Intersubjective strategies in deontic modality: evidential functions of Spanish deber ‘must’

The principal aim of this study is to examine the Spanish modal verb deber ‘must’ in its deontic readings, relating it to the notions of evidentiality and intersubjectivity. Deber has often been compared to the modal verb tener que ‘have to’ and described in rather vague terms, for example as an expression of weak, internal obligation, but this paper proposes that it is better understood as an intersubjective verb. Both quantitative and qualitative analyses have been carried out, with a special focus on the in-depth qualitative study. It will be shown that deontic deber can convey evidential meanings when used in the conditional form. First, it can refer to a norm shared between the speaker and the hearer, and, second, it can convey an inferential process, a conclusion presented by the speaker, which is based on shared information, available to a larger group (or all) of the interlocutors. Evidentiality is regarded here as an intersubjective strategy, used when the speaker wants to reach consensus, arguing for the most reasonable, morally defensible way to act. Thus, this study also offers a new perspective of evidentiality, looking at this notion in interaction with deontic modality instead of epistemic modality, which is usually the case.


Introduction
The notion of modality has received a substantial amount of attention within linguistic research during the last few decades. Especially the category of modal verbs has been investigated in a large number of languages worldwide, Spanish being no exception.
However, in spite of the considerable quantity of descriptions of Spanish modal verbs, many researchers offer a somewhat vague and simplified picture of this linguistic category, not least when trying to explain the differences between the verbs deber 'must' and tener que 'have to'. These two verbs are the two most frequently used when expressing obligation and necessity 1 in Spanish (Bauman 2013) and they are often considered as near synonyms, being interchangeable in certain contexts (Gómez Torrego 1988;Olbertz 1998;Bauman 2013). When accounting for the differences between deber and tener que, the former tends to be described as a weaker, moral obligation 2 leaving the door open for alternative actions, while the latter is associated with strong, external necessity which requires action and is impossible to avoid. This paper aims to offer a new perspective on especially one of these verbs, arguing for an intersubjective understanding of deber, contrasting with the more subjective tener que. As will be shown in the analysis, deber expresses two types of evidential meaning when used in the conditional tense. The evidential use of debería 'should' is considered here as an intersubjective strategy used in order for the speaker to orient him/herself towards the interlocutors. The paper will be organized in the following manner. Section 2 will provide the theoretical background of the study, namely a summary of previous studies of deber and tener que, a brief introduction to the research field of evidentiality, a short description of two modal functions associated with the conditional tense, and a definition of the notion of intersubjectivity. In section 3, the data and method are briefly described. Subsequently, in section 4, the results are presented and commented upon, consisting of both quantitative data and individual examples. Finally, in section 5, the analysis is summarized and some conclusions are drawn, based on the previous discussion.

Theoretical background
This section offers a theoretical basis for the present study. Section 2.1 briefly summarizes previous studies of deber and tener que while section 2.2 is dedicated to the research field of evidentiality. Section 2.3 describes two important modal functions of the Spanish conditional tense, and how they are presented in recent studies. In section 2.4, the notion of intersubjectivity is defined and outlined, focusing on its interaction with modality. 1 In modality research both necessity and obligation are used when referring to the core meaning of some of the modal verbs. In this article necessity will be used as the cover term for the verbs deber and tener que in their deontic readings. 2 In this article only deontic modality is considered. The epistemic and dynamic readings of these modal verbs are therefore not regarded here. For a detailed discussion about the relationship between dynamic and deontic modality in the use of deber and tener que, see Thegel (2017).

Previous studies of deber and tener que
As stated before, Spanish modal verbs have been studied to a fair extent especially, during the last thirty years. However, as Cornillie (2007, 5) puts it, being labelled as "verbal periphrases", the modal verbs tend to be regarded as a subgroup within a larger field of verbal constructions, which means that the stress is often put on their syntactic behavior instead of their semantic-pragmatic properties. As far as deber and tener que are concerned, formal, syntactic qualities such as word order, interaction with negation and the combination with other verbal periphrases have been taken into account (cf. Gómez Torrego 1988;Silva-Corvalán 1995;RAE 2009).
