They seem to be alike, but it appears that they are not

A contrastive analysis of seem and appear and their Norwegian correspondences

Synnøve Bolstad

ENG 4191 - Master’s Thesis in English Language 60 ECTS credits

The Department of Literature, Area Studies and European Languages The Faculty of Humanities

November 2019

Abstract How similar or different are seem and appear? And how are the differences and similarities reflected in the Norwegian correspondences and chosen translation strategies? This thesis explores similarities and differences of seem and appear with respect to grammar, syntax, semantic meaning and pragmatic functions, and demonstrates how interaction between form, meaning and function determines the meaning and function of the two – and it uses the Norwegian translations to do it. The comprehensive study of seem and appear uses translation data from the English-Norwegian Parallel Corpus (ENPC) in two different methods; Semantic Mirrors and contrastive analysis, to show that even though seem and appear are very similar and have many of the same correspondences in Norwegian, they differ with respect to grammar, syntax, semantic meaning and pragmatic function. The study shows that the interaction between form, meaning and function determines the meaning and function of the two verbs, and provides cautious support for the claim that seem and appear are going through a process of semantic bleaching and grammaticalization and are turning into semi-modals like other emerging modal auxiliaries in English, possibly filling a lexical gap in the range of the modal auxiliaries.

Table of contents Abstract ...... iii Table of contents ...... v List of tables ...... vii List of figures ...... vii 1.Introduction...... 1 2. Modality, epistemicity and ...... 4 2.1. The rise of modality at the expense of mood ...... 5 2.2. Modality as a grammatical concept ...... 5 2.3. Modality as a semantic concept ...... 6 2.4. Modality as a pragmatic concept ...... 9 2.5. The concept of evidentiality ...... 9 2.6. The relationship between evidentiality and ...... 10 2.7. Type and source of evidence ...... 12 3. Grammar, syntax, semantic meaning and pragmatic function of seem and appear as presented in the literature ...... 15 3.1. The grammar of seem and appear ...... 15 3.2. Syntactic structure ...... 16 3.3 Syntactic function ...... 19 3.4. Semantic meaning ...... 21 3.5. Pragmatic functions ...... 23 3.6. The relationship between form, syntax, meaning and function ...... 28 3.7. Previous studies of seem and appear ...... 28 3.8. Focus of the present study ...... 29 4. Method and Material ...... 30 4.1. Equivalence and correspondence ...... 30 4.2. Semantic Mirrors ...... 31 4.3. Corpus linguistics, contrastive analysis and translation correspondences ...... 35 4.4. About the chosen corpus ...... 40 4.5. The data set ...... 40 5. Analysis...... 45 5.1. The senses of seem ...... 45 5.2. The senses of appear ...... 49 5.3. Grammatical form and syntactic structures of seem and appear and their correspondences ...... 53 5.4. Evidential and epistemic meaning ...... 60

5.5. Pragmatic functions of seem and appear ...... 65 5.6. The relationship between syntactic form, evidentiality and epistemic modality and function ...... 72 6. Discussion and conclusion ...... 75 6.1. Seem and appear differ with respect to grammar and syntax ...... 75 6.2. Seem and appear have semantic meanings that are unique to them...... 75 6.3. Seem and appear differ with respect to evidentiality and epistemic modality ...... 77 6.4. Seem and appear have different pragmatic functions that reflect their epistemicity and evidentiality ...... 80 6.5. Form, meaning and function ...... 81 6.6. Seem and appear are becoming like the modal auxiliaries ...... 81 6.7. Conclusion ...... 82 7. Summary ...... 84 8. References ...... 86 Online sources ...... 90 Primary data ...... 90

List of tables Table 1. The Norwegian correspondences of seem and appear with a to- clause or a that-clause………………………………………………………………………………………………….42

Table 2. Congruent, non-congruent and zero correspondences of appear and seem, raw numbers and percentages in relation to the total number…………………..…………..53

Table 3. Type and source of evidence in sentences with appear and seem, raw frequencies and percentages………………………………………………………………………………………………..61

Table 4. Distribution of sense-groups of seem and appear, raw frequencies and percentages……………………………………………………………………………………………………….64

