Contrastive linguistics

Université de la Sorbonne Paris IV 2014/2015 Master 1 Elise BOUCHÉ A contrastive study of English and Swedish modals Sous la direction de M. le Professeur P. COTTE. Contents Introduction 5 1 Modality in English 6 1.1Root/Epistemic interpretation 8 1.2 Aspectual change and subject's type 15 1.3 Dynamic or Deontic? 20 2 Modality in Swedish 22 2.1.syntactic features 23 2.2 Special cases 31 2.3 Aspectual change, introductory subject and negation 35 3 Modality in abstraction 41 3.1Ambiguity and indeterminate utterances 41 3.2 Agentivity 44 3.3 Subject, Verb, Temporality 45 3.4 Contrastive observations 50 Conclusion 57 Acknowledgments I would like to thank M. Cotte who let me do as I pleased and believed in me even though I spent half the year in Australia. Thank you to M. Carruthers who supported me and always answered to my off topic emails. Many thanks to Starbucks which let me use the internet for free. I would also like to thank the State Library of Melbourne which enabled me to study peacefully and quietly. Finally I would like to thank my parents who have been very supportive especially when I was feeling lonely on the other side of the world. List of abbreviations MOD modal PART participle INF infinitive PRES present tense SUB subject WH Wuthering Heights GE Great Expectations DM “(De)coding Modality” Introduction In the first part of my dissertation, I present what modality is in English and discuss the different types of modality; ranking modals on an epistemic and deontic scale. I also study the use of modals and their interpretation, underlining the impact of grammatical and syntactic features on that interpretation (introductory subject, aspectual change, state and event verbs), especially focusing on may and must. Moreover, I discuss the nature of will and research the modal can in its deontic and dynamic uses. In the second part, I briefly review what modality is in Swedish and compare it to modality in English in the light of what has been explained in part one. I also open a parenthesis on some grammatical aspects which should be introduced to the reader in order to fully understand how verbs work in Swedish. In a sub part, I also cast a light on some particular cases such as the modals lär and vore which do not have equivalents in English. I also discuss the nature of the modal vill, comparing it to ska in Swedish and will in English as I already discussed the nature of will in the first part. The last sub part is a contrastive study of the modals måste and kan which respectively correspond to the modals must and may, already reviewed in part one. Finally in part three, I present modals in abstraction, discussing the ambiguous interpretation, agentivity and its impact on modality along with the impact of temporality. The reason why I decided to focus on måste, may, can and kan for a contrastive study is that these modals are on both ends of the epistemic and deontic scales which makes them more interesting to study because they are each other exact equivalent in English and Swedish whereas most of the other modals in English cannot be translated by only one verb in English. The examples I am using to illustrate my argument are either from papers and books listed in “Primary sources” or, when not between quotation marks, of my invention. Linguist Charles Hockett says that human language is defined by a set of features which differentiate us from animals. First of all, we would distinguish nine features for primate communication and four which are only used by humans (productivity, cultural transmission, duality and displacement). “Productivity” is the ability of creating new words, the fact that language is constantly evolving. “Cultural transmission” is the fact that culture and language are deeply entangled. We are born in a society and we are taught to interact withing that society. “Duality” is the making of words using different sounds combined together Finally “displacement” means that humans can talk about something that does not exist (such as abstract concepts) or something that is not physically present. This last feature is what interests me right now as it is what enables us to make assumptions, express a possibility, a necessity, based on will, law, goals, beliefs etc. In order to express a degree of possibility, we need to use a linguistic mean called modality which can manifest itself in various expressions and lexical items: Nouns (“there is a possibility that...”), adjectives (it is possible that...), adverbs (maybe...), conditionals (if...), semi-modals (she needs to/ought to), lexical verbs (believe, think...), hedges (like, you know...) and modals (she can/should/must/might/could/may/shall...). The present dissertation is exclusively about modals and their various use in English and Swedish. 1 Modality in English “Mood is one of a set of distinctive forms that are used to signal modality.” “Modality is a facet of illocutionary force, signaled by grammatical devices (that is, mood), that expresses: the illocutionary point of view or general intent of the speaker, or a speaker's degree of commitment to the expressed proposition's believability, desirability, or reality.” Illocutionary force distinguishes the following types of acts: Asserting/Promising/Excommunicating/Exclaiming in pain/Inquiring/Ordering One of the most interesting thing about modality is that in other languages, it can be expressed in a whole different way and yet conserves some uncanny similarities with the English modality on a syntactic level or on a semantic level. There are five modals in English (can, will, shall, must and may). What separates those verbs from the others, what make them unique in a way and forces us to put a label on them? The first syntactic properties are shared by other verbs: to be, to do, to have, and these properties are called the NICE constructions according to The Cambridge Grammar of the English Language. The first distinction would be the negative form. Just as the verbs TO BE, TO DO and TO HAVE, the modals have a special negative form. In other words it is possible to use the abbreviation/cliticization -n't. "She mustn't come." - "I haven't done that yet." The resemblance with “to be” and “to have” does not end here. Indeed, they share another form which is the interrogative form. There is a subject-verb inversion. "Is he ready?" - "May I come?" The letter C of NICE stands for "code". It means that some sentences can be understood without repeating the whole clause such as in: "She will too." which presumes that a preceding clause contains a verb which has not been repeated here and is not needed to understand the sentence either. Finally the letter E stands for emphasis which simply means that there is a semantic stress on the verb. "He does smoke" "He mày smoke" Of course, even if those characteristics make the modals stand out from the gigantic crowd of verbs, they also make them closer to other verbs. Although one should distinguish modal auxiliaries from modal main verbs. Actually, those were the properties of the auxiliary verbs. They now just stand out from the non auxiliary verbs, that is to say the rest of the verbs which are also called lexical verbs. 1.1Root/Epistemic interpretation Root interpretation or deontic modality (from Ancient Greek “deon” “obligation”) necessitates the involvement of will/permission/obligation to accomplish the action. Epistemic interpretation (from Ancient Greek “episteme” “knowledge”) involves the speaker's opinion/personal involvement/subjective way of explaining/interpreting things. In English, with the same modals, two interpretations can be understood. Ex: “She must be home.”--> root: she has to be home in the sense that she is obligated to be home --> epistemic: she is probably home based on what me (the speaker) knows about her life/agenda. All in all, epistemic modality can be regarded as the view of the speaker towards the truth, the speaker is actually expressing a judgment, he is giving his opinion on whether or not “she” is “home”. One could consider that there is a ranking of the degrees of commitment to the statement's truth. A scale of insensitivity can be imagined: (1) She can be home. (2) She may be home. (3) She must be home. (1) is a possibility, (2) is a probability and (3) is a strong probability or inferred certainty What has to be remembered is that modals can be interpreted either in an epistemic way or in a root/non-epistemic way. It can be difficult to distinguish an interpretation from another but if there was to be a way of ranking them, this is how it can be imagined. 1.1.1 The epistemic scale This scale was suggested by Anna Wärnsby (“(De)coding Modality”, 2006) Confident Inference Reasonable Inference Tentative Inference Possible conclusion must will should may might can could Non-epistemic modality or root modality is clearly distinguished from epistemic modality, as the difference lies in the degree of the speaker's subjectivity regarding a proposition. As I mentioned earlier, there are a lot of non-epistemic modalities. As can be seen on the scale, must can express a “confident inference” (epistemic certainty) but can actually also express a deontic obligation. (1) “The grass is wet, it must have rained.” Epistemic (2) “You must come to the party.” Deontic In sentence (1), must is actually expressing a “certainty” from the speaker which is based on what he can see (“the grass is wet”) and what he knows (the rain can make the grass wet). In sentence (2), must expresses a direct order from the speaker to the addressee. In the Epistemic Scale, only must from sentence (1) is represented. As can be seen on this scale, will is close to must as it expresses a “reasonable inference”. In fact, epistemic will and epistemic must can even have the same meaning. (3) “-Somebody is at the door. - Oh, that will be the postman.” Epistemic The only difference between epistemic will and epistemic must would be that will is based on experience and common sense rather than on a “confident inference”. In sentence (3), the speaker assumes it is the postman because the postman usually comes at that time of the day. In fact one could say that epistemic will is based on an assumption and epistemic must on a deduction. In the scale, should is associated to a “tentative inference”. (4) “She should be at the party by now.” Epistemic Indeed, in sentence (4), the speaker is making an attempt to guess the subject is at the party at the time of the utterance. The main difference between epistemic should and epistemic must is that should denotes some kind of personal wish in the actualization of the predicate. Besides, should refers to something in the future whereas must refers to something in the present (simultaneous to the time of the utterance). May is situated between “tentative inference” and “possible conclusion” in the Epistemic Scale. May is only epistemic if the speaker is trying to make a “tentative conclusion” which makes it weaker than must and should. (5) “She may be at the party by now.” (weak) Epistemic Sentence (5) could be paraphrased this way: “(Based on general knowledge and what I know)there is a possibility she is at the party by now” whereas sentence (4) could be paraphrased like this: “(Based on what I know) she is probably at the party by now.” In the scale, might is considered to me “more” tentative than may but they are usually interchangeable even though might is considered to be the modal representing epistemic modality. The difference between might and epistemic could is that might shows a little bit more commitment to the truth of the proposition. The following examples illustrate: (6) “She might be at the party.” (weak) Epistemic (7) “She could be at the party.” (weaker) Epistemic Epistemic could is therefore weaker than might. Same goes for epistemic can. As for epistemic may it is based on common knowledge and personal knowledge. Yet, it can be ambiguous whether can is dynamic or weak epistemic. (8) “Smoking can be dangerous.” Weak epistemic or dynamic 1.1.2 The Deontic scale Deontic modality is discourse-oriented. Deontic