Studies that focus on the semantic nature of the two verbs often describe deber as an expression of moral obligation, with a weaker level of necessity than tener que, which is considered to convey an inexorable necessity, leaving the modal subject 3 without any other option than to act (cf. Sirbu-Dumitrescu 1988;Olbertz 1998;Müller 2001;2005;RAE 2009). Müller (2001) applies a substitution test in order to argue for a difference in strength between deber and tener que, which is reproduced below: (1) Siento *deber/tener que molestar-le a usted. feel-1sg *should/have to-inf bother-inf-obj acc you 'I am sorry that I *should/have to bother you.' (Müller 2001, 57) According to the author, the use of tener que in this context conveys the meaning "I am sorry, but there is no other way out than bothering you" while deber would receive the interpretation "I am sorry, and actually I could have chosen another solution than bothering you" (Müller 2001, 58). Therefore, tener que is perfectly fine in the actual context, whereas deber would not be an alternative, sounding "pragmatically contradictory" (Müller 2001, 58).
From my point of view, the reason why tener que is appropriate while deber sounds odd in (1) is not due to a difference in strength but to the fact that the two modal verbs express different types of modal necessity 4 . I regard the notion of strength as a problematic parameter, because of its diffuse nature and the vague way in which it has frequently 3 Modal subject is used here to refer to the person or entity upon which the necessity is imposed, i.e. the responsible person for carrying out the action regarded as necessity.

4
Deber, when appearing in a non-epistemic context, almost exclusively receives a deontic interpretation, as in example (1); it is rare in dynamic modality. In contrast, tener que can convey both deontic and dynamic modality, the last type being the case in (1), expressing an inevitable, factual situation (Thegel 2017). been used (cf. Verhulst 2012, Verhulst et al. 2013, for a more detailed description of the previous use of strength), often applied without a defining description or criteria.
Another parameter that has been taken into account when examining the differences between non-epistemic functions of deber and tener que is the notion of internal or external obligation. Deber is regarded as an expression of obligation stemming from internal factors, often referring to a situation considered as "necessary and convenient for everyone" (Fernández de Castro 1999, 186, my translation), including even the modal subject itself (Sirbu-Dumitrescu 1988, 141). The following example is offered by Sirbu-Dumitrescu (1988) to support her view: (2) Pues a mí concretamente el casamiento de Jacqueline (Kennedy -n.n) well to me concretely the marriage of Jacqueline me ha parecido pues bastante normal ya que es una me have-3sg seem-prf well quite normal because be-3sg a mujer como otra cualquiera y…por tanto pues debe woman like other anyone and…for so-much well must-3sg buscar… una satisfacción para ella seek-inf a satisfaction for her 'In my opinion, Jacqueline's marriage seems, well, quite normal, as she is a woman like any other and…therefore…well…she must look for…a satisfaction for herself.' (Sirbu-Dumitrescu 1988, 144) According to Sirbu-Dumitrescu (1988), the obligation in (2) conveys a beneficial action for the subject in question, which will have a positive influence in her life and health, i.e., deber in this case refers rather to a convenient action than a pure necessity. In contrast, tener que is frequently described as an expression of external, severe obligation, indicating a necessity imposed by circumstantial factors or a strong authority (Sirbu-Dumitrescu 1988;Fernández de Castro 1999;RAE 2009). Fernández de Castro (1999 gives the example in (3) to explain the difference: (3) Tiene que/debe pagar las cuentas con la justicia have to must-3sg pay-inf the bills with the justice 'S/he has to/must face justice.' (Fernández de Castro 1999, 186) The author states that tener que conveys an external obligation, clearly in conflict with the will of the subject, whereas deber expresses an internal obligation, a recommended action that is the most convenient one. Unfortunately, the broader context of the example is not provided, which makes it extremely difficult to test Fernández de Castro's proposal.
A striking observation, when examining previous studies of deber and tener que, is the fact that many researchers resort to intuition-based examples when explaining differences between the two verbs, or, which is also problematic, offer a very limited linguistic context, thus making a proper interpretation impossible. This study offers new empirical data, aiming to contribute to a more nuanced description of especially deber (and more indirectly tener que) in order to explain their differences.