Table 5. Pragmatic functions of appear and seem, raw frequencies and percentages…..…66

Table 6. Syntactic structures with seem and appear, source of evidence and pragmatic function, raw frequencies……………………………………………….…………………………………72

List of figures Figure 1. Evidential hierarchy after Willett (1988), based on adaptation by Faller (Brugman, 2015)……………………………………………………………………………………………….13

Figure. 2 First t-image of Norwegian tak………………………………………………………………..………..32

Figure 3. The inverse t-image of Norwegian tak…………………………………………………………….…32

Figure 4. Second t-image of Norwegian tak……………………………………………………………..………33

Figure 5. Restricted second t-image of Norwegian tak…………………………………………………….34

Figure 6. Sense groups of Norwegian tak………………………………………………………………………..34

Figure 7. Sense groups of seem and appear ……………………………………………………………………53

Figure 8. Sense groups and degree of evidentiality and epistemic modality…………………….65

1.Introduction When I say that somebody appears to have done something or seems to be doing something, I do not commit myself totally to the truth of that statement, and I hint that there is some evidence – or lack of it – that form the basis for how committed I am. The expressions seem and appear modify what would otherwise have been a categorical statement and introduce epistemic modality and evidentiality into the proposition. The choice of these verbs over other verbs, and the choice of one over the other, makes a difference in meaning and communicative function, and we should take care both when we use them and when we translate them.

On the face of it, the two expressions and their translations are very similar. They are used as synonyms, and dictionaries often use one expression to define the other. They also share pragmatic functions. However, there are obvious differences between the two expressions when it comes to grammatical properties and syntax, and they also differ in semantic meaning and communicative function.

So how similar or different are seem and appear really? And how are the differences and similarities reflected in the Norwegian correspondences and chosen translation strategies? The present study tries to give a comprehensive understanding of the similarities and differences of seem and appear with respect to grammar, syntax, semantic meaning and pragmatic functions, and how the interaction between form, meaning and function determines the meanings and functions of the two verbs – and it uses the Norwegian translations to do it.

The research questions for my thesis are:

1. How different or similar are seem and appear with respect to grammar, syntax, semantic meaning and pragmatic functions? 2. How does the interaction between form, meaning and function determine the meaning and function of seem and appear? 3. How are these differences reflected in the translation?

To answer these questions, the present study uses translation data from the English Norwegian Parallel Corpus (ENPC) and two different methods, Semantic Mirrors (Dyvik, 1998) and contrastive analysis, to investigate the extent to which seem and appear differ. Translation data are thus used both as a source of information about the semantic meaning and pragmatic function of these two verbs in English, and as a source for detecting differences and similarities in the translations themselves.

For a productive discussion on whether the differences or similarities between seem and appear are related to grammatical properties, syntactic form, modal or evidential meaning and pragmatic effect, it is necessary to distinguish between evidentiality and epistemic modality as grammatical concepts, semantic concepts and pragmatic concepts. Any conclusion on how epistemic modal or evidential seem and appear are, depends entirely on the definitions and the boundaries of the concepts of evidentiality and epistemic modality. If evidentiality and epistemic modality are defined as separate grammatical concepts, only grammatical markers or members a closed grammatical class would qualify as evidentials or modals. As we shall see, seem and appear would not be evidentials or modals in this strictest definition. If, however, evidentiality and epistemic modality are defined as semantic concepts, the present study will show that both seem and appear clearly have evidential and epistemic meaning in certain constructions and could potentially qualify as evidentials or

1 modals. The analysis of seem and appear also depends on pragmatic factors such as subjectivity and intersubjectivity, stance, hedging and politeness. It is difficult to draw strict boundaries between grammar, semantics and pragmatics when it comes to seem and appear, since the different dimensions interact and mutually affect each other, but this study is an attempt at a comprehensive study of seem and appear that include these dimensions and their interdependency.

