Norwegian Modal Verbs and Attitudinal Modality

Terminology In the following, I will simply use the term ‘attitude’ rather than repeating ‘attitude towards the trustworthiness of propositions and / or the desirability of states of affairs’, since these are the only areas I treat as modal. In my understanding, modality is not concerned with people’s attitude towards other properties, such as good / bad, clever / stupid, easy / difficult, big / small, important / insignificant, cheap / expensive, and so on. Only trustworthy / untrustworthy (of propositions) and desirable / undesirable (of states of affairs) are truly modal. As a general term, I will use OK-ness, covering both trustworthiness and desirability. As technical terms, I will use epistemic attitude when talking about the evaluation of the trustworthiness of a proposition, and non-epistemic attitude when talking about the evaluation of the desirability of a state of affairs.


Terminology
In the following, I will simply use the term 'attitude' rather than repeating 'attitude towards the trustworthiness of propositions and / or the desirability of states of affairs', since these are the only areas I treat as modal. In my understanding, modality is not concerned with people's attitude towards other properties, such as good / bad, clever / stupid, easy / difficult, big / small, important / insignificant, cheap / expensive, and so on. Only trustworthy / untrustworthy (of propositions) and desirable / undesirable (of states of affairs) are truly modal. As a general term, I will use OK-ness, covering both trustworthiness and desirability.
As technical terms, I will use epistemic attitude when talking about the evaluation of the trustworthiness of a proposition, and non-epistemic attitude when talking about the evaluation of the desirability of a state of affairs.
The word epistemic derives from Ancient Greek έπιστήμη meaning 'knowledge, science' . However, the epistemic kind of modality deals with what is believed to be true rather than what is known to be true. This follows from my definition of modality as an attitudinal category, and also harmonizes with most other current definitions of 'epistemic' . In this respect, then, the term is somewhat unfortunate. However, it is so conventional in the linguistic literature that I do not see any point in trying to replace it with a new term.
The speaker may refer to her own or to someone else's attitude, and I will use the term participant which also has become conventional in modern linguistic literature on modality. As several authors have pointed out (e. g. van der Auwera & Plungian 1998: 83;Andersson 2007: 13f.), it is to be preferred to the alternative term, agent (used by, e. g., Bybee et al. 1994), since the person referred to does not necessarily have the agent role in an actual utterance -it may also be the patient, benefactive or have some other role.
I shall treat attitude as a notion with two values, neutral and positive. These can be combined with negation to form negative attitude. Negative attitude is thus seen as a composite category.
Negation may be treated as a separate feature or factor, which may be added to modal expressions in utterances, so that this combination expresses the participant's negative attitude. This is obviously correct with respect to the use of modal verbs, such as must, shall, can or will, all of which can be combined with the negative particle not. Although there exist lexical items, such as scarcely, hardly, prohibited which represent the participant's negative attitude, I choose to analyze the negative attitude as a complex value, resulting from a combination of (non-neutral) attitude and negation. Therefore, I will only operate with two types or degrees of attitude -positive (non-neutral) and neutral, which partly correspond to the traditional terms necessity and possibility, used in most literature on modality. However, there are some important differences between what is called neutral attitude and possibility, and to an even greater extent between positive attitude and necessity, as I will try to demonstrate in the following paragraphs.
Neutral attitude means that the speaker has no objections to accept a proposition as correct or a state of affairs as worth to occur. However, she may equally accept that the same proposition may turn out to be incorrect, or the same state of affairs may turn out not to be worth to occur. In either case, no problems (no conflict) will arise for the speaker with respect to her beliefs or expectations.
Positive attitude means that the speaker is willing to accept a proposition as correct or a state of affairs as worth to occur. If the proposition turns out to be incorrect, or the state of affairs turns out not to be worth to occur, a conflict arises between the speaker's beliefs and / or expectations and the reality, i. e. there is a problem.
However, the speaker may indicate in the utterance that she admits that other attitudes are possible. To put it in other words, the speaker may signal that other participants may have different attitudes than her own, but this does not mean that the speaker is unsure about her own attitude (if this were the case, one would have to do with neutral attitude, cf. above).