The notion of evidentiality
Evidentiality has been defined as "the kinds of evidence a person has for making factual claims" (Anderson 1986, 273), or the quality of the evidence put forward for an epistemic qualification (Nuyts 2001, 386). In many languages worldwide, especially in non-Indo-European languages such as Kashaya, Tibetan, or Japanese, evidentiality is marked grammatically, often by adding a suffix to the verb (Anderson 1986;Willet 1988). In contrast, in other languages, for example Spanish, the notion of evidentiality is manifested rather through "evidential strategies", which means that "categories whose main meaning do not reflect information source can acquire evidential extensions" (Aikhenvald 2007, 210).
Researchers within this field differentiate between several types of evidence, originating from direct and indirect sources, respectively. Willet (1988), for instance, organizes evidentials according to three main categories, namely, attested evidence, reported evidence and inferring evidence. The first category can be further divided into subgroups such as visual evidence or auditory evidence. For the two other categories, distinctions are sometimes made between second-hand knowledge and common information (Plungian 2001), or between inferences based on physical observations and inferences based on logic reasoning (Willet 1988).
When it comes to the relationship between evidentiality and modality, researchers tend to agree on the close connection between the former notion and epistemic modality; however, what they disagree on is how these concepts are connected. While some authors, like Aikhenvald (2007), regard them as two distinct categories, Willet (1988), van der Auwera & Plungian (1998) and Cornillie (2007) consider them to be partly overlapping, especially in the case of inferential evidentiality (or modality). As far as Spanish is concerned, the inferential value of deber has been studied by Cornillie (2007) and Hennemann (2013), concluding that the non-deontic readings of this verb often express an inductive, inferential process.
While several studies in Spanish focus on the relationship between epistemic modality and evidentiality, there is, according to my knowledge, no research dedicated to the interaction between the latter notion and deontic modality, which is the aim of the present article. The in-depth analysis of this study, where two evidential functions of the conditional tense will be outlined, is inspired by Bermúdez (2006;2016). His model, reproduced in Figure 1, is based on a cognitive semantic understanding of evidentiality: Bermúdez does not use the classical categories of evidentiality, such as those described in Willet (1988), but presents a deictic view of the domain of evidentiality, based on the model of spatial deixis. The three general parameters of that model, namely direction, distance, and reference points, can be translated into manner of access to the information, distance to the information source and access to the information source when adapted to the domain of evidentiality. The first parameter considers the way in which the information is obtained, where one pole refers to purely sensorial evidence, for example something that was heard or seen, whereas the other pole refers to cognitive evidence, obtained through a mental process. The second parameter takes into account the distance to the information source from the speaker's perspective, where an internal source is equivalent to the speaker him/herself, i.e. first-hand information, while an external source concerns information from someone else, be it second-hand or even third-hand information. Finally, the third parameter regards the exclusivity of the access of the information: it is exclusive if the speaker presents information as new to the interlocutors, but could also be universal, referring to e.g. folklore. A point in between the two poles would be a situation in which the speaker and some other people, but not everyone, share a piece of information. As will be explained in the analytical chapter, the two evidential functions found for debería 'should' are related to the first and the third parameter in Bermúdez' model.

Modal functions of the conditional tense in Spanish
The conditional tense in Spanish has received a considerable amount of attention recently, with several authors focusing on its modal functions, especially those related to reported information, the so-called condicional de rumor 5 (Hennemann 2013;Kronning 2015;Bermúdez 2016), and to politeness (Vatrican 2011;. The first type usually appears in newspaper discourse and is associated with epistemic modality or evidentiality, where it conveys that the information is not first-hand and that the speaker cannot guarantee its certainty. As Hennemann (2013, 351) puts it: The journalistic conditional's primary meaning is to indicate foreign text import. The speaker, or rather journalist, may be fairly committed, or may not be committed or may take a position in between regarding the transmitted information. But it is assumed that he is never totally committed… Hennemann (2013) offers the following example: of-the commander of Al Shabaab Mohammed Garmashago 'According to the newspaper 'De Volkskrant', which cites sources from Holland, Kenya and Somalia, one of the twelve arrested is supposed to have family ties with the commander of Al Shabaab Mohammed Garmashago.' (Hennemann 2013, 351) As is shown in example (5), the journalistic conditional usually indicates that the information in question is uncertain, originating from an undetermined or unknown source.