To answer the question of how similar or different seem and appear are, it is especially important to disentangle the complex relationship between reference to evidence (evidentiality) and speaker's evaluation of truth value of the proposition and his or her commitment to it (epistemic modality). Different models describe the relationship between evidentiality and epistemic modality differently, but the relationship itself is not disputed: Type and source of evidence determines how strong the evidence is perceived to be. Direct evidence is more reliable than indirect evidence, and visual evidence is generally regarded as more trustworthy than hearsay, for example. Evidence strength speaks to how reliable the information is thought to be, which in turn influences how certain the speaker is. The degree of certainty reflected in in the speaker's statement is an epistemic quality and influences how committed the speaker is. The relationship between evidentiality and epistemic modality is particularly relevant for inferences, assumptions and implicatures. Inferences is a way of acquiring information through reasoning, at the same time the reasoning can prompted by and dependent on evidence. This study will show that evidentiality and epistemic modality interact in inferences, and that in most cases evidentiality and epistemic modality become a matter of degrees: whether the evidentiality is stronger than the epistemic modality, or the epistemic modality is stronger than the evidentiality. This study will show that evidentiality and epistemic modality are evenly distributed in inferences, but that in inferences with seem, the balance tips slightly towards epistemic modality: Inferences with seem tend to have a higher degree of epistemic modality, but not by much.

Previous studies (of seem) have shown that there is a relationship between syntactic form on the one hand and evidential and epistemic modal meaning on the other hand. Seem can for instance have the meaning ‘appearance’ and hedge probability or indicate hearsay depending on the construction it occurs in. According to Aijmer (2009), seem to is the most grammaticalized form and can be regarded as an evidential marker, and it seems (that) is similar to an adverbial. Previous studies have labelled seem in combination with a that-clause as a hearsay marker, signalling that knowledge has been acquired through language. The present study shows that both seem and appear have different meanings and functions in different syntactic constructions, and that they are not identical in this regard. While I confirm what others have found before me - that seem in catenative constructions is predominantly evidential and seem plus a that-clause is more epistemic - I find that the pattern is a little different with appear. This study finds that appear is predominantly evidential in both in catenative constructions and constructions with a that-clause.

Through the Semantic Mirrors-method, the present study was able to determine several different senses for seem and appear. The different sense partitions could then be placed on a scale ranging from having a high degree of evidential meaning and little or no epistemic meaning, to sense partitions with a high degree of epistemic modal meaning and little or no evidential meaning.

The relation between form, meaning and function is central to another question underpinning my thesis: whether seem and appear in certain constructions are going through a process of semantic bleaching and grammaticalizing and are becoming like the modal auxiliaries – so much so that they should perhaps be considered part of the group of emerging auxiliaries in English, like need to or used to.

2

Previous studies have shown - and this thesis will confirm - that seem is rarely translated by a modal auxiliary, and it often disappears from translations without a trace. I find that the same is true for appear. Translation strategies like omissions, which entail a loss of modal or evidential meaning in the translations, and restructurings, e.g. with one of the Norwegian modal particles or a modal adjunct, offers insight into the different meanings and functions of the original expressions. The study also provides insight into the prototypical translations of seem and appear and how these are not the same as the most frequently used Norwegian correspondences for seem and appear. The study also shows that some correspondences are frequently used for some pragmatic functions, but not others, for example that translation with a Norwegian is the preferred translation strategy for seem as a politeness marker.

Moreover, translation strategies matter. Omission and restructurings that downplay the epistemic modality or upgrade the evidential strength or epistemic modal meaning could result in loss of meaning in the translated texts and subsequent misunderstandings.

The structure of the thesis is as follows:

Chapter 2 provides the theoretical framework for epistemic modality and evidentiality and introduces the definitions and distinctions that are made. In this chapter I provide an overview of the rise of the concept of modality at the expense of mood and discuss modality as a grammatical, semantic and pragmatic concept. I also provide a typology of modality and discuss the relationship between evidentiality and epistemic modality. Chapter 3 presents the grammar, syntax, semantic meaning and pragmatic functions of seem and appear as it is presented in the literature. This chapter also contains an overview of previous studies of seem and appear and previous studies of some of the most frequent correspondences, e.g. the Norwegian modal particles. Chapter 4 presents the method of Semantic Mirrors (Dyvik, 1998), and describes the design of the contrastive analysis of seem and appear and their correspondences. Chapter 5 presents the results of the analyses with examples from the corpus, and these are discussed further in Chapter 6. This chapter is structured to answer the research questions of my study. Chapter 7 sums up the thesis and suggests further work.