As technical terms, I will use simple and complex attitude to distinguish between cases where the speaker in her utterance expresses only her own positive (non-neutral) attitude without approving of any alternative attitudes, and cases where she expresses her own positive attitude at the same time as she signals in the same utterance that other participants may have a different attitude towards the status of the attitude target, i. e., the proposition or the state of affairs in question.
The distinction between simple and complex attitude is only relevant in connection to non-neutral attitude. Neutral attitude is automatically simple since the speaker does not -and cannot -invite anyone to a discussion or to negotiations about the trustworthiness of a proposition or about the desirability of a state of affairs. Such discussion or negotiations are only possible when the speaker has a non-neutral attitude and is willing to listen to alternative attitude(s).
In practice, the complex attitude may be perceived as a lower degree of commitment on the part of the speaker toward the attitude target, as compared with the simple attitude which sounds more categorical and uncompromising. Thus, the complex attitude can easily be interpreted as containing a certain element of doubt or non-assuredness and consequently as representing lesser confidence from the side of the speaker. However, the speaker does not actually need to be unsure about her own attitude in order to be able / willing to allow the other participant(s) to express their (alternative) attitude.1 As to terminology, traditional terms in the literature on modality are necessity and possibility. However, there are reasons to avoid using them in the description of my model of modality. It would not be logical to distinguish between simple (non-negotiable) vs. complex (negotiable) necessity, as the term necessity refers to something absolute and undisputable. The distinction between something the speaker agrees to dispute and something she does not, is an essential part of my model of the semantics of modality. Therefore I stick to the use of the terms neutral and positive attitude in preference to possibility and necessity.
Actually, many authors have operated with terms like strong vs. weak obligation and tentative vs. confident conclusion, which point in the direction that modality is treated as a gradable domain or even as a continuum (van der Auwera & Plungian 1998: 82;van der Auwera et al. 2005: 251-252). But since it is not logical to talk about strong (confident) necessity / possibility as opposed to weak (tentative) necessity / possibility, the terminology gets unnecessarily complicated. To my mind, the cleaner the terminology is, the more adequate the analysis one can achieve. I choose therefore to abandon the traditional distinction between necessity and possibility, in favour of talking about different types of attitude and complexity, which are represented in figure 1 below.
1 As an alternative set of terms for the description of this distinction between the different types of non-neutral attitudes, I have considered negotiable vs. non-negotiable attitude. These terms also represent the idea of the speaker's willingness (or unwillingness) to accept alternative attitudes. There is however some risk that the reader will misinterpret these terms as necessarily invoking some actual negotiations between the speaker and the other participant(s) in a concrete communication situation. The idea here is actually that the speaker may signal her readiness or willingness to accept alternative attitudes in the very same utterance where she expresses her own attitude. Nothing is said about whether any actual negotiations between two or several participants will ever take place. Therefore, I have finally chosen to use the terms simple vs. complex attitude throughout the article; of course, the adjective "simple" has here nothing to do with "simple-minded", "simplistic" or "naïve (attitude)". As already mentioned above, the term attitude is used in connection with the speaker's evaluation of the trustworthiness of propositions (epistemic attitude) and her evaluation of the desirability of a state of affairs to occur (non-epistemic attitude). When there is no need to specify whether epistemic or non-epistemic attitude is meant, I use the term OK-ness which covers both trustworthiness and desirability. This leads to the preliminary representation of modal domains given in figure 2.  In the following sections I will discuss the different types of attitude in the epistemic and non-epistemic domain in greater detail and provide examples that may serve as empirical evidence that my model of modality is not only based on theoretical considerations, but also represents linguistic reality well.
Finally, I would like to mention that I have chosen to use the pronoun 'she' when referring to the speaker, and 'he' when referring to the hearer or other participant(s) in a communication situation.

Non-epistemic modality
Strictly speaking, modality refers to the attitude in both epistemic and non-epistemic modality. In the case of non-epistemic modality, the attitude is pragmatically connected with expectations about the participants' actions and therefore with certain speech acts. This is, in principle, a secondary effect. These speech acts are determined not only by modality itself, but also by the communication situation, which in its turn is primarily determined by the (number of) participants involved. Nonepistemic modality differs from epistemic modality in that it is connected with reactions and / or actions, besides describing the participants' attitude. Epistemic modality is only connected with the participant's attitude towards the OK-ness of a proposition or a state of affairs.
