used as a deontic modal, (4), or as an epistemic modal, (5).
3 The Deliberative Use is Not an Epistemic or Deontic Use
3.1 The Deliberative Use is Not a Special Case of a Deontic Use
The first question to ask is whether the deliberative use of sollte is a special case of an deontic use or an epistemic use of sollte. The first alternative of these is not very plausible to begin with. The closest we find to a deontic use of sollte in a conditional antecendent is a use that refers back to a given recommendation or a previous use of deontic sollte.
In any case, this is not the use that we are interested in.Footnote 4
More evidence that the deliberative use is not a special case of a deontic use comes from the fact that we find deliberative uses of sollte in the antecendent of conditionals with progressive aspect and forms of the German “Perfekt” that are strongly marked on a deontic interpretation for sollte without an additional specification of a temporal reference point.
3.2 The Deliberative Use is Not a Special Case of an Epistemic Use
In this section, we will present our reasons why we think that the deliberative use is not a special case of a use as an epistemic weak necessity modal. First, while other modals that allow for an epistemic use can be substituted for sollte in an underspecified context like in (9) (ignoring for the moment the subtle differences in conditions of use and meaning), these modals cannot be substituted for deliberative sollte as illustrated in (10) – irrespective of their modal strength in a corresponding context.Footnote 5
Second, there clearly is no local interpretation of sollte in the sense of ‘there is a weak epistemic necessity that p’ as in the unembedded case.Footnote 6
That we don’t find local epistemic interpretations for sollte doesn’t already decisively show that sollte couldn’t be interpreted epistemically. As [7] shows for the reportative use of the indicative form soll, we sometimes find global interpretations of the relevant modal element. Global uses can be paraphrased as parentheticals. The corresponding paraphrases would be as follows:
At first sight, this looks like a reasonable interpretation. But we also find examples of the following kind:
A paraphrase that assumes a global parenthetical interpretation results in a clash in the second conjunct as can be seen by the following paraphrse:
The given context indicates that the matter of whether the person under discussion is found guilty or not is not settled yet and therefore cannot be known. But still the interpretation of should is deliberative in the relevant sense. We take these examples to show that we do not have any commitment at all to the (global) truth of the proposition on a deliberative reading – not even a weak one.
Other examples that can help to make the same point are examples with explicit parentheticals that deny any commitment, as in (17), uses with the focus sensitive particle selbst (‘even’) that indicates that the antecendent proposition is the least likely of the relevant propositions in the alternative set, as in (18), and the modal particle doch that indicates that the antecendent proposition is not in agreement with what was previously assumed or expected, as in (19).
It seems that the deliberative use in these examples marks that it an open question whether the antecedent proposition holds. We take this to show that the deliberative use is not a global epistemic use.Footnote 7
4 The Use of Deliberative sollte is Not Simply a Way to Express Subjunctive Mood
If sollte on its deliberative use doesn’t have its usual interpretation as a weak necessity modal, what does it contribute? [4] suggest that the use of sollte is maybe a way to express subjunctive mood (Konjunktiv). This would be in accordance with its characterization as “hypothetical” in reference grammars of German.
“[German] reference grammars discuss a special use of sollte that often occurs in conditionals [...]:
The meaning contribution as hypothetical (Zifonun, Hoffmann and Strecker 1997b: 1893) might suggest that the Konjunktiv II is semantically interpreted here, while the stem of the modal is not semantically interpreted.”
Here are some reasons why we think that the assumption that sollte is just a way to express subjunctive mood doesn’t get the distributional facts right. First of all, substituting Konjunktiv for sollte doesn’t always lead to an adequate paraphrase.
In the DWDS corpus search, we found for the first 84 occurences of deliberative sollte in the antecendent of a conditional 51 occurences of indicative mood in the consequent (including 12 cases of reportative present subjunctive mood that are not interpreted as conditional subjunctive mood), and 26 occurences of conditional subjunctive mood (including 7 occurences of modal verbs in subjunctive mood); the rest being infinitival and imperative forms. So the combination of deliberative sollte in the antecendent with indicative mood in the consequent doesn’t seem to be an exception from the rule.
Third – and this is the most important aspect – the use conditions of sollte-antecedents are more restricted than the use conditions of plain subjunctive antecedents. In fact, the use conditions of sollte-antecedents match the use conditions of conditional antecedents with the complementizer falls (‘in case’) in German.Footnote 8
5 sollte-antecendents and Degrees of Commitment
In this part of the paper, we are going to show that the use conditions of sollte-antecedents are more narrow than the use conditions of subjunctive conditionals. We will show this by testing whether the use of deliberative sollte is acceptable against the background of a particular degree of commitment by the speaker to the truth of the antecedent proposition. We consider the following range of possible (modal) commitments of the speaker to the truth of the antecendent proposition.
This discussion follows a similar discussion in [2] for the German conditional complementizer falls.