3

2. Modality, epistemicity and evidentiality This chapter attempts to describe and disentangle the complex relationship between modality, epistemic modality and evidentiality. These concepts and the demarcations between them are central to the analysis of seem and appear. Both expressions can refer to evidence. And both can convey something about speaker's assessment of truth value of the proposition and his or her commitment to it and function as epistemic modal markers. However, how evidential or modal they are, is a point on which they are thought to differ. A comprehensive understanding of what is comprised in the concepts of modality, epistemic modality and evidentiality is thus necessary to understand crucial similarities and differences between the two verbs. The concepts are also intertwined: Speaker's commitment depends on how certain or uncertain he is of the factuality of the proposition, which in turn depends on the reliability of the information, which depends on the source and type of evidence. Any conclusion on how modal seem and appear are, thus depends entirely on the definitions and the boundaries of the concepts. The chosen approach to evidentiality and epistemic modality matters, because the question of whether seem and appear should be considered part of a category of modals or evidentials depends entirely on how the category and subcategories of modality are defined and where the lines are drawn.

Another important discussion is whether the differences or similarities between seem and appear are related to grammatical properties, syntactic form, modal or evidential meaning or pragmatic effect. As such, it is necessary to distinguish between evidentiality and epistemic modality as grammatical concepts, semantic concepts and pragmatic concepts. If we see modality as a grammatical concept, linguistic items such as seem and appear must share the grammatical properties of a closed grammatical class to qualify as a member. Alternatively, modality could be a semantic category, in which case the modal meaning qualifies an expression as modal. The function of modals in context or interaction is a third way of categorizing modals.

Halliday (2004) called modality the region of uncertainty that lies between 'yes' and 'no'. A more detailed definition of modality is that it is a feature of language that allows the speaker to communicate some nuance of reality, truth, probability, possibility and inherent obligation pertaining to a situation or event, to communicate the speaker's degree of commitment to an expressed proposition, and the speaker's attitude and judgement towards the propositional content of a proposition. This definition is a synthesis of different definitions of modality that captures different notions of modality. It shows how modality is a very broad linguistic category – "a super category" according to Nuyts (2005). The present chapter provides the theoretical background for parsing this super category into manageable concepts suited to provide answers to the research questions. It starts with a brief outline of how the concept of modality has evolved over the years (2.1). Modality as a grammatical category is presented in 2.2. The definition of modality as a semantic concept is presented in 2.3. This section includes a typology of different types of modality, including epistemic modality. Modality as pragmatic concept is discussed in 2.4. The concept of evidentiality is discussed in 2.5., and the relationship between evidentiality and epistemic modality is discussed in 2.6. Next, different models for categorizing source and type of evidence is presented in 2.7.

4

2.1. The rise of modality at the expense of mood This section comprises a brief outline of how the concept of modality has become increasingly popular and widespread in its use at the expense of the concept of mood (adapted from Nuyts and Auwera, 2019). The concept of modality is closely linked to the concept of mood, and different definitions of modality will relate to the mood system in different ways.

Early Greek scholars defined mood as part of a language system that allows us to express ideas and moods by making a statement, posing a question or expressing an order, through sentence types and speech acts. In other words, the focus was mainly on form. This is the Protagorean notion of mood. Later on, Roman grammarians viewed mood as "different inclinations of the mind as marked on the verb" (Nuyts and Auwera, 2019: 23), thereby introducing a definition that referred to both form and meaning. With this definition, the number of moods grew as new types of moods were introduced, for example the infinitive and impersonal moods. This is the Dionysian notion of mood. In the Middle Ages, grammarians introduced the idea of propositional content (dictum) that could be characterized according to a set of labels (modi): the propositional content that it is raining could, for example be either 'necessary', 'possible', 'impossible', 'contingent', or even 'true' or 'false' (Nuyts and Auwera, p. 17). This is the Boethian notion of mood.