In the traditional literature on non-epistemic (specifically deontic) modality, one usually speaks about permission and different types of so-called mands (commands, demands, encouragements, requests, entreaties), which are called non-epistemic (deontic) possibility and non-epistemic (deontic) necessity, respectively. In my view, permission, encouragement or command are not different (sub)types of nonepistemic modality, but rather different types of speech acts, the use of which depend both on the speaker's (or some other person's) attitude and on the communication situation.
The neutral attitude in connection with non-epistemic modality (non-epistemic attitude) carries the meaning that, in the participant's view, there are no obstacles for the state of affairs to occur -it is OK that the state of affairs occurs, but it is also OK if it does not occur. The reason for why it is OK that a state of affairs occurs does not need to be expressed in an utterance. Thus, the reason is actually not relevant for the identification of the attitude as neutral. This is not to say that the reason is of no relevance for the choice of lexemes in concrete utterances, as we will see in the examples below.
The non-neutral attitude in connection with non-epistemic modality means that, in the participant's view, matters are OK only if the state of affairs occurs. If the state of affairs fails to occur, there is a conflict between the participant's expectations or interests and reality.
An utterance may, of course, contain certain information about the obstacles for a state of affairs to occur. Likewise, one can state that no obstacles are present in a given situation. The speaker may for example indicate that there is no prohibition (either by the speaker herself or by law) or that there are no physical, material obstacles for the state of affairs to occur. An utterance may also contain information about reasons for why it is important to ensure that a state of affairs does not fail to occur. Such information may be explicitly expressed by lexical means, but may also be indicated by the speaker's choice of modal verb. A typical example from Modern Norwegian is the use of the modal verb får in utterances expressing permission.
Du får gå nå (= you may go now, you are permitted to go now) differs from Du kan gå nå (= you can go now) in terms of explicitness regarding the obstacles. The latter utterance simply indicates that there are no obstacles for the participant to leave (without indicating what kind of obstacles could prevent him from being able to do so), while the former indicates that there is no prohibition (= the obstacle) to leave. By using the verb får, the speaker thus grants her own permission or refers to someone else's permission for the participant to leave.
Another typical example is the use of the modal verb skal in utterances expressing command, with 2nd person subject. Du skal gå nå (= you are obliged to go now, you are commanded to go now) differs from Du må gå nå (= you must go now) in that the latter utterance more neutrally indicates that the participant is forced to leave (one does not indicate what kind of circumstances force him to do so), while the former indicates that it is someone's will (= the obstacle) which requires that the participant leaves.
A similar difference may be observed between utterances containing the modal verb bør as compared to utterances containing the modal verbs skal or må. Du bør gå nå (= you ought to go now) means that, in the speaker's view, it is in the other participant's own interests that he goes now. By such an utterance, the speaker sends a signal that other people may have different attitudes towards whether the matters really only are OK provided that the person in question goes now. In practice, such an utterance may be interpreted as expressing a lower degree of confidence on the part of the speaker regarding what her own attitude actually is. Both Du skal gå nå (= you are obliged to go now, you are commanded to go now) and Du må gå nå (= you must go now) disallow any other points of view towards the OK-ness of the matters and consequently can be interpreted as expressing a higher degree of confidence on behalf of the speaker.
As already mentioned, in the traditional literature on modality one usually speaks about different (sub)types of non-epistemic modality, depending on the nature of the obstacles. The most established notions are participant-external vs. participant-internal modality, and dynamic, deontic and boulomaic (boulethic) modality. In my view all of these notions refer to different communication situations, rather than constituting different types of modality. By communication situation, I mean first of all the number of participants involved in a conversation. This determines the nature of the speech (monologue or dialogue), and allows for variation regarding the source of attitude, that is, the person whose attitude is being reported. The speaker does not necessarily report her own attitude.