The use of conditional tense to express politeness is traditionally associated with deontic modality. Both in Spanish and in other languages, temporal distance, i.e. using a nonpresent tense, has been confirmed as a strategy to convey social distance, showing deference (cf . Fleischman 1989;Palmer 2001). The politeness conditional in Spanish has been studied especially by Vatrican (2011Vatrican ( , 2013 but is also mentioned in RAE (2009); Azzopardi (2013) and Hennemann (2013). The hypothetical nature of the 5 In English, hearsay conditional or journalistic conditional (Hennemann 2013). conditional, according to Vatrican (2013), creates an attenuating effect at the pragmatic level. Therefore, utterances considered as possible threats to the face of the addressee, such as a request, are often formulated with the conditional tense, as in the following example, reproduced from Azzopardi (2013): give-inf-obj a little of bread 'Could you give me some bread?' As has been stated here, when looking for a connection between the conditional tense and modality, researchers tend to relate the evidential functions of the conditional tense to epistemic modality, especially that of reported information, while conditional morphology in a deontic environment is seen as a sign of politeness. Therefore, based on earlier studies, one could easily classify cases of deontic debería as politeness conditional instead of evidential conditional, without any further reflection. Nonetheless, as will be shown in the analytical chapter, in the context of the European Parliament the conditional marking of deber is not regarded as politeness, but is rather seen as a sign of evidentiality, working as an intersubjective strategy. Moreover, the evidential functions found here are not related to the journalistic conditional and do not express uncertain information. Instead of referring to second-hand or third-hand knowledge, which could, in our view, weaken the argumentation, the conditional tense studied in our material functions as a justification for a deontic evaluation, i.e. a recommendation or a request for action.

The concept of intersubjectivity
The concept of intersubjectivity could hardly be defined without the related concept of subjectivity. The latter rose as a field of interest with works such as Lyons (1982), Traugott (1989) and Stein & Wright (1995), looking at subjectivity both in modern language communication and in a historical context, for instance its relation to grammaticalization. While the linguists generally agree on the definition of subjectivity as "the expression of self and the representation of a speaker's (or, more generally, a locutionary agent's) perspective or point of view in discourse" (Finegan 1995, 1), the notion of intersubjectivity is somewhat more debatable, being defined differently depending on the author. Two of the most influential definitions are put forward by Traugott (2010) and Nuyts (2012), respectively. From Traugott's perspective, intersubjectivity concerns the expressions that "mark attention to the addressee's self-image". She also coins the term intersubjectification which refers to the diachronic process in which expressions gain intersubjective meanings, normally from expressions that are already considered subjective. Thus, according to Traugott (2010), intersubjectivity and subjectivity are closely related, with the former notion being seen as a further development of the latter. In contrast, Nuyts (2012) regards subjectivity and intersubjectivity as two related but distinct notions, defining subjectivity as an (epistemic or deontic) attitude held by the speaker and intersubjectivity as an (epistemic or deontic) attitude shared between the speaker and a larger group of people, thus applying these concepts to the field of modality. The analysis offered in this paper focuses primarily on Nuyts' (2012) definition. However, it also takes into account his earlier work, Nuyts (2001), where he relates intersubjectivity to evidentiality in his description of epistemic modality, stating that an intersubjective meaning expresses information that is generally known or available to a larger group of people "who share the same conclusion based on it" (Nuyts 2001, 393). From my point of view, there is no evident contradiction between a piece of information or an attitude being shared by a wider group; instead it seems logical that when the addressee shares the speaker's knowledge, there is also a (greater) chance that s/he shares his/her point of view of the best measures to be taken, which is usually the case when speaking of deontic necessity.