In later studies of English as a vernacular language, the modal auxiliaries were usually described and analysed as part of mood, but at the same time they were often linked to the ideas of necessity and possibility, which is very similar to the modern concept of modality. The concept of irrealis – that something is not real or not factual – was also introduced as a type of mood in this period (Bybee 1994, Palmer 2001).

Even though the modern concept of modality, which has at its core the notion of possibility and necessity and can be traced back to Aristotle and his square of oppositions that explains the relationship between necessity and possibility, it was Kant (1781) and his knowledge theory that caused the increased use of the term modality, which replaced the Boethian mood. Kant's philosophy was concerned with knowledge and truth, e.g. how we acquire knowledge and whether it is possible to have certain knowledge. According to Kant, modality is one of four categories of human judgment: quantity, quality, relation and modality. Modality can be either problematical (Is the proposition possible?), assertorical (Is the proposition real or true?) or it can be apodeictal (Is it necessary?). Kant was the first to use the term modality to refer to the necessity and possibility of a proposition.

From the 1960s and onwards, modality entered English linguistics, and was increasingly used to analyse modal auxiliaries and speech acts (Leech, 1969, Lyons 1977, Halliday 1970), modal logic and types of modality (von Wright 1951, Palmer 1979).

2.2. Modality as a grammatical concept Modality as a grammatical concept focuses on form and linguistic properties that place modal expressions in a grammatical category. Modality is defined as a category of grammatical forms which express modal meaning. Traditionally, the only modal expressions which constitute a well- defined class in English are the modal auxiliaries (Palmer, 2011). The modal auxiliaries can-could, may-might, shall-should, will-would and must all have clear formal markers that distinguishes them as a grammatical class. These are the NICE-properties:

5

• Negation (Auxiliaries can be negated • Inversion (they can be inverted in questions) • Code (they can be left stranded as in John can't sing, but I can) • Emphasis (they can be used in emphatic form).

Modal auxiliaries do not have an -s in singular third person and they cannot co-occur (there can be no double modals).

Seem and appear are not part of the closed class of modal auxiliaries (they don't fulfil the NICE- properties), nor have they yet found a place among the semi-modals, some of which are on the rise, aspiring to become part of the class of modals through the process of grammaticalization (Bybee et al, 1994, Leech, 2004, Krug, 2000). Semi-modals, or marginal modal auxiliaries, are need (to), dare (to), have to, have got to, used to, used to, be supposed to and be going to. Other emerging modals are had better, want to, be going to and would rather. This new generation of auxiliary verbs has been progressively emerging in Modern English and have taken over some of the functions of the traditional modals (Mitchell and Leech in Facchinetti, 2003).

However, there is an argument to be made that seem and appear have gone through a process of grammaticalization and have become and much like the modal auxiliaries in form and meaning (Johansson, 2001). The point to be made at this stage is that if we operate with a definition of modality as a grammatical concept, seem and appear would only qualify as modals if they were members of a closed grammatical class, no matter how modal their meaning. That's why grammatical properties of seem and appear and the syntactic constructions they occur in are presented in detail (3.1. and 3.2), and why a central discussion in this thesis is whether seem and appear in certain syntactic constructions could qualify as modals or as auxiliaries (6.4.)

2.3. Modality as a semantic concept As a semantic concept, modality is a category that covers the many different nuances of modal meaning conveyed by modal expressions. Even though the concept of modality has been studied extensively over the years, there is no unified view among linguists on "how to cut up the semantic space of modality into types of modality, or even the boundaries of modality" (Lewis, 2015). Several ways of organising and categorizing modality has been proposed in the literature. A three-way division between dynamic, deontic and epistemic modality is presented here. The typology below (adapted from Nuyts and Auwera, 2019) is based on shared semantic properties. The category of epistemic modality is more or less consistent across all typologies, but there is an ongoing debate on whether evidentiality is part of epistemic modality or not, and whether a speaker's attitude and opinion fall under the category of epistemic modality or is understood as a category of its own. Examples in this section are from Nuyts and Auwera (2019). Relevant parts are marked in bold.