By separating modality, which deals with the attitude, from information about the kinds of obstacles or reasons behind the attitude, we are able to avoid the major problem with the traditional interpretation of modality, namely, the difficulty of proving that all the different subtypes are actually parts of the same linguistic category. In other words, we avoid the difficulties by formulating a definition of modality which is equally well suited for all the subtypes of modality, and at the same time excludes other categories, such as tense or aspect.
The neutral attitude in connection with non-epistemic modality, as has already been pointed out, may be paraphrased as no obstacle for a state of affairs to occur, or it is OK if a state of affairs occurs (but also OK if it does not).
The positive (non-neutral) attitude in connection with non-epistemic modality may be paraphrased as the matters are OK if -and only ifa state of affairs occurs, but the speaker may signal that it is "allowed" for other people to have different attitudes towards the same state of affairs. Such a "democratic" attitude may be expressed explicitly by lexical means or by the choice of modal auxiliary in a concrete utterance.
Let us now take a look at different types of attitude in connection to non-epistemic modality. If the speaker holds that there are no obstacles for a state of affairs to occur, she may say a sentence like one of the following.
1. Neutral attitude in connection to non-epistemic modality a) Du kan reise til Paris.
'You can go to Paris. ' b) Du får reise til Paris.
'You may (are allowed to) go to Paris. ' c) Du må gjerne reise til Paris.
'It is fine with me if you go to Paris. ' d) Bare reis til Paris, du! 'Just go to Paris!' All these utterances may be paraphrased as It is OK if you go to Paris (but also OK if you don't). The utterance in (1a) is the most unspecified one in the sense that it simply states the absence of obstacles for the state of affairs to occur, saying nothing about the nature of the obstacles. (1b) expresses permission, that is, absence of prohibition, which constitutes information about the obstacle. (1c) and (1d) sound most natural in situations where the speaker reacts to the other participant's attitude, that is, where the other participant has shown his own non-neutral attitude towards the state of affairs (in this case, willingness to go to Paris) and the speaker is now giving her approval.
If the speaker considers that matters are OK only if the state of affairs occurs, she may use an utterance like one of these.
2. Positive (non-neutral) attitude in connection to non-epistemic modality simple attitude a) Du må reise til Paris.
'You must go to Paris. ' b) Du skal reise til Paris.
'You shall go to Paris. ' c) Du vil reise til Paris, altså.
'You ought to / should go to Paris. ' = It is desirable with respect to your own interests that you go to Paris (it is not OK for yourself unless you go to Paris).
All these utterances may be paraphrased as It is OK if, and only if, you go to Paris = It is not OK unless you go to Paris.
The utterance in (2a) is the most unspecified one in the sense that it only shows the speaker's attitude towards the state of affairs, namely, that the participant must leave for Paris in order for matters to be (become / stay) OK. Nothing is said about the reasons for why it has to be so.
The utterance in (2b) means that someone has planned the participant's journey to Paris. This utterance thus contains information about the nature of the obstacle.
The utterance in (2c) means that the reason for why it is necessary to go to Paris, is the participant's own will.
The utterance in (2d) is a command and will typically be used in situations where the speaker has authority to decide what is desirable and what is not.
The utterance in (2e) differs from the utterances in (2a-d) in that it signals that the speaker dissociates herself from being the only licit source of norm: the decision to go to Paris or not rests with the other participant himself. Thus, the speaker expresses her point of view quite unambiguously, but at the same time (and in the same utterance) she indicates that the other participant(s) can have different attitudes. In the traditional literature on modality such utterances are said to express weak obligation, but in my view it is more precise to analyze them as representing the speaker's complex attitude towards the state of affairs. To put it in other words, the speaker does not express that, in her view, it is less necessary for the other participant to go Paris by choosing the auxiliary bør (ought to) instead of må (must) or skal (shall). What the speaker does say, is that she is not expressing the only possible attitude towards the necessity of the journey to Paris.
Thus, the choice of a modal verb in a concrete utterance may reflect considerations which are primarily related to the identity of the source of attitude and the complexity of the attitude (simple or complex). The consequence of the pragmatic interpretation of these considerations is assigning the utterances in (2a-e) different degrees of OK-ness of the state of affairs.