Data and method
The data used in the present study consist of political debates from the European Parliament, conducted by 28 Spanish MEPs between 2010 and 2011. The written protocols (the official version) of the oral debates, in total 613 interventions, from the plenary session were extracted from the webpage of the European Parliament (http:// www.europarl.es/), creating a corpus of approximately 180 000 words. In total, 587 occurrences of deber and 287 of tener que were found and classified as deontic modality. Cases categorized as epistemic, dynamic, as well as ambiguous cases were excluded from this analysis. The 860 cases of deontic modality were later classified according to tense 6 and run through the statistic program SPSS, carrying out a chi square test in order to verify the statistical significance of the data. Five tense categories with a very low frequency (a total of only 14 occurrences) were grouped together in a category labelled other, thereby avoiding cells with very few cases. After the statistical analysis, all the cases of the conditional tense were contextualized and studied in detail, to examine their possible function in the political debates. Two tendencies were found, labelled shared norm (based on shared information) and inferential process. 6 Tense is here understood in a broad sense, including both finite classes, such as present, past and conditional, and non-finite classes, like the gerund.

Results and analysis
In this section both the quantitative and the qualitative results will be taken into account, with a special focus on the qualitative part. First, a table from the statistical analysis is presented, followed by a brief commentary. Second, the conditional occurrences will be accounted for and the two tendencies exemplified.

Quantitative results
The statistical analysis, presented in Table 1 below, shows a clear preference for the present tense, both for deber and tener que, reaching 73.3% and 89.2%, respectively. However, there is a difference of 16 percentage units between the verbs, which is mainly explained by differences in the conditional category. Almost one out of five cases of deber, 19.4%, occurs in the conditional, while for tener que it is a highly non-prototypical category, including only 6 cases. The differences in tenses between the verbs were statistically significant (χ 2 =51,059, df=4, p=<.001). Post-hoc tests (Bonferroni corrected) confirmed that there was a significant difference of distribution in the conditional and the present tense categories. The other tense categories only reach a few percent of the total number of occurrences and, more importantly, there are no significant differences in the use of these tenses. As far as the differences in the present and the conditional tenses are concerned, the question arises why deber is more frequent in the conditional tense and less frequent in the present tense than tener que. Following Narrog (2005), the tense differences could be explained by the dimension of speaker-orientation, a notion related to subjectivity.
As Narrog puts it, the use of a non-present tense is regarded as less speaker-oriented than the present tense, since the speaker expresses a distanced position in accordance to the deictic center, the here and now of the actual speech situation. Therefore, due to its higher frequency in the present tense, tener que could be classified as a more speakeroriented verb than deber. Even though this seems reasonable, an interpretation based on the quantitative results was not considered enough for this study and was followed up by a more in-depth analysis. Since present tense is usually the most frequent tense category, which has been observed in other works about modal verbs (Bauman 2013;Blas Arroyo et al. 2013), and, as shown in Table 1, also the dominant tense for both of the verbs studied here, a decision was made to focus on the conditional category for further analysis, where the largest difference was to be found between deber and tener que.

Qualitative results: an in-depth study of the conditional tense and its functions
As is shown above in Table 1, there are 117 occurrences of the conditional tense in the corpus, of which 111 are uses of deber. In section 2.3, when summarizing previous descriptions of two modal functions of the conditional, it was stated that the conditional tense generally tends to be seen as a politeness strategy in deontic contexts. However, in the debates of the European Parliament, the conditional tense clearly assumes other functions, namely two functions related to evidentiality, which can be regarded as evidence towards an intersubjective understanding of deber (as opposed to the more subjective tener que). As will be seen in examples from the corpus, the conditional is used to reinforce the sense of 1) a general norm, shared between the speaker and a larger group, usually the interlocutors, and/or 2) an inferential process the conclusion of which points towards the most reasonable way of acting.