2.3.1. Dynamic modality (sometimes called inherent modality) has to do with capability and ability. A distinction is made between participant-inherent dynamic modality, i.e. abilities that are fully inherent in the first-argument participant, for example

(1) He can stand on his head and participant-imposed dynamic ability, i.e. ability that is determined by external circumstances and may be beyond the participants control

6

(2) The garage is free, so you can park there.

2.3.2. Deontic modality is traditionally defined in terms of permission and obligation, including related notions of interdiction and advice. Most scholars include volition and intention in the category of deontic modality, but this is debated. An even broader definition of deontic modality is that it is an indication of the degree of desirability of the state of affairs in the utterance (3). In this definition, the concept of desirability should be taken widely: it could be anything from societal norms to personal ethics.

(3) It must be done

2.3.3. Epistemic modality The term 'epistemic' derives from episteme, the Greek word for knowledge, and epistemic modality has to do with knowledge and belief – or the lack of it. However, there are different takes on epistemic modality, too, and the fault lines goes between modal meaning connected to (i) speaker's subjective attitude and judgement, and (ii) modal meaning qualifying the factuality of a proposition:

(i) Epistemic modality expresses the attitude of the speaker towards the propositional content and marks the extent to which the speaker can commit himself to the state of affairs in the proposition – or the truth value of the proposition (Lyons, 1977, Palmer, 1986, Halliday 2004). Epistemic modal forms are thus "defined semantically as those linguistic forms which are used to indicate the speaker's confidence or lack of confidence in the truth of the proposition expressed in the utterance" and they can express the whole range of attitudes from confidence to doubt" (Coates, 2013). In this view, epistemic modality expresses the subjectivity of the speaker.

(ii) Epistemic modality is a category that qualifies the factuality of a proposition, and whether something is objectively true or not. More specifically, modal elements mark the state of affairs as undetermined in terms of factuality. This concept has its roots in the early ideas of the realis-irrealis-dimension of language and the notion of possible worlds, in which propositions or events were seen as 'real' or 'true', and modal expressions were defined relative to other world’s distance to the real world in the theory of modal logic (Perkins, 1983; Palmer 1986). This definition of epistemic modality is thus an estimation (typically by the speaker, but not always) of "the likelihood that the state of affairs expressed in the clause applies in the world" (Nuyts, 2019, 38).

There is some debate as to whether the truth value of a proposition should be a category of its own (alethic modality) pertaining to modes of truth. The underlying premise for separating truth value from epistemic assessment is that whether something is true or not is a separate notion from the speaker belief and commitment to the truth of a proposition. It is full possible to be strongly committed to the truth value of a proposition, even if what you are committed to is objectively wrong. With this division, whether something is true or not (modes of truth) is covered by the category alethic modality, while epistemic modality comprises modes of knowing and reliability of evidence. Coates (2013) for example, stresses that in everyday spoken interaction, epistemic modality is used to convey the speaker's attitude to the proposition, not to convey some objective

7 truth. Others rejects this division, and states that there is no difference between what is logically true and what the speaker believes to be true (Palmer, 1986 in Nuyts, 2019).

Epistemic modality is closely linked to evidence. Speaker's commitment to the truth value of a proposition may depend on factors connected to the evidence itself, such as source of evidence, and factors that are more epistemic in nature such as mode of knowing, strength of evidence, reliability of knowledge, speaker's deduction and personal judgement based on evidence which in turn influences speaker's certainty and commitment to the proposition. In some typologies, evidentiality is singled out a category of its own, separate from epistemic modality, others subsume evidentiality as part of epistemic modality. Evidentiality and the demarcation between epistemic modality and how that affects the analysis of seem and appear is discussed in more detail in 2.6.

Some typologies place the speaker's (or someone else's) attitude towards a state of affairs as a category of its own (boulomaic modality, sometimes called attitudal modality or emotional attitude). This is not very common, most likely because this type of modal meaning is not very relevant for the grammatical category of the modal auxiliaries – the study of which dominates the literature on modality. Instead, attitudal modality is usually perceived as a shared semantic characteristic of epistemic modality. This is the approach chosen in this thesis: Speaker's attitude is part of epistemic modality, and the division is made between what is primarily evidence-based and thus evidential, and what is primarily epistemic and expresses the speaker's attitude and belief.