5.1 Factual Conditionals
Conditionals with sollte-antecendents cannot be used as factual conditionals (‘given (the fact) that’/‘assuming that’) – even if the consequent clause is in indicative mood. This is similar to falls according to [2].
5.2 Strong Epistemic Necessity
Conditionals with sollte-antecendents cannot be used if the antecedent proposition is considered to be an epistemic necessity. We illustrate this point with epistemic muss (‘must’) in (24).
5.3 Circumstantial Necessity
Conditionals with sollte-antecendents cannot be used if the antecedent proposition is considered to be a circumstantial necessity – as for example with promises:
We assume that, given the promise, p is a circumstantial necessity, i.e., for all future situations compatible with the (relevant) circumstances now (including the promise) that are most normal: it is the case that p. Circumstantial necessity associated with promises is in conflict with the use conditions of sollte.
5.4 Weak Epistemic Necessity
If the antecendent proposition is given in the discourse context as a weak epistemic necessity, introduced by the use of the weak epistemic necessity modal müsste in the example in (26), then the use of sollte in the antecendent of the conditional is possible.
5.5 Varying Degrees of Possibility
The use of sollte as in (28) is fine against the background of discourse contexts in which the antecendent proposition is given as a possibility with varying degrees of commitment.
5.6 Counterfactual Conditionals
sollte-antecedents cannot be used in counterfactual contexts for the antecedent-proposition.Footnote 9
In summary: Our discussion of the data supports a similar conclusion as [2] reaches for falls-antecedents: Deliberative sollte seems to require that the antecedent proposition is a “truly open possibility” against a given epistemic (or circumstantial) conversational background.Footnote 10
6 The Proposal
We propose the following semantics for sollte in its deliberative use (here illustrated for the use in the antecendent of a conditional).Footnote 11
f and g can – but don’t have to – be the relevant conversational backgrounds for the interpretation of the conditional. g is a stereotypical ordering sourceFootnote 12 and f is either an epistemic conversational background or a circumstantial conversational background.
The meaning contribution of deliberative sollte is purely presuppositional. It presupposes that the antecendent proposition is a simple possibility with respect to the conversational background f in the world of the world of evaluation w and at the same time it must not be a human necessity – in the terminology of [3] – with respect to the modal base f and the ordering source g in the world of evaluation.Footnote 13
7 Deliberative sollte in Relative Clauses
There is another context were we typically find deliberative readings for sollte: free relative clauses.
As already seen with conditional antecendents in (10), other modals that in principle allow for epistemic interpretations cannot be substituted for sollte.
We also find deliberative readings for sollte in restrictive relative clauses to universal quantifiers, (33), “generic” indefinites, (34) and plural definites, (35).
Interestingly, all these sentences seem to have a modalized or generic interpretation. This is supported by the observation that with none of these sentences there even has to be a single individual of which the main predication of the relative clause is true. The meaning of (32) can be paraphrased by the conditional in (36).
If (32) had the truth-conditions of (36), the fact that the main predication of the relative clause doesn’t have to be true of any individual would readily be explained since the conditional in (36) gives wide scope to sollte with respect to the existential quantifier jemanden (‘someone’)/the generic pronoun einem. At this point, we don’t have more to say about the use of deliberative sollte in relative clauses.
8 Conclusion
In this paper, we have discussed a particular use of the German modal sollte in the antecendent of conditionals that we called “deliberative”. We presented arguments that the deliberative use of sollte is not a special case of a deontic use or an epistemic use of the modal sollte. By going through a range of contexts with varying degrees of (modal) commitment of the speaker to the truth of the antecendent proposition, we could show that the use of sollte marks the antecendent proposition as a truly open possibility against a given conversational background. The results are summarized again in the left table on the next page. We proposed that sollte on its deliberative use introduces a presupposition that restricts the use of the conditional to conversational backgrounds in which the proposition is given at least as a simple possibility and at most as a weak necessity.
In this final section, we want to take a step back and end with a few comments. As for the content of the presupposition: We are aware that the proposal is tailored to fit the observations and doesn’t give us any deeper explanation. The main point of the proposal is to precisely illustrate the contrast between the deliberative use of sollte and its epistemic use. Secondly, we are aware that the status of the condition of use as a presupposition hasn’t sufficiently been argued for. Since the use conditions of sollte-antecedents seemed to us to be similar to the use conditions of mood marking in conditionals and since mood marking is usually associated with a presupposition, we assumed that sollte contributes a presupposition, too. More interesting than the details of our proposal is a pattern that emerges from our generalizations and that might even shed new light on sollte (and English should) on its epistemic use: While the epistemic weak necessity reading of sollte is considered to be part of the asserted content and seems to be (mostly) restricted to syntactic root contexts, deliberative readings appear to be presuppositional and are restricted to non-root context.Footnote 14 The table summarizes this pattern.
This pattern lets one wonder whether one should look out for a single sollte after all that flips its interpretation depending on its context of use.