The two evidential functions found here, are related to the first and the third parameter of Bermúdez' (2006; 2016) model of evidentiality, described in 2.2. The inferential conditional, by which the speaker presents a conclusion based on contextual facts, is linked to the manner of access of the information and belongs to the right pole of that dimension, associated with cognitive processes. Similarly, the idea of a shared norm could be related to the (exclusivity vs. universality of the) access to the information source, meaning that this evidential function of the conditional tense would be placed near the right pole, as it refers to universal or, at least, shared knowledge. Perhaps the link between shared information and a shared norm is not crystal clear. Nonetheless, as was proposed in section 2.4, it is logical to relate a shared norm to shared information, regarding the former as a possible consequence of the latter: if a group of people are equally informed about a situation, they may also agree upon the appropriate way of acting. To be able to share a norm or a principle, the people sharing this common ground have to be aware of the existence of the norm. Thus, shared knowledge can be regarded as a prerequisite for a shared norm. An example where the conditional tense functions as a reference to a norm, or a general principle, is presented here: In (6) the speaker criticizes the behavior of a group of other parliament members, which he regards as disloyal, meaning that they are acting against the norms of the European Parliament. In accordance with Thegel (2017), deber refers to a norm in all its tenses, be it the present tense, past tense or the conditional tense. However, according to this analysis, the conditional strengthens the sense of norm, indicating that it has been violated; there is a previous agreement that someone has broken. Thus, the evidential function here is a reference to the common ground between speaker and hearer, a norm that both sides are well aware of. In this manner, the speaker reminds the audience of the guidelines of the European Parliament, "these, as you know, are the rules that we are supposed to follow".
For a comparison, to see what the conditional value adds to the meaning, we will look at an example of deber in the present tense: (7) Context: Además, para afrontar una verdadera reforma fiscal, que incluya recursos europeos para hacer política social y que tenga claramente una dimensión verde, ecológica, es preciso dejar muy claro 'Furthermore, in order to tackle a real fiscal reform, which includes European resources in order to form social policy and which clearly has a green, ecological dimension, we need to make it very clear' que quien contamina debe pagar, y que quien actúe de that who pollute-3sg must-3sg pay-inf and that who act-3sg-sbjv of forma fraudulenta e irresponsable, también. manner fraudulent and irresponsible also 'that those who pollute must pay, and that this also applies to those who behave in a fraudulent and irresponsible manner.' In (7), the speaker refers to the environmental dimension of European policy, stating that the parties guilty of pollution must also pay the cost; that is a principle which must be followed. Even though deber certainly has a future reference here, pointing towards actions that are necessary after the moment of speech, it could be implied that the norm described also constitutes a general rule which has been followed (at least to a certain extent) before. One could say that the present tense use of deber in (7) has a more descriptive character, of the type "We need to make it very clear how our regulation works". In contrast, the use of the conditional deberían in (6) reminds the audience of a norm that has not been followed by the MEP; due to their disloyal behavior, they have acted against the principles of the European Union. This function of the conditional tense, i.e. a reference to a shared norm and the violation of it, is perhaps even more clearly manifested in the following example: (8) Context: Pero esta no es la directiva que habíamos prometido en Tampere y que habíamos prometido durante diez años en el Libro Verde; esta no es la directiva de la igualdad de derechos: <…> siguen quedando en manos de los Estados miembros decisiones que son decisiones 'But this is not the directive that we had promised in Tampere and that we promised for ten years in the Green Paper; this is not the directive on equal rights: <…> there are still decisions left in the hands of the Member States' que no deberían ser discrecionales, sino que deberían ser that not must-3pl-cond be discretional but that must-3pl-cond be-inf regladas. Por tanto, es un paso adelante pero es insuficiente; regulated. For so-much be-3sg a step forward but be-3sg insufficient está muy lejos de la ambición que deberíamos tener be-3sg very far from the ambition that must-1pl-cond have-inf como europeos as Europeans 'that should not be discretional, but should be regulated. Thus, this is a step forward but it is insufficient; it is very far from the ambition we should have as Europeans.' 8 In (8), we have three instances of debería, where the speaker refers to a directive that has not been obtained and to an agreement that has not (yet) been fulfilled. The idea of the violated norm is expressed explicitly through the reference to a promise that was made in Tampere, but that, apparently, has not been kept. According to the speaker, the European Union is far from its ideal, the shared idea about how it should act as a defender of equal rights, and by using the conditional tense, s/he is reminding the interlocutors of this European ambition that they all agree upon.