modals are used to give permission, allow, impose obligation. (a) “She can leave.” Meaning that “she” is allowed to leave, “she” is given the permission to leave. Deontic modality can be roughly divided into three subcategories; directives (possibility and necessity), commissives (promises), and imperative. (a.1) “She may leave.” possibility (a.2) “She must leave.” necessity (a.3) “You shall receive a gift.” promise (a.4) “You must leave!” imperative This scale was suggested by Anna Wärnsby (“(De)coding Modality”, 2006) Obligation Recommendation Permission must should can could may might Evidently, must is the modal that expresses the strongest authority or pressure over the addressee. (1) “You must come to the party.” Deontic As we can see, must is at the left end of the scale, both in the deontic and the epistemic figure. In the epistemic scale, it was associated to a “confident inference”, here it is associated to a strong obligation and it can express a directed and a non-directed order. As can be seen on the scale, should tends to express a strong recommendation which is weaker than an obligation expressed by must. (2) “You should come to the party.” Deontic Sentence (1) could be paraphrased like this: “I urge/demand you to go to the party” whereas sentence (2) would be paraphrased like this: “I recommend you to go to the party”. The speaker in sentence (2) is putting less pressure on the addressee than the speaker in sentence (1). Lastly, can, could, may and might are associated to permission. That is to say that the speaker is allowing the addressee to actualize the main predicate in the utterance. (3) “You may come to the party.” Deontic (4) “You can come to the party.” Deontic (5) “You could come to the party.” Deontic These three sentences basically carry the same meaning. The only difference between sentence (3) and sentence (4) is that may is considered to be more “formal” than can. So the only difference would be the degree of formality. Same goes for sentence (5) since could as the past tense form of can is considered to be more “polite”. Even though Wärnsby (2006) put might in the scale, I must say that might is actually only used in epistemic contexts today and its deontic use is quite archaic. 1.1.3 Dynamic modality Dynamic modality is subject-oriented. It deals with the ability of the subject to accomplish what is stated in the sentence. (b) “She can swim.” Meaning that “she” is able to swim, she has the ability to swim. What distinguishes dynamic modality from deontic modality is whether the action is controlled internally or externally. Deontic modality is a permission/obligation which is given by someone else whereas, dynamic modality is the subject's internal ability. Unlike deontic modality, dynamic modality is neither contextual nor performative. Another type of non-epistemic modality that can be mentioned would be existential modality: which has to do with the possibility of something/ a phenomenon existing in the world. (c) Lions can be dangerous. Meaning that there is a possibility for lions to be dangerous even though they are not dangerous all the time (which in this case I would also call sporadic modality, since the phenomenon can occur sometimes on a non regular basis). Certain auxiliaries can only bear one interpretation such as MIGHT which can only have an epistemic interpretation, whereas others can have multiple interpretations (deontic/epistemic/bouletic/ teleological/circumstancial...) The following example illustrates: ex: “I must write my essay” (in view of my goal to validate the year) → teleological (Ancient Greek telos “goal”) What really make the modals unique are these criteria: they are defective, which means that they lack inflectional forms: they have no gerund-participle, no past participle and a single present tense form (that is to say no -s form of the third person singular). Even so, one has to consider that not all the modals actually agree to these rules. Ought to differs for being the only one requiring to (although I consider it to be a semi-modal such as need to), may for example has no -n't form in the present, and ought to and must don't have any past-tense forms. However, they all have a great deal in common syntactically and one could wonder how it is actually possible. It is plausible to suggest that they have been initially taken apart by the native speakers who were aware that these verbs were different from the others. This hypothesis could only mean one thing: first of all, the modals shared something semantically which led them to be syntactically similar. Before digging deeper into this semantic part, I would like to quickly review the ancestors of the English modals. Those verbs are now known as the preterite-presents; they have the form of a Proto-Indo-European perfect but have a present tense value. In Proto-Indo-European, the perfect tense was used to express the result of an action which occurred in the past. Yet, this result is non past and is currently happening. For example in Proto-Indo-European I know equals I have seen (“woida” in P.I.E, “oîda” in Ancient Greek). From a diachronic point of view, modal verbs in Modern English and Modern Swedish are very similar since they come from the same language group even if today their properties differ considerably. → Old Norse → Norwegian North Germanic/Scandinavian → Swedish/Danish Proto Indo-European → Germanic → East Germanic → Gothic West Germanic → Germanic → Anglo-Frisian → Old English → English English and Swedish belong to the same group (Proto Indo-European), although their family is different (Scandinavian/West Germanic). Even though those two languages are linked, and their modal verbs seem similar, are they used in the same way/as frequently? Frequency of modal verbs in the article “Modern English Grammar”, JESPERSEN (11 pages) Will May Shall Must Should Can Temporal 14 0 root 7 3 0 3 0 0 epistemic 8 14 2 0 14 14 The first thing that strikes is that will has a temporal interpretation more often than a root or epistemic interpretation. The other thing would be the fact that there are only two occurrences of shall which seems to be disappearing in Modern English. If I compare these facts to the use of modal verbs in Swedish, the outcome is different. Will and shall; in Swedish vill and ska function differently. Vill can only be used in an epistemic way. Ex: “Jag vill gå med dig” → I want/ I wish to go with you The difference between vill and ska is the same difference between the epistemic and the temporal interpretation. vill is closer to wish as ska is used to express simple futurity. Ex: “Vill du gå med mig?” → Do you want to/Would you like to go with me? “Ska du gå med mig?” → Are you going/ Will you go with me? But more on will and ska later. 1.2 Aspectual change and subject's type 1.2.1 Must Frequency of must and its aspects in chapter 2, 3 and 4 of Great Expectations, DICKENS Frequency Frequency of epistemic interpretations Perfect aspect 5 5 Other 11 1 At first, it seems that every time must is used with a perfect aspect it has to be interpreted epistemically. Indeed, out of 5 propositions with a perfect aspect, five have epistemic values whereas out of the 11 other uses of “must” only a single one is to be understood epistemically while the rest has a root interpretation. The following examples demonstrate: (1) It must have been a very long trip. Epistemic (2) You and I must have traveled at the same time. Epistemic Even though sentence (1) and (2) have different predicates (sentence (1) contains the verb “be” and sentence (2) the verb “travel”), it doesn't change the fact that the interpretation is epistemic. It is actually odd if you consider that must itself is supposed to express a non-epistemic modality with its sense of obligation (“you must do this.”). It is performative (deontic) and it has to be immediate to the time of the utterance. Only one explanation seems possible: the perfect aspect would gives must its epistemic interpretation. Actually, the main verb can only be actualized if the subject to whom the utterance is addressed chooses to act accordingly to what has been commanded/allowed/prohibited. (3) You must do this. command (4) You must have done this. command In sentence (4), the time reference for “do” is anterior to the time of modality which means that the action can no longer be controlled by the speaker since it is already in the past. In sentence (3), the action is expected to be actualized after the time of the utterance. The addressee is given an order and it is their choice to obey or not. Although, one can notice that the posterior reference has a huge impact on whether the interpretation will be epistemic or not. (5) You must have done this, since you are doing something else. Epistemic (6) You must have done this by midnight. Deontic The adverbial “by midnight” is the only difference between sentence (5) and (6). The adverbial places the proposition in the future which gives the speaker some power over the actualization of the action, whereas in sentence (5), the proposition is already in the past and the utterance is only a supposition from the speaker, hence the epistemic interpretation. At first sight, the progressive aspect often gives must an epistemic interpretation too. (7) She must be taking a shower now. Epistemic Although it does not mean that aspectual modification always has something to do with epistemic modality. (8) I must be leaving soon if I want to see this movie at 8. Deontic (9) I can see that the lights are off so the movie must be starting soon. Epistemic In sentence (8) and (9), the time reference for the proposition is posterior to the utterance which basically means that it is incorrect to assume that the epistemic/root interpretation is only based on the sentence's aspect. In sentence (9), the speaker is assuming something based on what he is seeing which can be considered a valid evidence but in (8), the statement is conditional (if). I would like to consider another element which has to be considered when it comes to analyzing modals' interpretations: the subject. In the case of must one has to notice that inanimate subjects can be a factor on whether the sentence has to be understood in a deontic way or an epistemic way. It seems that inanimate subjects encourage an epistemic interpretation. It actually makes a lot of sense since deontic utterances are often involving a person's authority on another person. Yet, in Great Expectations some utterances with an inanimate subjects are to be read in a deontic way (10) “This description must be received with a week-day limitation.” (From GH) Deontic Although, as expected utterances with an inanimate subject mostly have an epistemic interpretation. (11) “The talk must have been all on my side, for Mr. Wopsle parted from us at the door of the Jolly Bargemen, and Joe went all the way home with his mouth wide open, to rinse the rum out with as much air as possible.” (from GH) Epistemic (12) “This must be a dream.” (from GH) Epistemic Sentence (11) has an epistemic interpretation which results from the combination of an inanimate subject and a state verb (“be”) since both encourage epistemic understanding. Besides, perfect aspect (“have been”) indicates an anterior time reference which as mentioned before, usually indicates an epistemic interpretation. Same goes for sentence (12) which has an inanimate subject (deictic) and a state verb (“be”). In this utterance the fact that the main verb is a state verb is very important. Indeed if the state verb is replaced by an event verb, then the interpretation changes too. (13) “This must become a dream.” Deontic Sentence (13) is deontic as it can be paraphrased this way: “this has to become a dream”. What can be remembered is that even though inanimate subject is not a feature of all epistemic interpretation, it certainly promotes it. Combined with a state verb, the chances of the utterance to be understood epistemically are even higher yet not systematic. 