The source of modality need not be expressed in an utterance. If we only have an utterance like Du må reise til Paris, and no additional information, we can say that the attitude is non-neutral and simple, but the source of attitude is unspecified.
Optative is another example where the source of attitude remains unspecified.
3. Unspecified source of modality, optative a) Leve kongen! 'Long live the king!' b) Må kongen leve lenge! 'May the king live long!' The attitude represented by such utterances is non-neutral and simple. In my model, then, non-epistemic modality partly overlaps with what has traditionally been called boulomaic (boulethic) modality, deontic modality and dynamic modality, or, in terms elaborated by van der Auwera and Plungian (1998), participant-external and participantinternal modality. The non-epistemic modality also covers wishes and fears, which are treated as partly deontic and partly epistemic by Palmer (2001: 13).

Epistemic modality
As mentioned, epistemic modality is pure attitude, in my view. Unlike non-epistemic modality, it does not require any reaction and does not involve any actions from the side of the participants of the communication situation. The speaker expresses (her own or someone else's) attitude towards the trustworthiness (likelihood) of a proposition. As in the case with non-epistemic modality, one can speak of the source of attitude here, as well. The speaker is identical with the source of modality when she expresses her own attitude, and not identical with the source of modality when she refers to someone else's attitude.2 The source of modality can also remain unspecified.
The neutral attitude in connection with epistemic modality simply means that the participant has no reason to believe that the proposition is wrong or false; there is no obstacle to accept the proposition as potentially true. It does not say anything about whether the participant expects the proposition to be true or not; it says only that it may well be true (but may also be false).
The most typical means of expressing the neutral attitude in English are utterances containing modal verbs can and may, while in Norwegian it is utterances with the modal verb kan. Such utterances can usually be paraphrased by It is possible that…, It is possibly the case that…, or, to include the notion of obstacle, by There is no obstacle to assuming that the following proposition is true (although it may also be false).

Neutral attitude in connection with epistemic modality
Han kan ha reist til Paris. 'He can have left for Paris. ' = It is possible that he has left for Paris. = There is no obstacle to assuming that he has left for Paris.
The positive (non-neutral) attitude in connection with epistemic modality means that the participant has essential willingness (not reason -although willingness may be invoked by some reason, it may also remain unmotivated) to believe that the proposition is correct; or, to employ the notion of obstacle, the participant will encounter obstacles to accepting any other conclusion than the proposition being correct. The nature of the obstacle does not need to be expressed in the proposition, but if it is expressed, it may be related to knowledge, information, reasonable thinking or simply the participant's belief. 5. Positive (non-neutral) attitude in connection with epistemic modality a) Han må ha reist til Paris.
'He must have left for Paris. ' = It is not OK for me to believe anything else than that he has left for Paris. b) Han vil ha reist til Paris.
'He will have left for Paris. ' = It is not OK for me to believe anything else than that he has left for Paris.
The utterance in (5a, b) represents simple attitude, which is to say that the speaker is not accepting any other attitude besides that the proposition Han er reist til Paris 'He has left for Paris' is correct. The difference between (5a) and (5b) is not related to the attitude as such. The utterance in (5a) may be characterized as a deduction, while (5b) may be characterized as a prediction. In other words, by choosing the modal verb må the speaker signals that she has some kind of evidence to base her conclusion on. By contrast, the modal verb vil is chosen when the speaker predicts something without necessarily having any evidence (or when she refers to something that is known commonly). Thus, the choice of the modal verb provides more information than barely the speaker's attitude sensu stricto.
As in the domain of non-epistemic modality, the speaker may also possess a complex attitude, as is demonstrated by the utterances in (6).

Complex attitude in connection with epistemic modality
De bør ha reist til Paris. 'They ought to have left for Paris. ' Complex attitude means that the speaker considers the propositions in (6) to be correct, but at the same time she signals that different points of view may be accepted as well. In practice such utterances as in (6) may be interpreted as representing a lower degree of confidence (certainty) compared to the utterance in (5a, b).