2.3.4. Other possible schemas for organising modality Several other ways of organising and categorizing modality has been proposed in the literature (Quirk et al 1985, Palmer, 2001, Nuyts, 2019), but are only mentioned briefly here, since they are not used in this thesis.

The classic division is between root modality (the root sense of modal auxiliaries, e.g. obligation permission, ability etc, i.e. deontic and dynamic ability), and epistemic modality. This typology fits the closed class of modal auxiliaries, but it is less useful to study the many different lexical items with modal meaning that does not form a grammatical category, e.g. seem and appear. Another similar approach proposes two broad categories for modality: 'event modality' and 'propositional modality'. Event modality covers dynamic and deontic modality; propositional modality covers epistemic and evidential modality. In this typology, epistemic modality is further subdivided into speculative, deductive and assumptive modality, and evidential modality has the subcategories reported and sensory.

Another way of looking at modality is to regard all categories of modality as being concerned with possibility and necessity. The idea is that all categories of modality can be characterized in terms of these notions. The categories will have a value for necessity and possibility, and propositional content can be characterized as being more or less possible or more or less necessary. These schemas are not particularly useful approaches for seem and appear, therefore I have chosen not to pursue these dimensions in my discussions in this thesis.

Distinguishing between 'speaker-oriented' and 'agent-oriented' modality is another way of categorizing modality. Speaker-oriented modality covers markers of directives (e.g. imperatives, optatives and permissives). In other typologies these are categorized under the heading mood, not modality. Agent-oriented modality covers any meanings which predicate conditions on the agent,

8 e.g. obligation, desire, ability, permission and root possibility. Agent-oriented modality is thus identical to root modality mentioned above, while speaker-oriented modality covers epistemic modality. Speaker-oriented modality is in many ways a pragmatic approach to modality (2.4.)

2.4. Modality as a pragmatic concept Sometimes, the modal meaning of a lexical item is not coded in a word class, nor inherent in the lexeme, but is acquired when language is put to use. This brings the analysis of seem and appear into the realm of pragmatics. Epistemic modal markers and evidentials have pragmatic functions when they convey something which is not (or is not just) about the source of information in the utterance nor about the assessment of the truth value of the proposition.

For seem and appear, which do not fit neatly into a well-defined grammatical category, and are vague and multifunctional, a pragmatic approach is sometimes needed to discuss the modal and evidential qualities of the two expressions. The functions of seem and appear are typically to show stance, to hedge or to construe shared knowledge, to be a marker of politeness or a face-saving device, or to serve other pragmatic functions, such as topic transitions and making counter proposals. The different pragmatic functions of seem and appear are presented in Chapter 3.

2.5. The concept of evidentiality Evidentiality is a category of linguistic means that has source of evidence as its primary meaning (Brugman, 2015). The definition seems straightforward, but in linguistic literature there is little agreement as to what the category really covers. As is the case for modality, evidentiality can be defined as a purely grammatical phenomenon, a semantic category, or it can be described and discussed in terms of functions of evidential devices (Almeida, 2015: 126).

Evidentiality was first studied in languages that have evidentiality coded into their grammatical structure, e.g. verb suffixes in Native American languages. Later studies have shown that there is a wide range of evidential markers such as verb suffixes, sentence-affixes, lexical predicates or fully- fledged morphological categories. In English, evidentiality is expressed through a variety of devices. The most common ones are lexical verbs, but modal auxiliaries, , discourse particles and miscellaneous idiomatic phrases can also express evidential meaning.

Seem and appear are lexical verbs with semantic meaning relating to visual, sensory or inferred evidence and are examples of linguistic devices with evidential meaning (semantic approach) or lexical items that can function as evidentials (functionalistic approach). Some scholars argue that evidentiality should not be defined as a strictly semantic, grammatical or modification phenomenon, but should be understood as a functional-conceptual substance domain (Boye and Harder 2009 in Brugman, 2015). In such a schema, the function of seem and appear would determine if and how they belong to a category of evidentials or epistemic modals.