1.2.2May May seems to have an epistemic interpretation more often than must. Frequency of may and its aspects in Wuthering Heights, BRONTË, Emily (pages) Frequency Frequency of epistemic interpretations Perfect aspect 3 3 Other 104 50 It is clear that out of the 104 occurrences, less than half are to be understood epistemically whereas the occurrences using the perfect aspect, even though they are only a few, are all epistemic examples. The correlation between perfect aspect and epistemic interpretation is absolute. (1) “He may have done a little in those vocations, Mr. Lockwood; but I couldn't give my word for any.” (from WH) Epistemic (2)“[...] as you may have noticed, the sough that runs from the marshes joins a beck which follows the bend of the glen.” (from WH) Epistemic As can be observed, the time for the proposition is actually anterior to the time reference of the utterance, which is why these sentences have an epistemic interpretation. Yet would it change if the time reference was posterior? (3) You may have done this by midnight. In sentence (3) the adverbial “by midnight” indicates that the time reference is in the future. Unlike must, we know that the modal may cannot allow a root interpretation in this case. Obviously the deontic use of may is not as frequent. Must can bear a deontic interpretation since it expresses obligation/necessity, whereas may which is at the other end of the epistemic scale expresses a possibility and thus allow an epistemic interpretation. Probability---------------------------------------------------------------------------------------> certainty may must 1.2.3Will It is clear that would can have an epistemic value when used with a state verb (1) That would be Jane knocking at the door. Epistemic The speaker assumes it is Jane at the door, based on his personal knowledge of the situation. Yet, present form will is usually associated to deontic modality. (2) You will be here at 8. Deontic (order) Although can would have a non-epistemic value? (3) Back then, I would take a walk in the morning. Deontic In sentence (3), “take a walk” has been actualized in the past on a regular basis. Would expresses a habit, something which has happened often before. If it had to be translated into French, a modal would not even be needed and it would be translated with an imperfect. (“Je marchais/J'avais l'habitude de marcher”). Would is more temporal than it is a modal. Can will have an epistemic value? (4) It is possible that I will take a walk in the morning. Epistemic Here, the modal has en epistemic value thanks to the epistemic word “possible” which indicates that the context is hypothetical. The actualization of “walk” is not systematic and cannot be assured. Once again, epistemic interpretation is related to contextual information. Does perfect aspect have an influence on the epistemic nature of will? Of course, will and the perfect aspect is an impossible combination. (1) “I will have done that.” Incorrect (2) “I would have done that.” Correct Will has to be in its past form to function with perfect aspect. Frequency of would and its aspects in Wuthering Heights, BRONTË, Emily (pages 1 to 7) Frequency Frequency of epistemic interpretations Perfect aspect 3 2 Other 3 1 As can be observed in the table, there is no correlation between perfect aspect and epistemic interpretation. Out of 3 occurrences, only 2 are to be interpreted epistemically. (3) “I would have made a few comments, and requested a short history of the place from the surly owner; but his attitude at the door appeared to demand my speedy entrance.” (from WH) Epistemic It is natural to think that would always has an epistemic value since it is the conditional form of will. In sentence (3), the speaker explains that he would have acted that way if [his] (third party) attitude had been different. “Make” was never actualized and will never be actualized since the time reference of the proposition is anterior to the utterance. (4) “If I had been [bitten], I would have set my signet on the biter.” (from WH) Conditional Most of the time, if would is found in an apodosis, it only has a temporal value. Would can only be either epistemic either deontic if the proposition is independent. 1.3 Dynamic or Deontic? According to A Comprehensive Grammar of The English Language, QUIRCK, 1985, can/could is used to express possibility and ability. (1) Back then, I could take a walk in the morning. Deontic Sentence (1) is not a prototypical example of deontic modality since what gave permission is not indicated; it is not performative. Ability and permission are combined. This sentence could be paraphrased with “be able to” → “Back then, I was able to take a walk in the morning.” Can is usually associated to dynamic modality. Consider the following examples: (2) “She can enter the room.” Deontic (3) “She can easily climb the mountain.” Dynamic Can is polysemous and can express deontic modality and dynamic modality. It is important to distinguish deontic modality from dynamic modality even though, as event modalities they have a great deal in common. They both are considered to be categories of root modality and agent-directed modalities. The main difference is that in deontic modality, the agent is controlled by something external while in dynamic modality, the agent is self-controlled meaning that control/power is inherent to the subject. In sentence (2), the ability of entering the room is given by the speaker to the subject. In sentence (3) the ability of climbing the mountain easily is coming from the subject herself as it her own ability. The subject is the initiator of the event whereas in deontic modality, the subject is subjected to the speaker's control. Deontic modality involves contextual features. The time of the order/permission is relative to the time of the utterance. In sentence (2), the actualization of the main verb “enter” is expected simultaneously to the utterance. Although it is interesting to notice that the past form of can is not the deontic past time equivalent of can in its present form. (4) She could enter the room if I she had the key. There is a permission given by the speaker but this permission is denied and will not carried out in the future (“if”). This shows that contextual features are extremely important to understand modality. The first contextual features are related to linguistic features (time/subject/aspect), the second contextual features are situational. Besides, dynamic can and dynamic could are completely different: (5) I can climb the mountain. (6) I could climb the mountain (until I broke my leg). Without the parentheses, sentence (6) would be ambiguous because there is no temporal indication. It could have been paraphrased like this: “I was allowed to climb the mountain.” in which case the reading would have been deontic. Without contextual features, dynamic could is easily mistaken as deontic (weak permission). On the other hand sentence (5) is a strong permission if read in a deontic way. If read in a dynamic way, the speaker is just stating the fact that he has the ability to climb the mountain; this ability belongs to him and comes from him alone. Can is usually associated to dynamic modality. Dynamic modality lacks subjectivity unlike deontic and epistemic modality which both bear a subjective point of view. 2. Modality in Swedish Just as English, Swedish is an SVO language, which means that the word order is the same: subject/(finite)verb/object. Usually verbs, prepositions, word order are used to express a syntactic and semantic relation. For example a modal can be used to express this kind of relation to a head verb. “Jag får äta” → “I get to eat”/ “I am allowed to eat” → Swedish is partly analytic. But, the semantic and syntactic relations can also be expressed morphologically. “Jag äter” → “I am eating” → Swedish is partly synthetic. There are five groups of modal verbs in Swedish, they are called “hjälpverb” (literally help-verb). Temporals (ha, skola,kommer att) Modals (måste, kunna) Passive (bli, vara) Actionnals (börja, bruka) Causals (låta) On top of that, Swedish has a lot of “modal equivalents” such as behöva (need) or lova (promise) (1) Hon behöver äta. She needs to eat (2) Hon älskar att äta. She loves eating/to eat As can be seen in sentence (1), behöva takes the bare infinitive and doesn't need to be introduced by “att” (to). 2.1.syntactic features Auxiliaries in Swedish can only be followed by a subordinate infinitive verb phrase which cannot be substituted by a noun phrase. Its complement cannot be paraphrased by a nominal sentence (non pronominal). (3) Du bör se hans senaste film. You MODAL see (INFINITIVE) his latest movie. Secondly, an auxiliary cannot be an antecedent of the pronominal göra (to do). Just like English modals, they cannot be introduced by “att” (exception: “kommer att”) so they take their verbal complement without “att”. They cannot be in a passive form but are needed to passivize (“BLI” cf 2.1.1). Their complement is not promoted to subjects in passive forms. Let's consider the following examples (found in ANDERSSON Erik. Modality in Swedish. Revue belge de philologie et d'histoire. Tome 81 fasc. 3, 2003) (4) “Att omarbeta förslaget beslöts.” (4.1) “Omarbeta förslaget villes inte” “It was decided to revise the proposal.” “Revising the proposal was unwanted.” Their complement has no nominal paraphrase. (5) “Hon borde resa till Paris.” opposed to “Hon planerar en resa till Paris.” “She should go to Paris.” “She plans a trip to Paris.” Another characteristic of those verbs is that they can occur in “piles” as this example illustrates: (6) Han måste försöka börja arbeta. SUB/ MOD/ infinitive verb 1/ infinitive verb 2/ infinitive verb 3 He/must/try/start/work → “He must try to start working”. This characteristic is shared by modals only, yet a few verbs share it as well (“hota”, “threaten”, “ämna”, “intend”). Lastly, these verbs are defective or have irregular proprieties. From a diachronic point of view, they used to be strong verbs which were grammaticalized, which explains why it is so hard to define exactly which verb is a modal and which verb isn't. In this mémoire, I will especially focus on the modals which are characterized by these morpho-syntactic criterias: måste (must/have to), kunna (can, may), vill (want), ska (will). It is possible to distinguish epistemic from deontic modality, just as in English. 2.1.1 The epistemic scale This scale was suggested by Anna Wärnsby (“(De)coding Modality”, 2006) Confident Inference Reasonable Inference Tentative Inference Possible conclusion måste bör bör börde må kan kunde lär törde måtte (I deleted the column “Hearsay”as I do not see how it can be related to this study) As can be seen on the scale, måste just like must expresses the strongest commitment to the truth. Just like its English equivalent, måste can be epistmic or deontic (directed or non-directed). (1) Han måste vara på festen nu. Epistemic He must be at the party now. The modal måste shows that the speaker is almost certain that the subject “han” is at the party. The speaker bases himself on what he knows of the schedule of the subject, thus, sentence (1) can be paraphrased like this: “It is very likely that he is at the party now.” (2) Han måste vara på festen vid 2 tiden. Deontic He must be/has to be at the party at 2 o'clock. As the examples illustrate, must and måste seem to function the same way when it comes to deontic and epistemic interpretations. The scale shows us that two modals can express “reasonable inference”: lär and bör. Yet, even though many studies recognize lär as a modal, it is still quite vague which is why I will enlarge on that subject later (cf.2.2.1). On the other hand, bör is generally used to express a suitability or a strong recommendation. (3) “Du bör se hans senaste film.” Deontic “You should see his latest film.” (4) “Man bör göra sin plikt.” Deontic “One should do one's duty.” Yet, in its epistemic sense, bör expresses an assumption or a possibility. (5) “Jag bör vara hemma vid sex tiden.” Epistemic “I should/will be home around six.” Although in the English epistemic scale, the modal associated to reasonable inference is will even though (5) can be translated by should. My