The use of modal verbs such as kan, bør, vil, må does not in itself contain information about the reasons to believe that the proposition is true (or the nature of obstacles to reject it), other than what has been said about the difference between deduction (expressed by må) and prediction (expressed by vil). The situation is somewhat different when the modal verb skal is used, as discussed below in the subsection on evidentiality.
Dynamic modality and evidentiality -modal or just modality-related domains? In this section, I will briefly discuss the status of dynamic modality and evidentiality in relation to my model of the semantics of modality.

Dynamic modality
Dynamic modality is usually defined as dealing with a participant's ability and, according to some authors, willingness to perform actions or get involved in states of affairs.
Examples of these subtypes of dynamic modality would be utterances like (7) and (8).

Ability
Han kan spille piano. 'He can (is able to, knows how to) play the piano. '

Willingness
Han vil spille piano. 'He will (wants to, is willing to) play the piano. ' Some authors extend the notion to also cover enabling conditions that are external to the participant(s). This type of modality is sometimes called circumstantial modality or circumstantial possibility. It may be exemplified by the utterance in (9). 9. Circumstantial possibility Man kan spille piano der (det fins nemlig et piano der borte). 'One can play piano there (there is in fact a piano available over there). ' As already mentioned, I have chosen to define modality as an attitudinal category, and by attitude I mean people's evaluation of the trustworthiness of propositions or the desirability of states of affairs to occur. So the question is how dynamic and circumstantial modality fits into my model of modality.
The status of willingness seems to cause no problems. Willingness is a kind of positive attitude, and therefore utterances expressing willingness are treated as modal. In my model, willingness is a kind of non-epistemic modality.
The status of ability and circumstantial possibility is less clear. If ability is taken to mean nothing else but a person's mental or physical powers, it falls outside the range of what can be called modal (= attitudinal) meanings. However, the Norwegian verb KUNNE often is used to signal that it is OK if a state of affairs occurs or if a proposition turns out to be true without specifying why it is OK. Or, to put it in other words, the verb KUNNE is often used to express that there is no obstacle for a state of affairs to occur or for a proposition to be true. Since there is no obstacle, the speaker does not need to say anything about the nature of the obstacle(s) which might potentially prevent the state of affairs from occurring or rule out the chances that the proposition could be true. The utterance in (7) Han kan spille piano 'He can play the piano' may be interpreted as meaning 'He agrees (is not unwilling, has nothing against) playing the piano' , as well as 'He is able, knows how to play the piano' -and 'He may be playing the piano' . It is impossible to tell which of the interpretations was intended by the speaker unless additional information is provided. This is not to say that there is no difference between the three interpretations, or that it is impossible to tell them apart in principle. The speaker may make clear which of the interpretations she intends by lexical means, but she also may fail to provide any additional information and in so doing leave it for the hearer to choose which one of the interpretations he prefers. The fact that a person knows how to play a piano does neither require him to be agreeable nor prevent him from being agreeable to play a piano. There is no automatism in the relationship between ability and agreeability (positive attitude). So, it is clear that the two interpretations are distinct in principle. The one of them (ability) is not related to attitude sensu stricto, while the other one (agreeability) clearly is. Ability and agreeability may, but do not need to, coincide. Therefore it seems reasonable to claim that the ability-reading and the agreeability-reading of the verb KUNNE belong to different squares on a figure representing the semantics of the Norwegian modal verbs rather than to the same square. But those squares must be adjacent to each other, since the speaker can fail to indicate the boundary between them. The same goes, mutatis mutandis, for the relationship between epistemic and dynamic modality.
The relationship between circumstantial modality and non-epistemic modality (attitude) is of the same kind. The utterance in (9) above, Man kan spille piano der (det fins nemlig et piano der borte) 'One can play piano there (there is in fact a piano available over there)' , will unambiguously be interpreted by the hearer as an example of circumstantial modality only if the remark about availability of a piano is included. This interpretation involves no attitude. Without additional information the utterance Man kan spille piano der 'One can play piano there' may also be interpreted as expressing someone's permission, i. e. attitude. Availability of a piano and someone's permission may, but do not need to, coincide. The speaker may say explicitly whether she speaks about the availability of necessary resources or about authorities' permission. But she may choose just to say that there is no obstacle for playing a piano, leaving it for the hearer to decide which one or both of the interpretations he chooses.