To end on a speculative note: If we were to assume that the common core to deliberative and epistemic sollte is the contribution it is not the case thatpis a strong necessity, it would be intruiging to think of the assertion of sollte in a root context as coming with an exhaustification of the scale of graded modalities resulting in a reading as a weak necessity modal as suggested in the right table below. Since it is not at all clear how such an exhaustification should come about and what should account for its obligatoriness in root clauses, this is mere speculation at this point.
Notes
1.
We assume that everything we say in this paper holds in the same way for wenn...sollte- antecedents, as in (1) and (2), as for sollte-V1-antecedents, as in (3). Wherever we choose to illustrate a point with a wenn...sollte- antecedent, we might aswell have chosen a sollte-V1-antecedent and the other way around. We will refer to both types of antecedents as “sollte-antecedents”.
2.
[4] report that “English should shows the same reading.” As for example:
As in German, one also finds conditional antecendents with should in first position. More data from English can be found in Daan Van den Nest’s dissertation:
(ii)
Should they use what is regarded as excessive or unnecessary force, they, too, might well become the targets of aggression. Daan Van den Nest (2010)
3.
We use the term “deontic modal” here in a rather wide way corresponding to what [6] calls a “priority modal”.
4.
As in the example in (6), the comparative adverbials besser (‘better’) and lieber (‘preferably’) can in principle always accompany a deontic use of sollte. It cannot accompany the relevant deliberative use.
(i)
Wenn das dir (*lieber/*besser) zu früh sein sollte, dann komm einfach später. If this should (*preferably/*better) be to early for you, then just come later.
5.
We don’t want to say that modals that in principle do have epistemic interpretations never occur in the antecendent of conditionals, see [5] for a discussion. But these uses seem to be rare. The rareness of real epistemic readings of modals in the antecendent of conditionals is confirmed by a comprehensive corpus search in the DWDS subcorpus “DWDS-Kernkorpus (1900–1999)” (https://www.dwds.de/; date of search: October 06, 2018).
6.
The assumed paraphrase is of course a simplification. For concrete proposals of the meaning of English should as a weak epistemic necessity modal/normality modal: see [1] and [8]. The same point could be made if we were to assume a similar contribution for the German modal sollte on its epistemic use as [1] and [8] assume for should.
7.
In a later paragraph, we will argue that certain occurences of sollte in relative clauses are also deliberative uses of sollte. With these examples, it can be clearly seen that the deliberative use of sollte is not an epistemic use since we also find clear cases of epistemic uses in (appositive) relative clauses.
In contrast to (i), the reltative clause in (ii) clearly has an epistemic interpretation.
8.
For a detailed discussion of the use conditions of conditional antecedents with the complementizer falls see [2].
9.
Here is the only difference we found to the use conditions of falls-antecendents: Since falls-antecendents can in principle be marked with additional subjunctive mood, we find a difference in certain counterfactual contexts. Against the same background as (29), the falls-antecendent is fine:
This example shows that Hinterwimmer’s generalization that falls-antecedents cannot be used in counterfactual contexts for the antecendent proposition has to be modified. At the same time, it seems to be the right generalization for sollte-antecendents.
10.
[2]: “falls seems to require that the speaker considers the antecedent proposition to be a truly open possibility.”
11.
This is very close in spirit to the proposal in [2] for falls.
12.
Hinterwimmer also assumes a stereotypical ordering source in the context of his proposal for falls.
13.
We use a syncategorematic meaning rule in (30-b) since this is the direct way to spell out our proposal. Here is the non-syncategorematic rule:
This semantic rule gives us an interesting additional insight since it forces us to distinguish between the local world of evaluation and the local context world (for which we write “\(c_{\textsc {w}}\)”). This might have to be reconsidered in the light of the considerations at the end of this paper.
14.
Deliberative sollte could be characterized as an anti-root-phenomenon. This is the reason why we introduced the condition that the local world of evaluation must be different from the world of the local root context in our definition, compare condition (i–c) of footnote 13 .
References
Copley, B.: What should should mean. Manuscript. CNRS (2006)
Kratzer, A.: The notional category of modality. In: Eikmeyer, H.J., Rieser, H. (eds.) Words, Worlds, and Contexts, pp. 38–74. de Gruyter, Berlin (1981)
Schenner, M.: Double face evidentials in German: Reportative ‘sollen’ and ‘wollen’ in embedded contexts. In: Groenn, A. (ed.) Proceedings of SuB12, pp. 552–566. ILOS, Oslo (2008)
Sode, F., Sugawara, A. (2019). On the Deliberative Use of the German Modal sollte.
In: Kojima, K., Sakamoto, M., Mineshima, K., Satoh, K. (eds) New Frontiers in Artificial Intelligence. JSAI-isAI 2018. Lecture Notes in Computer Science(), vol 11717. Springer, Cham. https://doi.org/10.1007/978-3-030-31605-1_25