assumption would be that tentative inference and reasonable inference are hard to distinguish for the speakers which explains why sometimes bör can be translated by will or should. (6) “Han bör vinna valet.” Epistemic “(Based on what I know) He will win the elections.” Yet, it is clear that the differences between must and should are the same between måste and bör (Börde in the “tentative inference” on the scale is the past form of bör). Both, bör and should express a weaker commitment to the truth of the proposition. Under “tentative inference”, we also find torde, (present tense) usually used to express assumption or probability and mostly found in written Swedish. (7) “Leveransen torde bli nästa vecka.” Epistemic “Delivery should/ is supposed to be next week.” The biggest difference between torde and börde would be that the speaker tends to dissociate himself from the proposition by using torde. Indeed, just like should, the Swedish modals bör and börde express some kind of wish or desirability from the speaker. On the scale, “tentative conclusion” is associated to the modals må and måtte. These modals commonly used to express wishes, hopes, and possibilities in fixed expressions. “Vad som än må hända” (“Whatever may happens...”) “Det må jag säga!” (“I must say!”) “Länge må hon leva” (“Long may she live!”) “Måtte du aldrig glömma det!” (“May you never forget it!”) When used in an epistemical way, the speakers is basing himself on circumstantial evidence, but they appear very rarely and thus, will not be researched in this study. Finally, “possible conclusion” is linked to kan and kunde just like in the English scale (can and could). Yet, kunde (past form of kan) seems to be weaker in its commitment to the truth than its present form. The following examples illustrate: (8) Han kan ha stulit. Epistemic It is possible that he stole. (9) Han kunde ha stulit. Epistemic It was possible for him to steal (but he didn't). Actually kan expresses a “possible conclusion” and for that reason it can have the meaning of these two English modals: may and can. On the other hand, kunde which seems to express something unreal is closer to might and could which are associated to “tentative conclusions. 2.1.2 The deontic scale This scale was suggested by Anna Wärnsby (“(De)coding Modality”, 2006) Obligation Recommendation Permission Volition måste ska bör börde få kunna må måtte The first noticeable thing is that the Swedish scale has one more category which is “volition”. As expected, måste just like must is at the extreme left of the scale and is associated to a strong obligation. (1) Du måste gå på festen. Deontic Just like its English equivalent (“You must go to the party”) the speaker has a strong authority over the addressee and is urging him to carry out the proposition. The second modal liked to obligation is ska. Ska will be explained more thoroughly later on (cf. 2.2.3) but what has to be remembered for now is that bör, börde and ska share the same meaning. Yet, ska is a little bit stronger and deprives the addressee of his say in the matter. Consider the following examples: (2) Du ska gå på festen. Deontic (3) Du bör gå på festen. Deontic (4) Du börde gå på festen. Deontic These three sentences could all be translated into English like this: “You should go the party.”Yet, there is a slight shift in meaning in every one of them. The strongest pressure over the addressee is expressed in sentence (2). This should can be paraphrased by “have to”. In sentence (3), bör is also putting some pressure over the addressee but this pressure has to do with moral values. Indeed, bör, used in the present tense with an even verb which has to be actualized in the future usually carries a morality meaning. Therefore, the should in sentence (3) is more of a moralizing recommendation than the should in sentence (2) which is a strong obligation. Finally, börde in sentence (4) is the softest order out of the three. Usually börde is associated with unrealistic statements, meaning that the speaker himself does not believe that the proposition will actually be carried out. When saying “you should go to the party”, the speaker sort of already knows that the addressee will no do so. Interestingly enough, these subtleties cannot be expressed within three different verbs in English which is why it is quite difficult for an English speaker to understand the semantic differences between ska, bör, and börde. Permission in Swedish is associated to modals få and kunna (kan). Få has various uses. Not only does it have to do with permission but it can also express necessity when its past form (fick) is used. (5) “Jag fick ligga kvar på sjukhuset.” Deontic “I had to stay in hospital.” (from DM) Something has to be notice though, the usual non-modal sense of få is “receive”. As a modal, få means receiving a permission. (6) Du får gå på festen. Deontic “You are allowed/ get to go to the party.” Another fact should be mentioned here, få does not exist in the epistemic range. It can never be interpreted epistemically. Now what is the difference between få and deontic kunna? Utterances with få may be more authoritative then those with kan as kan can be understood as a polite request. (7) Du kan gå på festen. Deontic Just like in English, when past-tense modals are used, it is usually more polite and respectful than present-tense forms. I already mentioned börde but same goes for kunde (past form of kan) (8) Kan du ge mig saltet? “Can you pass me the salt?” Deontic (9) Kunde du ge mig saltet? “Could you pass me the salt?” Deontic Just like in English, the only difference between sentence (8) and sentence (9) is the degree of politeness. Lastly, må and måtte are in the “volition” section. As I said in the previous part, these modals are usually used in fixed expressions and saying. They are in a way the equivalent of få and kan in the sense that they also give permission except that they are slightly less authoritative. Yet because they are archaic and almost never appear in verbal speech, I will not enlarge on the subject. 2.1.3 Grammatical aspects 2.1.3a Passive in Swedish – singular use of the modal BLI Passivization is the process of shifting the grammatical assignments. a. Active: Jens serverar Lena → Jens is serving Lena b. Passive: Lena serveras av Jens→ Lena is being served by Jens When the verb is morphologically marked with the suffix “-s” it is passivized. Although there are two ways. The modal “BLI”(become) and the verb “VARA (be) can be used and thus co-occur with the past participle (usually formed with suffix “-d”) a. äpplet äta-s → the apple is being eaten b. äpplet blir ätit → the apple is being eaten (blir=present tense/ ätit=past participle) 2.1.3b Double passive in Swedish In Swedish it is actually possible for some verbs to take an infinitival complement. An infinitive complement is a verb which is used as a the complement of another verb. In English, it has to be introduced by a preposition. “I need to work”. “work” is the infinitival complement of “need”. In Swedish: (1) Jag planerar att äta äpplet. → att äta äpplet is the infinitival complement clause. I plan (present tense) to eat (infinitive) the apple. → I am planning to eat the apple. Although if this sentence is transformed into a passive clause, then two passive verbs will be necessary thus forming the Swedish double passive. (-r indicates present tense/ -s indicates passive) (2) äpplet planera-s att äta-s av mig. → As for the apple, it is planned to be eaten by me. Thus, the infinitival complement was actually transformed and in a way morphologically passivized. Yet, there can be a reduced form of (2). (3) äpplet planera-s äta-s → “as for the apple, it is planned to be eaten.” Of course, some information were left out. Who is going to eat the apple? Who is planning for the apple to be eaten? I have shown that Swedish has two passives: a periphrastic one using “BLI” (or “VARA”) as the auxiliary and the synthetic passive- those two forms have actually different distributions because they are two different types of modals. (4) Han vill bli arresterad. → “he will become arrested” → “he will be arrested” / “he would be arrested” (5) Han måste arresterades. → “he must be arrested” I would say that the auxiliary use of the verbs bli, få and kommer att assigns a lot of different semantic roles. If we analyze sentence (4), it is obvious that “han” (he) has three thematic roles, that is to say one from bli, one from vill and finally one from “arresterad”. Should the interpretation be root or epistemic? In sentence (5) the participle of the main verb “arresterades” assigns a semantic role which is stronger than the semantic role given by the modal “måste”, whereas in sentence (4) “vill” has actually the strongest semantic role. 2.2 Special cases 2.2.1 The case of lär Just like other modals, lär precedes other verbs but cannot be considered to be a main verb. (1) “Han lär komma.” “He is said to come.” In such a sentence, lär is definitely not the main verb. Another fact which could identify it as a modal is that it cannot be followed by the preposition “att”. Yet, this characteristic is not only shared by modals. Indeed, numerous verbs function that way and are not called “modals” or “auxiliaries” (“ämna”, “hota”, “lova”). Actually lär has more differences than similarities with modals. First of all, it can only occur in the present tense. This particularity is called “a defective paradigm of lär” in Swedish grammars and it justifies the fact that its occurrence is restricted. Actually, lär cannot be found with a verb's past form or a non finite verb. Nilsson (1969) gives these examples: (2) “Han tvingade henne att sjunga.” “He forced her to sing.” subject/ past-tense/ object/ infinitive (3) “Han tvingade henne att lär sjunga.” Incorrect The reason why lär cannot occur in sentence (3) is because such a construction calls for an infinitive (to sing/ att sjunga) as the main verb “tvinga” (force (someone) to) has to be followed by an infinitive. Besides, I said earlier that modals can be piled up in a sentence which is also the case for lär except that lär must be at the left end of the pile. Even though some modals share these characteristics, (torde and borde also only occur at the left end of the pile, borde cannot occur in the infinitive form -only in some Finland-Swedish dialects), none share them altogether. If we compare lär to another modal, the differences are clearly noticeable. Let's take the modal vill (“want to”) which has a past form (ville) and does not share the previous characteristics. (4) “Han lär vilja studera koreanska.” “He is said to want to study Korean.” subject/modal/modal/infinitive verb/object As can be seen, vill or vilja is not compelled to be at the left hand of the pile. Now that I have covered most of the syntactic part, I must elaborate on the semantic part regarding lär. Basically, this verb means “is said to”. (5) Han lär studera koreanska. He is said to study Korean. This can only mean that lär is usually used in reported speech. Let's consider an interrogative sentence in which lär occurs: (6) Lär han studera koreanska? Is he said to study Korean? This question is not about him studying Korean or not but it is about him being said to study Korean or not. Yet, the same construction with another modal would be much different. (7) Måste han studera koreanska? Must he study Korean? Actually lär could be paraphrased like this, according to Nilsson (1969): “It has been reported to me and I believe it to be true that...”which is why sentence (8) is incorrect. (8) Jag tror att han lär studera koreanska. Indeed, the veb “tro” means believe which is already present in the semantism of lär. Basically, such a construction means that the speaker believes something he believes to be true. I would say that lär is too different from modal verbs and it shares too little with them. Even if it has some similarities with borde and torde, those two verbs still have different functions just like måste.The following examples demonstrate: (9) Han måste vara hemma nu. He must be home now Epistemic (10) Han måste studera koreanska. He must study Korean Deontic In sentence (9), måste functions as lär but in sentence (10), måste functions as any other modal. What I would conclude is that because lär can only be involved in reported speech due to its meaning, it can only be interpreted in an epistemic way. A deontic reading of lär cannot exist. As I mentioned before, when using lär in an interrogative sentence, the speaker is wondering about the likelihood of the reported proposition to be true rather than on the actualization of that proposition. 