The conditional tense sometimes appears for grammatical reasons, being motivated by a subordinated conditional if-clause. We will look at one of the few cases from our corpus, an example where tener que is used: (9) Context: No entiendo por qué determinados sectores o intereses deberían 9 salir perjudicados en esta negociación. <…> 'I do not understand why certain sectors or interest groups should come out of this negotiation at a disadvantage. <…>' En todo caso, si esa situación se produjera, es evidente in every case if that situation refl happen-3sg-sbjv-ipfv be-3sg evident que la Comisión tendría que arbitrar medidas that the Commission have-3sg-cond to come-up-with measures compensatorias para los sectores afectados. compensatory for the sectors affected 'In any case, if this situation were to come about, it is clear that the Commission would need to come up with compensatory measures for the sectors affected.' Some researchers argue that the conditional tense can be used when there is an underlying subordinate hypothetical condition, which, despite being implicit, creates the suitable context for the conditional use (cf. RAE 2009; Vatrican 2014). Kronning (2001, 252) 8 This intervention has not yet been translated into English on the European Parliament webpage, and was therefore translated by me. develops a similar argumentation for a modal function 10 of the French non-deontic devoir in the conditional tense, stating that the subjacent clause could be paraphrased like "if everything goes according to plan/as normal [then X would happen]". In my view, it is possible to apply this idea to a deontic environment, where the underlying clause referring to the norm or agreement serves as the background to the conditional clause. The underlying clause would here, in contrast to Kronning's proposal about nondeontic modality, not express an idea like "if everything goes as normal" but rather "if the responsible agent acted in accordance with the norm/our agreement" [then s/he should X]".
As stated above, the second tendency within the conditional use is to express an inferential process leading to a conclusion, which is presented to the audience. The inferential conditional is often associated with epistemic modality, but here, in a deontic context, it is used not to convey probability, but rather what can be termed reasonableness, pointing towards the most appropriate way of acting given the circumstances. The inferential conditional usually appears at the end of an intervention, when the speaker, after having explained a situation, presents his/her conclusion based upon the premises, as exemplified in (10) and (11) Por lo tanto, nosotros creemos que, a partir de esa consideración, therefore we believe that based on that consideration deberíamos apostar por un sistema de seguridad en transición must-1pl-cond bet-inf for a system of security in transition que permita la desmilitarización progresiva de toda la that permit-3sg-sbjv the demilitarisation gradual of all the seguridad en el mundo security in the world 'We therefore believe that, based on this consideration, we should commit to a transitional security system that will enable the gradual demilitarisation of all security in the world.' In both (10) and (11) the conditional tense is preceded by a certain contextualization; the audience is provided with background information, serving as a foundation for the conclusion which follows. In (10), the speaker claims that it seems more appropriate to use the term irregular immigration instead of illegal immigration, thus avoiding a criminalization of these activities. That would be the most logical and reasonable alternative, according to the speaker, taking into account that immigration is not illegal in some Member States. Similarly, in (11), the speaker argues towards a demilitarization of security at a world level, given that the principal threats in our time, hunger and poverty, cannot be defeated through military force. Both in (10) and (11), there are other linguistic means, apart from the conditional tense, which indicates that a conclusion is being presented, such as, which is why, therefore and based on this consideration.