Thus, it may be argued that the difference between neutral attitude on the one hand, and absence of (physical, material or any type of) obstacles on the other hand, is linguistically irrelevant, at least in the standard bokmål variety of the Norwegian language. By linguistically irrelevant I mean that the speaker does not necessarily need to express her choice vis-à-vis the mentioned difference.
Returning to the question about the position of dynamic (and circumstantial) modality in my model, it seems reasonable to claim that dynamic (and circumstantial) modality is connected to epistemic and non-epistemic modality via underspecification. The attitudinal and non-attitudinal meanings of the verb KUNNE are clearly distinct, and the speaker may express explicitly which of the meanings she intends. But the speaker may choose to fail to draw the boundary between attitudinal and non-attitudinal meanings, leaving it to the hearer to choose between interpretations.

Evidentiality
The modal verb skal is not only used to express the participant's attitude. It is also frequently used in utterances where the speaker refers to someone else's words. That usage represents one of the so-called evidential meanings. Thus, by uttering Han skal ha reist til Paris 'He "shall" have left for Paris' , the speaker indicates that it is someone else that claims that the proposition Han har reist til Paris 'He has left for Paris' is true. The degree of the speaker's own commitment to the proposition is not unambiguously shown by the modal verb in this case. Of course, it may also be shown by some other means of expression. The speaker who refers to someone else's words may also want to express to what degree she herself is committed to the trustworthiness of the proposition, but in such a case she must choose some other means of expression. It seems that it would be "too much work" for an auxiliary verb to indicate both the source of information and the degree of trustworthiness.
In a sentence which contains both the information that the speaker is reporting someone else's words and information about the degree of trustworthiness, one has to employ two means of expression.
Evidential skal is thus not related -not directly, at least, -to the speaker's own attitude towards the trustworthiness of the proposition. However, the preterite form skulle may be used to indicate a lower degree of the speaker's commitment to the trustworthiness of the proposition than the present tense form skal. Consider the two sentences in (11). 11. Evidentiality and degree of trustworthiness a) Han skal ha reist til Paris.
'He is said to have left for Paris' (and I say nothing about the level of trustworthiness) b) Han skulle ha reist til Paris.
'He is said to have left for Paris' (and I see this information as less trustworthy) The difference between skal and skulle can most probably be accounted for from a diachronic point of view, treating the form skulle as subjunctive of the verb SKULLE. In Modern Norwegian, however, there seem to be no grammaticalised means to express both the fact that information is reported and the degree of the speaker's commitment to the trustworthiness of this information.
In the traditional literature on modality, reported information is treated as a type of evidentiality, but there is no consensus as to whether evidentiality is to be included into the domain of epistemic modality or if it should be considered as a separate, though adjacent, domain.
Since modality is defined as an attitudinal category, that is, a category dealing with the people's attitudes towards propositions or states of affairs, the question about the source of information is irrelevant, so to speak, for the decision whether a category is modal or not. A crucial question, however, is whether the category in question describes some participant's attitude towards the validity of a proposition, or not. In such a perspective, evidentiality should only be treated as a modal category to the extent that it involves an evaluation of the OK-ness of a proposition.
These considerations point in the direction of evidentiality being a non-modal domain, in principle. This interpretation is also supported by the fact that a feature "reported" may, but does not need to, be combined with information about the obstacles to accept a proposition as true / false -it follows from there that evidentiality and modality are two different categories.

Conclusion
To sum up the proposal above, we can represent the different types of modality as in figure 3.
A rough overview of the uses of modern Norwegian modal verbs in terms of the proposed model is given in figure 4 (see page 136).
The real picture is further complicated by pragmatic considerations, such as the use of kan / kunne in imperative utterances with a certain amount of politeness or, on the contrary, irony and impatience. Such utterances deserve a more detailed discussion (cf. Mikučionis 2009). Another important aspect of the use of the Norwegian modal verbs is preterite (past tense) forms, used non-temporally. They have not been covered in the current article. OK if a proposition is true (but also OK if it is false)