2.2.2 Vore- modal or state verb There used to be three different moods in Swedish: indicative, imperative and subjunctive. Past subjunctive is formed from the stem of the past form of strong verbs and it is supposed to express something that is only imagined. For weak verbs the past indicative is used as a subjunctive. In Modern Swedish, the subjunctive (konjunctiv) tends to disappear and be replaced by indicative past forms (just like weak verbs). Yet, one form remains: “vore”. Vore is the past subjunctive of “vara” (to be). It can be used to express modal meaning even though it has become extremely rare in Modern Swedish. Usually it is used to express a difficulty, an impossibility or the fulfilling of something imagined. (examples from Swedish GRAMMMARRR) (1) “Vore jag ung igen, skulle jag göra mycket annorlunda.” “If I was young again, there are many things I would do differently.” It can also express a wish: (2) “Om jag vore rik, skulle vi kunna köpa en bättre bil.” “If I was rich, we could buy a better car.” It is interesting to notice that vore can function as a modal: (3) “Det vore roligt att komma.” (3.1) “Det vore trevligt att se dig igen.” “It would be nice to come.” “It would be nice to see you again.” As can be observed in sentence (3), vore is no longer translated by “was” even though it is its original meaning. In constructions such as (3), in which vore function as a modal, it is translated by would be. I would consider vore as a modal and as an archaic equivalent of kunde vara (“would be”). 2.2.3 WILL/SKA /VILL auxiliary or modal? What is the use of an auxiliary? When an auxiliary is not a modal, it is used to express time. In English, the abbreviation “'s” ; “'d” is linked to its weak semantic value. Whereas modals cannot be cliticized because they are “too” semantic-except from will/would because of its temporal value. “I would do that”--> “I'd do that”. Here the modal “would” expresses a potentiality but it also expresses an anteriority which is even stronger. “I 'll do that” expresses the action's futurity more than its strong hypothetical dimension. We focus more on the future realization of that action than on the strong eventuality of this future. What matters is the temporal dimension, or even, the consequences of the action's actualization. We already think about what will follow this actualization as if, without being completely certain of its accomplishment in the future, we considered will not as a modal but as a pure temporal element. In Swedish it is a little bit more subtle than that. The Swedish will actually expresses an underlying will of the subject whereas ska expresses the almost impossibility of acting otherwise, as if the action's actualization depended on something independent and non related to the subject. I would say that the perfect translation of “'d” in English would be ska since it is more an auxiliary than it is a modal. In Swedish, will has various uses; a polite one: (a) “Jag ville helst åka utomlands.” I would prefer to go abroad. And one that expresses the subject's whishes: (b) “Vill du ha ett glas öl?” Would you like a glass of beer? Here lies the difference between vill and ska since vill is closer to the subject's wish and ska only expresses a simple futurity. (c) “Ska du gå på bio?” Are you going to the cinema? (d) “Vill du gå på bio?” Do you want/ would you like to go to the cinema? Although ska seems to be slowly falling out of use. It can be compared to ought to when used as a conditional. (e) “Om jag hade tid, skulle jag följa med.” If I had time I would come. 2.3 Aspectual change, introductory subject and negation It is fair to say that the most epistemic modals in Swedish are måste and kan as one is factual and expresses necessity and the other is potential and expresses possibility. 2.3.1 Måste Just like must, måste expresses a deep commitment to the truth form the speaker and a deep authority over the subject. For this reason we can consider måste to be the Swedish equivalent of must. Epistemic must is equivalent to epistemic måste. (1) Hon måste vara hemma. Epistemic She must be home Based on their personal knowledge of her agenda it is safe to assume that [she] is home. (2) Hon måste vara hem vid två. Deontic She must be home at 2. (2) can be paraphrased like this: She has to be/ is obligated to be at home by 2 o’clock. The adverbial “vid två” is what gives the modal its deontic value. Negation Måste expresses compulsion and concession. In positive expressions, måste is the equivalent of får and can be translated into English by must, “have to”, “be forced to”. In negative expression måste is the equivalent of “behöver inte” (“do not need/have to”). (1) “Jag måste tyvärr gå nu”= “Jag får tyvärr gå nu.” Compulsion “Unfortunately, I have to go.” (2) “Du måste ju inte röka.” Concession “You don't have to smoke you know.” Although, even though måste and får share the same meaning in an assertive statement, måste inte and fär inte do not. Får inte means must not. (3) “Du får inte röka.” Interdiction “You must not smoke.” / “You are not allowed to smoke.” Actually, the negated form of måste expresses a lack of obligation. (4) “Du måste inte äta.” Sentence (4) expresses a lack of obligation rather than an “obligation not to”. In English, it would be translated like this: “You don't have to eat.” Whereas in English must not is more likely to express an order. (5) You must not eat. If I was to translate “you don't have to eat” in Swedish, I would use the modal få or the modal bör (6) Du får inte äta./ Du bör inte äta. → You must not eat/ You are not allowed to eat An example is given in this Swedish restaurant: (7) “Det måste inte vara starkt men det måste vara gott!” → It doesn't have to be strong but it must be good! (Negated kan also expresses a non-factuality). Even though must and måste seem to have a great deal in common, they are not each other equivalent semantically speaking, especially in negated utterances. In assertive contexts, måste, just as must expresses a reasonable inference. Although its negated form is closer to a lack of the obligation itself rather than an interdiction. The negation “inte” (not) concerns the modal itself rather than the plain verb. In English though, must not is an interdiction in deontic contexts. Måste is always epistemic when used with a passive aspect. It doesn't matter if the verbs that follow are state verbs or event verbs. Just like in English, the time reference is often anterior to the proposition. (8) “Klockan måste ha varit närmare ett på natten när han ringade mig.” Epistemic “It must have been almost one o'clock when she called me.” (9) “Hon måste ha burit den med sig.” “She must have carried it with her.” Epistemic In sentence (8), a state verb (be) is combined with an inanimate subject (clock) but in sentence (9), the main verb is an even verb (carry) and the subjec is animate (she). Yet, both modals have an epistemic interpretation. Which really seems to matter is that the time reference is always anterior to the utterance, due to perfect aspect and that seems to be the key element for an epistemic interpretation. Inanimate/animate subjects, state/event verbs do not compromise the epistemical reading of these sentences. As must, måste carries a deontic sense. Epistemic adverbials and other features permit to differentiate deontic from epistemic (nog=probably) or it becomes ambiguous. What has to be remembered is that måste, even though it was presented as a modal expressing a factual case and necessity, cannot express a total factuality or a complete necessity. It actually expresses a very high probability. Andersson in his article Modality in Swedish (2003) distinguishes at least five options on the reality scale. Det har regnat: “it has been raining.” Det måste har regnat: “It must have been raining.” Det kan ha regnat: “It may have been raining.” Det kan inte ha regnat: “It cannot have been raining.” Det har inte regnat: “It has not been raining.” Basically, måste means that the speaker is hesitating between necessity and possibility but impossibility is not considered. 2.3.2 Kan Kan and its past form kunde can have epistemic and deontic interpretations. On the other hand English can is rarely epistemic and is usually associated to dynamic modality (1) She can swim. Dynamic Meaning [she] has the ability to swim. (2) Lions can be dangerous. weak epistemic or dynamic? The speaker based himself on general knowledge and personal knowledge/beliefs. In Swedish, kan usually has either an epistemic or a deontic interpretation. With the perfect aspect, kan seems to always have an epistemic value. (3) Vi kan ha varit för generösa. Epistemic We may have been too generous. In sentence (3), the utterance has an anterior time reference. Yet, perfect aspect does not guaranty an anterior time reference. The following examples demonstrate: (4) Hon kan ha avslutat sitt arbete imorgon. Epistemic She/ modal/ perfect aspect/ his work by tomorrow She may have finished his work by tomorrow. (5) Hon kan ha avslutat sitt arbete vid det här laget. Epistemic She/ modal/ perfect aspect/ his work by this time She may have finished his work by this time As can be seen in (3), (4) and (5), kan has been translated by may which cannot have a deontic interpretation in that context. I already mentioned the fact that may expresses a probability more often than it expresses an obligation. As can be seen on this epistemic scale, may, just like kan, is situated on the other end of the epistemic scale, opposed to måste. In (5), the time reference is simultaneous with the proposition. The adverbial “vid det här laget” (“by this time”) is contextual proof that the speaker refers to a present situation. Yet, the proposition's time reference is anterior (perfect aspect). Again, it is proved that perfect aspect encourages epistemic interpretation rather than deontic interpretation. As I said before, in English, can is usually associated to dynamic modality. In Swedish it is quite similar. (6) Hon kan (PRES)simma(INF). Dynamic She can swim. (7) Hon kan (PRES)ha simmat (PERF). Dynamic She could have swum. In all of these cases, (6) and (7) the ability of swimming belongs to the subject “hon”. I already said that dynamic modality is about the subject controlling the event whereas deontic modality is about the event being controlled by something external to the subject. It means that if [she] can swim, it does not mean that she will swim now. In (7), the fact that the subject is able to swim does not mean that the subject has actually swum in the past. The fact that the time reference is posterior to the proposition is crucial for an epistemic reading is also noticeable with “kommer att”. (8) Du kommer att äta applen. You kommer att PRES äta INF apples. You will eat apples. (9) Du kan komma att äta applen. You kan MOD PRES komma att INF äta INF apples. You may eat apples. “Kommer att” can function as a finite auxiliary or as an infinite auxiliary combined with a modal. I must mention that “kommer att”, or rather “komma att” (infinitive form) can only be used if the time reference remains vague. I would say that “komma att” can be compared to ska, even though “komma att” seems to be more subjective in the sense that it involves the speaker's will, whereas ska expresses a simple futurity without any involvement from the speaker. Actually, “komma att” is more epistemic than ska which is, most of the time, purely temporal. Yet “komma att” cannot be called an actual auxiliary because of its syntactic behavior. Although as language evolves constantly, it seems that “komma