Until now, four examples have been given where only one of the two evidential functionsshared norm or inference -is present. However, there are also cases where both functions are present, being difficult to separate one from another, as in (12) of age. 'They should not be considered as crimes in Iran -of course, they are not crimes in Europe -because Iran is bound by international instruments protecting human rights that establish that these are not crimes of which people can be convicted and that they certainly cannot be given such terrible sentences as the death penalty, especially when minors are involved.' The speaker refers to the penalty system in Iran, stating that this country has sentenced three people, in her opinion innocent, to death, for committing crimes that should not be regarded as such. In this context, it is possible to interpret the conditional use both as inferential and as a reference to a shared norm. A conclusion about the most reasonable way of acting, i.e. not considering adultery, homosexuality and peaceful participation in demonstrations as crimes, is based upon the fact that these are not considered crimes according to international laws of human rights. Besides, these actions are not criminalized in Europe, and if other countries do not criminalize these actions, neither should Iran, as is put forward by the speaker. The indication of a shared norm is found in the idea of the human rights; there are principles to be followed, regarding freedom of speech, being able to be with the person you love, etc., and Iran, in not paying attention to the human rights, is violating the shared norms of the international community.
I argue that these evidential functions point towards an intersubjective understanding of deber. As described in 2.4, intersubjectivity has been defined as a shared attitude by Nuyts (2012), by which the speaker shows that his/her view has support from a larger group of people. The intersubjective sense is especially strong where the speaker uses the conditional tense to refer to a norm, shared by the Parliament members but the inferential conditional can also be seen as an intersubjective strategy, where the speaker presents a conclusion, based upon information accessible to the interlocutors, as the most reasonable alternative, trying to make them feel included in the cognitive process leading to the conclusion.
The fact that deber, in this corpus, is fairly frequent in the conditional tense, while tener que is highly non-prototypical, occurring mainly when strongly grammatically motivated, as in (9), can be seen as an argument towards a description of deber as a more intersubjective verb than tener que. What is more, when having concluded that deber seems to be used to convey intersubjectivity, it is easier to understand previous descriptions of the verb, labelling it as weak obligation (Sirbu-Dumitrescu 1988;Silva-Corvalán 1995;RAE 2009) or internal obligation (Gómez Torrego 1988;Sirbu-Dumitrescu 1988;Fernández de Castro 1999). Taking into account that deber is used to refer to shared norms, necessities that are supposed to be defended by all (or many of) the interlocutors since they imply the most appropriate, reasonable way of acting, the link towards a weaker necessity, beneficial for the modal subject itself, is not entirely far-fetched. Nevertheless, it is considered here that deber is better described as an intersubjective verb than as an expression of weak, internal obligation, since these labels are often ill-defined (if they are defined at all), and also difficult to operationalize.
In contrast, the labelling in this study, preferring the term intersubjective, has been supported by empirical data, both statistical results and an in-depth study of a more limited number of cases.

Conclusions
This paper has argued for an intersubjective understanding of the Spanish modal verb deber, in comparison to the modal verb tener que; two expressions of deontic necessity often considered as near synonyms. A quantitative analysis carried out with the statistical program SPSS shows that there is a significant difference between the verbs regarding their tense distribution, namely their frequency in the present tense and the conditional tense. Most striking is the fact that deber is fairly frequent in the conditional tense, reaching 19.4% of the occurrences, whereas tener que is highly non-prototypical in this tense, with 2.1% of the cases. A more in-depth study of the 117 cases of the conditional tense reveals that it performs two main functions, labelled as evidential, when appearing with deber. Following the evidential model offered by Bermúdez (2006;2016), these two functions of the conditional can be related to the (exclusivity vs. universality of the) access to the information and the manner of access to the information, respectively. Firstly, the conditional form debería is used to reinforce the idea of a shared norm, a general principle which is supposed to be followed by its defenders. The conditional tense not only refers to the existence of the norm but also to the fact that is has been violated. Secondly, the conditional tense can have an inferential function, conveying a cognitive process. In these cases it strengthens the sense of a conclusion presented by the speaker, which is based on contextual information, available both to him/her and the audience. The fact that deber has a high number of conditional cases expressing evidential functions is regarded as an argument towards an intersubjective label of the verb, meaning that it, in contrast to tener que, is oriented towards the addressee, expressing a shared attitude. With these findings in mind, it is easier to understand earlier descriptions of deber, presenting it as a weak, internal obligation. This study also offers a widened perspective of the nature of evidentiality, as it describes the interaction between this notion and deontic modality. Contrarily to what has been stated in previous studies, deontic cases of debería do not necessarily have to be understood as politeness, but can also express evidential functions.