att” is slowly changing into an actual auxiliary. In informal Swedish, it has become more and more common to use it and dropping “att” (the infinitive marker, equivalent to “to” in English). Constructions such as “komma göra” (will do) have been used instead of “komma att göra”. (10) Hon ska åka till Paris imorgon. She ska MOD PRES åka INF to Paris tomorrow She is to go to Paris tomorrow. (11) Hon kommer att åka till Paris imorgon. She kommer att PRES åka INF to Paris tomorrow She will go to Paris tomorrow. As I said in the will/ska/vill paragraph, ska is “more objective”, meaning the event in the main predicate will happen because there is an undeniable authority which comes from the subject or from something/someone exterior. Sentence (10) could also be paraphrased like this: “She has to go to Paris tomorrow.” We don't know if that imperative was forced upon the subject by someone else or by herself, yet, she has no choice but to go to Paris tomorrow. On the other hand, sentence (11) is more subjective, and thus more epistemic. Introductory subject Another interesting fact is that kan is always epistemic when it has an introductory subject. (1) Det kan vara interessant. Epistemic It can be interesting. (2) Det kan väl hända att jag gå med dig. Epistemic I may as well go with you. (3) “Det kan förstås också ha varit som med havet” (from DM) Epistemic “Of course, it could have been as it was with the sea.” In sentences (1) and (2), the time reference is posterior to the proposition whereas is sentence (3), the time reference is anterior. Yet, it does not seem to change the fact that these examples are all to be interpreted epistemically. The introductory subject is in a way invalidating the subject's authority or influence over main predicate's verb. 3 Modality in abstraction 3.1Ambiguity and indeterminate utterances A modal can be ambiguous if the context does not clarify what is to be understood. (4) She may come home tomorrow. Ambiguous One of the interpretations of sentence (4) would be (a)“she is allowed to come home tomorrow”, the other one would be (b)“there is a possibility of her coming home tomorrow”. Interpretation (a) is deontic and interpretation (b) is epistemic and only one of those two interpretations is possible; they cannot coexist. Thus, it can be interpreted either as deontic or as epistemic. In sentence (4), if the addressee understands it in a deontic way, he may say something about the subject “she” coming home tomorrow and object, because he does not want anyone to visit. Although if the addressee understands the sentence as an epistemic statement, he may clean the house just in case “she” visits on the following day. Thus, the addressee will act accordingly to how he understood the statement: as a permission or a probability. If we take the Swedish modal kan, some utterances can easily become ambiguous (5) Det kan jag inte säga. Ambiguous Object/ MODAL/ subject/ main predicate I can't tell you this. In both sentences (5), the subject is animate (“jag” I), the time reference is posterior to the utterance and the main verb is an event verb (“säga”, say) which should indicate a deontic interpretation. In (5) we are not certain that the source of the deontic force is coming from the subject itself. Ambiguity → Interpretation 1 → Implication 1 Utterance → Interpretation 2 → Implication 2 Another type of ambiguity would be “merger”. It is an ambiguity in which interpretations do not have to be either epistemic either non-epistemic, which basically means that different interpretations can overlap each other. The addressee doesn't have to choose between one interpretation or the other because they can coexist. (7) Man kan inte röka. → you/one can not smoke Merger The subject “man” is generic; it can be translated by “one” and it would if the interpretation was deontic. Indeed, one cannot smoke because of health/legal reasons. On the other hand, “man” can include the addressee. Whatever the interpretation is, the result will be the same, the addressee will not be smoking. Even though sentence (7) can receive two different interpretations: “you cannot smoke” because it is forbidden to smoke in this area, or “you cannot smoke” because it is bad for your health, but there is only one possible pragmatic outcome. (8) “Then, it must be a shilling.” (from GE) Merger This sentence is from a conversation between the narrator and the coachman. The narrator just arrived where he wanted to go and asks the coachman how much he owes him. The coachman wonders if the narrator has more than a shilling since the ride was a shilling but he would like to be paid over that price. The narrator only has one shilling; hence the coachman's answer “then, it must be a shilling.” Two interpretations are possible, the first paraphrase would be “it is probably a shilling based on what I know of rides pricing”, the second one would be “it has to be a shilling because you don't have more than that.” Of course, the second interpretation is the right one because the context is known, but without it, the first interpretation would be valid as well. This shows how important contextual elements are. The right interpretation is a non-directed deontic modality. Yet, isolated, this utterance can be considered to be a merger case since it combines both deontic and epistemic interpretations which overlap each other. The outcome will be the same; the narrator will have to pay a shilling. Merger → Interpretation 1 \ Utterance Implication / → Interpretation 2 The third kind of ambiguity is called “gradience” and it involves different interpretations which are close to each other (dynamic/weak epistemic). It is particularly obvious with Swedish modal kan. (9) De kan ringa min mamma med en telefonautomat. Gradience They can call my mother with a payphone. First interpretation would be that it is theoretically possible for them to call if they use a payphone, the second interpretation would be that they will actually be able to call. Thus sentence (9) can have two possible interpretations, either weak epistemic or dynamic. Just like in merger cases, there is only one possible outcome. The difference with merger would be that in sentence (9), the interpretations can coexist because they are related. Indeed, if it is possible for “De” to call with a payphone, they will be able to do so. The dynamic interpretation is actually implying the weak epistemic interpretation. Gradience Interpretation 1 / \ Utterance Implication \ / Interpretation 2 3.2 Agentivity According to the Oxford Dictionary of English, “agentivity” is the quality or fact of being an agent. “Although agentivity is assigned to the verb, agentive verbs are defined by the referential properties of their subject nouns.” said Cruse (1973). He also claimed that agentivity can be defined by four features: volitive, effective, initiative and agentive. “Volitive” (from Latin volo “I want”) is when the power of will is stated or suggested. (1) “What John did was not eat anything for two days.” (from Cruse) The “effective” feature is when something has a power over something else (metaphorically or literally). (2) “The wind opened the window.” (from Cruse) In sentence (2) “the wind” is the agentive. “Initiative” is a term Cruse borrowed from Halliday (“initiator” of an action) and is explained as when something is initiated through a command. (3) “John galloped the horse around the field” (from Cruse) Lastly, “agentive” is a feature that is found in any sentence in which an action is performed by a subject/object using its own means to achieve it. (4) “The machine automatically switches (itself) off at 6 p.m.” (from Cruse) Now what I would like to say is that modal verbs can be agentive as well. Let's consider these three examples: (5) She may lose some shares. Epistemic (6) She may buy some shares. Deontic (7) She may gamble some shares. Ambiguous Although those utterances are almost identical (only the main verbs vary), they all have a different reading. Sentence (5) obviously has an epistemic value since the speaker is making an assumption based on what he knows of the subject's financial situation whereas sentence (6) is viewed as a permission, and thus has a deontic sense. On the other hand, sentence (9) is completely ambiguous and it is impossible to know without contextual features whether it is an epistemic statement or a deontic one. In sentence (5), there is no agentivity involved since it is highly improbable that it is the subject's will to lose her shares whereas in sentence (6), the subject becomes agent in control since it is possible for her to buy some shares (permission was given by the speaker). Sentence (7) becomes agentive only and only if it is interpreted as a deontic statement, in which case, just as in sentence (6), the subject becomes an agent who can decide to gamble or not. Other famous examples of the subtlety between deontic and epistemic would be the following: (8) You must be very careful. Deontic (9) You must be very careless. Epistemic It is clear that sentence (8) has a deontic interpretation since the speaker is obviously giving an advice/command to the addressee. “You” becomes an agent which has to carry out the action. Yet, in sentence (9) is an observation which encourages an epistemic reading as it is common sense that no one would ever advise someone to be careless. 3.3 Subject, Verbs, Temporality 3.3.1 Influence of the subject's proprieties on modals' interpretation I would like to consider the influence of the subject on the modal's interpretation. Here, I am only considering noun phrases, not verbs, even though a verb can have the role of a subject (“smoking can be dangerous”). Although I will focus on nouns. To briefly sum it up, a noun gives an identity to a stable reference. It designates an entity. Unlike verbs which are analytics (grasping reality “state by state”), nouns are more likely to be synthetic as they represent a reality as a whole. As a verb is dynamic (except for state verbs), it represents a movement, a trajectory; a noun is more likely to be static. Nouns categorize. A subject can either be animate or inanimate and belong to either the first, second or third person. As I previously said, an inanimate subject is more likely to be part of an epistemic utterance since it is very hard to imagine an order/a permission being given to an inanimate object which leads me to think that direct deontic utterances rarely include an inanimate subject. (1) “The wind may open the window.” Epistemic It is irrelevant to think that the speaker is actually giving the wind permission to open the window. As I said before, the subject can have a powerful influence on the modal interpretation. A subject can be definite or non definite. The following examples illustrate: (2) Någon kan komma. Epistemic (3) En flicka kan komma. Ambiguous (4) Flickan kan komma. Deontic (5) Den här flicka kan komma. Deontic A subject can be definite/non definite and it will affect modality. In sentence (2) (“Somebody may come”), the subject “någon” is non-specific and indefinite. Therefore, the interpretation is epistemic since it is hard for the speaker to control the actions of “somebody”. Because the agent is non-specific, the speaker cannot give an order/permission to that agent. A possible paraphrase would be “it is possible for someone to come”. On the other hand, sentence (3) is ambiguous because the subject is specific and indefinite (“A girl may come”). The subject refers to “any girl in the world”. Thus, the speaker is not giving a permission to someone in particular which leads me to think that the interpretation is epistemic. Although, it could also be deontic in the sense that the speaker is allowing girls to come but not boys. Yet the subject “en flicka” is not a typical subject towards which a directed deontic force can be directed. In sentences (4) and (5), the subject is definite and specific (“The girl”/ “This girl”) thanks to the definite article attached to the noun (“-n”) and the deictic “den”. In those cases, the interpretation is clearly deontic since the speaker is giving permission to someone in particular. One should notice that whether it is English or Swedish, the conclusions remain the same on that matter. 3.3.2 Influence of state verbs/event verbs on modal interpretations State or stative verbs are bounded to express a state which is (relatively) static. These verbs are usually verbs of perception, cognition, or emotions. Event verbs or dynamic verbs express processes and activities. Dynamic verbs tend to be bounded to a period of time which is definite whereas stative verbs are more likely to be non-bounded to time since there is usually no indication of when a state of being starts and ends. Evidently the verb's nature influences modality if you consider the following examples: (1) She must be tired. Epistemic (2) She must love you. Epistemic (3) She must wake up. Deontic In sentences (1) and (2), the main verbs are state verbs (“be”, “love”). It is obviously impossible for the speaker to order the agent to be tired or to love someone as it is impossible to control someone's emotions or physical condition. Obviously, state verbs cannot be controlled by an agent that is external to the utterance. In sentence (3), the main verb is a dynamic verb (“wake up”) and the speaker is ordering the subject to actualize this action, hence the deontic interpretation. 3.3.3. Influence of the time reference on modal interpretation When considering the influence of time reference on may, must, måste and kan, I differentiated the time reference of the main verb in the utterance and the time reference for the utterance itself. In Swedish, past time reference seems much more frequent with måste than in English in deontic interpretation. As I said earlier, must differs from måste even though they are considered to be each other equivalents. For example, måste can express past necessity while must cannot. (1) “Jag var så rädd att jag måste blunda.” (from DM) Deontic I PAST so afraid that I PRES form INF “I was so afraid I had to close my eyes.” In English, the same construction would be incorrect: “I was so afraid I must close my eyes.” Usually in English, deontic sense is expressed with “have to” or “allowed to” rather than with must in past time references. (2) You must come tomorrow! Deontic (3) If I feel like it, you may be able to come. Ambiguous In sentence (2), the time reference of the modal is present so the referee is the subject of the speaker's order and is expected to obey which is why the interpretation is deontic. On the other hand, sentence (3) is a little bit different. Because the time reference of the modal is actually not indicated, it is hard to decide between epistemic and deontic. The time reference is present if we consider that the speaker is actually judging the possibility of the proposition to become true. Although if the speaker is in fact saying that in some specific situation which has not come yet, he would indeed allow the referee to come, the time reference is future. This is clear evidence that time reference does matter when it comes to decide whether the modality is epistemic or deontic. Most of the time, past reference and simultaneous reference are associated to epistemic interpretation since most of the time, the speaker is expressing a personal judgment over something that has happened/is happening or is recalling something whereas future reference is generally associated to deontic meaning. Indeed, if the time reference is in the future it usually means that the speaker is giving an order/permission which has to be carried out in the future. The following examples illustrate: (4) You must have come to the party. Epistemic (5) You must be kidding. Epistemic (6) You must come to the party. Deontic In sentence (5), the time reference of the modal is simultaneous to the time of the utterance (progressive aspect). In sentence (4) the time reference is anterior to the time of the utterance (past perfect). Both those examples are interpreted epistemically which confirm the idea that past and simultaneous time reference promote epistemic interpretations. In sentence (4), the speaker is judging the likelihood of a past event while in sentence (5), the speaker is judging the likelihood of a present event. Besides, sentence (5) is describing a state which, as I mentioned before, encourages epistemic sense. On the other hand, the time reference of sentence (6) is posterior to the utterance which promotes deontic meaning. Yet, posterior time reference does not guaranty deontic meaning. (7) “They may finish the construction next year.” (from DM) Ambiguous Sentence (7) is ambiguous since it is not specified whether the action of “finishing the construction” is up to the agents or not. Yet because the time reference is posterior to the utterance, a deontic reading is possible. The speaker might be giving permission to the agents. Although epistemic reading is also possible since the speaker might be judging the likelihood of that action to be carried out in next year. Because the future is unknown, it is possible for the speaker to express doubts and hypotheses about what is going to happen. Another common theory is that passive aspect with must promotes deontic interpretation. This would not make sense if we consider that the subject of a passive sentence is “under control”, meaning that it cannot carry out the action as the action is actually having controlling the subject. Let's consider the following examples: (8) The paper must be handed out by tomorrow. Deontic (9) I thought that the building must be reconstructed. Epistemic In sentence (8), an order (directed or non-directed) is given. But in sentence (9), there is no intended agent. What is stressed upon is the fact that the speaker thought something which was not true. There is no agent to whom the order of being reconstructed is given. Besides, “I thought” clearly promotes the epistemic meaning. Not only that but time reference in sentence (9) is simultaneous to the proposition whereas in sentence (8), it posterior. I would say that progressive aspect does not promote more one interpretation or the other and such consideration has to be decided by looking at contextual elements. 2.4 Contrastive observations 2.4.1 Must and Måste As I stated earlier, aspectual modification is really important when it comes to decide whether the interpretation is deontic or epistemic. Yet, even if most utterances aspectually marked are epistemic, it is possible to find some that are not epistemic (cf 1.2.1). Actually a sentence which is marked by a perfect aspect and has a posterior reference can only be deontic, whereas a sentence modified by perfect aspect with an anterior reference tends to be interpreted epistemically. The following examples illustrate: (1) You must have chosen your outfit by tonight, otherwise you won't be able to go to the party. Deontic (2) “Joe an I going to church, therefore, must have been a moving spectacle for compassionate minds” (from GE) Epistemic Same goes for progressive aspect, usually, simultaneous time reference will be interpreted epistemically rather than deontically. (3) You must be choosing an outfit for tonight. Epistemic I would say that aspectual modification is in fact important for the interpretation, but it has to be considered in combination with time reference (posterior or anterior) when it comes to decide between deontic and epistemic. Most sentences with a posterior time reference will be interpreted in a deontic way as the following examples illustrate: (4) “‘You must give Pip to me, one of these days; or lend him, at all events” (GE) Deontic (5) “I must rob the pantry.” (GE) Deontic Sentences (4) and (5) are both deontic and I would attribute this to the fact that posterior reference promotes deontic interpretation on top of the strong force of must as a modal, which is at the extreme left of the deontic scale (“obligation”). On the other hand, a sentence with anterior time reference will be interpreted epistemically. (6) “I had fancied that it must necessarily be night-time” (GE) Epistemic Nevertheless, in English, anterior time reference without aspectual change is quite uncommon, unlike in Swedish. Let's discuss the case of måste as it is must's equivalent. (7) Du måste ha valt din outfit innan sju annars kommer du inte gå på festen. Deontic you MOD INF PART your outfit before seven otherwise PRES go to the party “You must have chosen your outfit before seven otherwise you won't be able to go to the party.” (8)Det måste ha varit lätt för dig. Epistemic it MOD INF PART easy for you It must have been easy for you. These examples show that, like in English, time reference is crucial for the interpretation. In sentence (7) the adverbial “innan” (before) indicates that the time reference is posterior to the utterance whereas in sentence (8) the time reference is anterior and the sentence is thus, interpreted epistemically. Yet, in Swedish occurs something that cannot be in English. Some utterances appear without a main verb and are interpreted in a deontic way. This is actually quite common in Scandinavian languages, especially in Norwegian. Although this used to be common in Early Modern English. (9) “Allas! Un-to the Barbre nacioun/ I moste anon.” (Canterbury Tales CHAUCER) Alas! To the Barbaric nations, I must go right away.” Deontic (10) Nu, måste han hem. Deontic Now MOD he home Now, he must go home. Usually in Swedish, when the main verb is missing, it can be guessed from the context and it is usually a motion verb such as “gå” or “komma”. This a constant, every time a modal is used in a sentence skipping the main verb, the interpretation is deontic. 2.4.2 May and Kan I said earlier that may is either a “possible conclusion” that the speaker is drawing basing himself on evidence and personal knowledge, or a permission given to carry out the proposition. Unlike must and måste, may combined with a change of aspect will automatically result into an epistemic interpretation. (1) “She mayn’t have know’d” (GE) Epistemic (2) “How much of my ungracious condition of mind may have been my own fault” (GE) Epistemic Another feature influencing the interpretation of may would be the notion of control. Indeed, it is very important to define whether the subject is in control of his actions or not to find out what the interpretation is. Usually, if the subject is in control of the uttered situation, may will have a non epistemic sense. (3) “Among those few, there may be one who loves you.” (GE) Epistemic (4) “It is not the least to the purpose what the reasons of this prohibition are; they may be the strongest and gravest reasons, or they may be mere whim.” (GE) Epistemic Both sentences (3) and (4) are to be interpreted epistemically even though the subject in sentence (3) “one” is animate and the subject in sentence (4) is inanimate “the reasons”. What bring them together is that both of these subjects have no control over the situation. Besides, their predicates are state verbs (“be”) which usually promotes epistemic reading since a state can hardly be controlled. Talking about main verbs, if the main verb is an event verb, and the subject is the first person, there are high chances of the utterance to have a deontic meaning. Let's consider the following examples: (5) “May I venture to congratulate you?” (GE) Deontic (6) “If I say yes, may I kiss the cheek again?” (GE) Deontic In sentences (5) and (6), the time reference is either simultaneous or posterior to the proposition. In sentence (5), the speaker is not actually waiting for a permission to congratulate the addressee as the simple fact that he is asking is already congratulating. Both predicates are event verbs and both subjects are the first person which both promote a deontic reading. Now, I might state the obvious, but it must be said that when an event verb is combined with the second person as the subject, the interpretation tends to be deontic. (7) “You may read the Lord's Prayer backwards if you like.” (GE) Deontic (8) “Very well; than you may go.” (GE) Deontic The deontic sense is even promoted by the fact that the time reference for the utterance is posterior; the speaker is expecting the addressee to carry out the proposition. On the other hand, if the subject is generic, the interpretation will be more likely to be epistemic, unlike with must and måste which tend to be deontic when combined with a generic subject. (9) Man måste dö några gånger innan man kan leva. Deontic One must die a few times before one can live. (10) “One must imagine Sisyphus happy.” (Camus translated) Deontic (11) “One may guess the power of the north wind blowing over the edge, by the excessive slant of a few stunted firs at the end of the house.” (WH) Epistemic I would say that sentence (11) cannot be interpreted as a permission but is rather the possible fact that someone could actualize the predicate if that someone was there in this instant, hence the epistemic reading. In Swedish, modal kan either express a “possible conclusion” if epistemic or a “permission” if deontic (cf. 2.1.1 and 2.1.2) just like may in English. First of all, an utterance containing an inanimate subject (lack of control over the subject) and perfect aspect is usually interpreted as epistemic, which is no surprise (as we saw with may, aspectual change automatically promotes epistemic reading). (1) Det kan ha varit ett misstag. Epistemic “It may have been a mistake.” In sentence (1), time reference is anterior to the utterance, yet I will demonstrate that time reference is not that important when it comes to “possible conclusion” or “permission”, unlike with “obligation” or “confident inference”. Let's consider an example with an inanimate subject and a posterior reference. (2) Det kan få konsekvenser om du gör ett misstag. (weak) Epistemic “They may be consequences if you make a mistake.” In sentence (2), the speaker is expressing an hypothesis based on his personal knowledge of the situation. Like in most weak epistemic statements, the time reference is posterior to the time of the utterance and the only way to verify that hypothesis is if the proposition is carried out. Another common feature of weak epistemic utterances would be passive voice in the utterance. (3) “Det kan faktiskt ha setts som ett misstag.” (weak)Epistemic “It may actually be seen as a mistake.” Sentence (3) combines posterior time reference to the time of modality, inanimate subject, passive voice (“setts” the “s” indicates passive cf 2.1.3a) and, important fact, a modal adverbial (“faktiskt” “actually”). This combination tends to demote a deontic reading and indicates an epistemic one (or a weak epistemic one as it is the case in (3)). On the other hand, when the voice is active, one must consider the context to determinate the interpretation. (4) “Det kan väl aldrig bli samma sak.” (from DM) Epistemic Inanimate subject/ MOD/ never become INF same thing “But of course, it can never be the same.” In sentence (4), the subject is an introductory subject which cannot allow a deontic reading because the subject itself cannot carry out the proposition. What can be concluded is that, with kan, if the subject is introductory, the interpretation is epistemic, yet, if it is not introductory but the subject is inanimate the interpretation is weak epistemic. To summarize, the features characterizing an epistemic interpretation involving kan are introductory subject, passive voice, inanimate subject, change of aspect, modal adverbial. Now, if the subject is animate, one must consider the aspect. Usually, an utterance combining past perfect and animate subject will be analyzed as epistemic. (5) Du kan ha bott här. Epistemic You may have lived here. Yet, if there is no aspectual change, one must consider whether the modality is negated or not. Most of the time, when kan is negated with an animate subject the modality is dynamic. (6) Kan man inte älska, kan man i varje fall hata. (from DM) Deontic “If one cannot love, at least, one can hate.” In sentence (6) the ability of “älska” (love) is lacking but it does not mean that the subject cannot “hata” (hate). In negated deontic utterances, the negation can concern the proposition or the modality whereas in negated epistemic utterances, only the proposition itself is negated. In assertive utterances, the interpretation is more likely to be dynamic if the subject has control over the situation. (7) Jag kan simma. Dynamic I can swim. Yet, if the subject has no control over the situation, the interpretation is more likely to be epistemic. Just like may, subject control is very important in utterances with kan when it comes to choose between epistemic and non epistemic. If the subject is a second person, one must consider the time reference (if it is posterior, the interpretation should be deontic, if it is anterior, it should be epistemic). In non assertive utterances, it seems that one again one must consider the time reference, and if it is posterior, the interpretation should be deontic. (8) Kan vi gå på festen? Deontic Can we go to the party? In sentence (8), the subject is asking for permission to go to the party, hence the deontic interpretation. To summarize, with kan, if an assertive utterance has a second person subject that has control of the situation, if the main predicate is an event verb with no change of aspect and if the time reference is posterior or simultaneous to the time of the modality, there is a strong probability for the interpretation to be deontic. If the utterance combines these features but is non assertive, the interpretation is more likely to be dynamic. (9) Kan du komma på festen? Dynamic Can you come to the party? (10) Du kan komma på festen. Deontic You can come to the party. Sentence (9) combines second person subject, event verb, no aspectual change and posterior time reference and is interpreted in a dynamic way whereas sentence (10) which share these same features has a deontic sense. The only difference between the two is that one is assertive (10) and the other one is not (9). Now, the difference between Swedish kan and English may is diachronic. Indeed, Modern English may comes from Old English mæg which meant “have power” (In Old Norse “mágr” and in Swedish “måg” which means “to make”). Mæg used to be employed in contexts of power but it later expended to contexts of permission and then, developed epistemic meaning. With time, may lost its verb proprieties and became more and more grammaticalized, whereas in Swedish, an utterance with the modal kan as the main predicate is absolutely possible. (11) Jag kan koreanska väl. I MOD korean well I speak good Korean. What can be concluded here is that kan is less grammaticalized than may as it can still function as a main verb. Even if may and kan are at the very right end of the epistemic scales, kan cannot be considered to be the exponent of epistemic modality as may is. Conclusion In this dissertation, I tried to present modality in English and Swedish. I have discussed modal verbs in English and Swedish focusing on their possible interpretation and what defined these interpretations. I organized modals in epistemic and deontic scales in order to compare them, choosing to cast a light on may, kan, måste and must as these four are located either on the extreme left or the extreme right of the scale and seemed to be each other equivalents in both languages. I did my best to show the influence of features such as introductory subject, perfect and progressive aspects, state and event verbs on the modality's interpretation. I also considered the impact of time reference, negation, and subject's nature on that interpretation, which turned out to be crucial. I have come to the conclusion that many features help us define whether an utterance has to be interpreted in a deontic way or in an epistmeic way. The first thing I noticed is that aspectual change has a huge impact on that decision. For example perfect aspect and progressive aspect seem to promote an epistemic interpretation for must even though it does not mean that aspectual modification has always something to do with epistemic modality (only for may, a change of aspect necessarily means epistemic interpretation). I also considered the type of subject involved. Inanimate subjects, for instance, encourages an epistemic interpretation since in deontic utterances, it is usually a person's authority over another person that is expressed. Another feature that I have studied is the type of verb involved. If the predicate is a state verb for instance, the utterances has more chances to be interpreted in an epistemic way since it is hard for the speaker to control a state or a feeling. I tried to cast a light on the verbs must and måste since they are usually considered to be each other equivalent but I made it clear that they are completely different in negated utterances. I also tried to compare kan and may, also considered to be each other equivalent since they have a great deal in common. I showed the influence of time reference on the interpretation and the fact that most of the time, a past reference or simultaneous reference is associated to an epistemic reading as it is usually a personal judgment of the speaker over something that is already in the past, when future reference, on the other hand is linked to deontic meaning as it usually express an order which has to be carried out in the future. Lastly, I showed that despite all of those features, it is sometimes hard to decide whether an utterance has an epistemic interpretation or a deontic interpretation and that the only way to know for sure is to consider contextual features. Table of contents Introduction 5 1 Modality in English 6 1.1 Root/Epistemic interpretation 8 1.1.1 The epistemic scale 9 1.1.2 The deontic scale 10 1.1.3 Dynamic modality 12 1.2 Aspectual change and subject's type 15 1.2.1 Must 15 1.2.2 May 17 1.2.3 Will 19 1.3 Dynamic or Deontic? 20 2 Modality in Swedish 22 2.1.syntactic features 23 2.1.1 The epistemic scale 24 2.1.2 The deontic scale 27 2.1.3 Grammatical aspects 29 2.1.3a Passive in Swedish – singular use of the modal BLI 2.1.3b Double passive in Swedish 2.2 Special cases 31 2.2.1 The case of lär 31 2.2.2 Vore- modal or state verb 32 2.2.3 WILL/SKA /VILLauxiliary or modal? 33 2.3 Aspectual change, introductory subject and negation 35 2.3.1 Måste 35 2.3.2 Kan 38 3 Modality in abstraction 41 3.1Ambiguity and indeterminate utterances 41 3.2Agentivity 44 3.3 Subject, Verbs, Temporality 45 3.3.1 Influence of the subject's proprieties on modals' interpretation 45 3.3.2 Influence of state verbs/event verbs on modal interpretation 47 3.3.3 Influence of the time reference on modal interpretation 47 2.4 Contrastive observations 50 2.4.1 Must and Måste 50 2.4.2 May and Kan 52 Conclusion 57 References 61 References Primary sources: JESPERSEN Otto Modern English Grammar, The School Review, Vol. 18, No. 8 (Oct., 1910), pp. 530-540 DICKENS Charles Great Expectations, 1861 (ebook) BRONTË Emily Wuthering heights, 1847 (ebook) Oxford Dictionary of English, Oxford University Press, 2010 HUDDLESTON Rodney and PULLUM Geoffrey K The Cambridge Grammar of the English Language, Cambridge University Press, 2002 GRAMBERRY Julian Essential Swedish Grammar, 1992 (ebook) HOLMES Philip and HINCHLIFFE Ian Swedish: A Comprehensive Grammar, Routledge Comprehensive Grammars, 2d edition, 2003 Secondary Sources: CRYSTAL David A first dictionary of linguistics and phonetics. Boulder, CO: Westview, 1980 SODEBERG Benny The Double Passive in Swedish, a case of creating raising verbs in the Scandinavian languages GISBORNE Nikolas Dynamic Modality SKASE Journal of Theoretical Linguistics, vol. 4, 2007 HACQUARD Valentine Modality, 2009 KAI VON FINTEL Modality and language, 2006 CRUSE D.A Thoughts on Agentivity Journal of Linguistics, Vol. 9, No. 1, 1973 MEAN Young Song Semantic of epistemic modals WÄRNSBY Anna (De)coding Modality, 2006 PALMER, F., Mood and Modality. Cambridge University Press, Cambridge, 1986 ANDERSSON Erik. Modality in Swedish. Revue belge de philologie et d'histoire. Tome 81 fasc. 3, 2003 QUIRCK, A Comprehensive Grammar of The English Language, 1985 NILSSON, Kim A note on the so-called “modal” “lär” in Swedish, Scandinavian Studies, Vol. 41, No. 4 (November 1969), pp. 320-329   62