Lexical modal verbs are considered borderline between the lexical and the grammatical category, but a watertight division between lexical and grammatical evidentiality may not be very useful (Aijmer, 2009). The grammaticalizing seem and appear may have the semantic category or their functions as a starting point for the process of becoming full-fledged grammatical items like the modal auxiliaries.

9

2.6. The relationship between evidentiality and epistemic modality Epistemic modality is strongly linked to evidence, but the relationship between epistemic modality and evidentiality is less clear-cut than one could wish for. The main question concerns that which is not purely source of evidence, but more epistemic in nature. These are mode of knowing, strength of evidence, reliability of knowledge, speaker's deduction and personal judgement based on evidence – all of which influences the speaker's certainty and commitment to the proposition. Are the epistemic features part of evidentiality or not? And how do the two concepts relate to each other? Three main approaches are described below: epistemic modality and evidentiality as one category (2.6.1.), Evidentiality as a category of its own (2.6.2.), or a combination of the two where evidentiality and epistemic modality are considered two distinct categories with overlapping functions (2.6.3). The definitions matter, because the analysis of seem and appear and their correspondences depends entirely on where the boundaries are drawn between evidentiality and epistemic modality. The relationship between evidentiality and epistemic modality is particularly relevant for inferences, assumptions and implicatures. On the one hand, inference is a way of acquiring information through reasoning, and reasoning might be considered evidence. Reasoning is also sometimes triggered by and dependent on evidence. On the other hand, inferences could be considered to represent the speaker's deductive epistemic judgements. Whether seem or appear could be considered evidential or epistemic or both would thus depend on the approach.

2.6.1. A broad approach to evidentiality One possible relationship between epistemic modality and evidentiality is to include evidentiality in the definition of epistemic modality, or to include epistemic assessment and degree of certainty in the definition of evidentiality (Chafe and Nichols, 1986, Squartini, 2019: 58). The category of epistemic modality is expanded to cover both evidentiality and the source of the information and the epistemic modality – the mode of knowing, the reliability of the information, degree of certainty and speaker's commitment to the proposition. Another take on the broad theory of evidentiality is to make the definition of evidentiality broad enough to include a wide range of epistemological and pragmatic properties. Chafe (1986) for example, defines evidentiality as "marking the speaker’s attitudes to knowledge of reality", a definition which subsumes degrees of reliability as well as evidential meaning related to hearsay, induction and deduction. With inferences, there would be no distinction between encoded evidential meaning (expressing inference based on observational evidence) and implicated meanings, e.g. speaker's evaluation and judgement about the truth value of the proposition in this approach (Aijmer, 2019).

2.6.2. A narrow definition of evidentiality The narrowest definition of evidentiality restricts the linguistic category of evidentials to have only two obligatory properties that hold fast across all languages: They specify the source of evidence and they are members of a specific grammatical system. Other properties of evidentials, e.g. semantic and pragmatic properties like epistemic meaning, illocutionary meaning, speaker deixis and truth- conditionality vary across languages and must be empirically determined for individual items and languages. They are not inherent features of evidentiality in the strictest definition (Brugman, 2015).

Even if we define evidentiality primarily as a semantic category, a narrow definition would mean that evidentiality is concerned with source of evidence, and only that. In the narrow theory of evidentiality, evidentiality and epistemic modality are two distinct categories, and the link between them is indirect. Evidentials code for strength of evidence only implicitly, relying on a presupposed

10 hierarchy of strength of evidence, in which direct attested evidence is considered stronger than indirect reported evidence, second hand sources are more reliable than third hand sources and so forth. It is this implicit connection that, in turn, links evidentiality to how certain the speaker will be with respect to the propositional content and perhaps to his or her degree of commitment. (Givòn, 2001 in Brugman, 2015). Reliability of information, degree of certainty and speaker's assessment of the truth value of the proposition belong to the category of epistemic modality.

The relationship between evidentiality and epistemic modality can more easily be discussed within a framework where the two notions are kept apart (Aijmer, 2009, Corinillie, 2012, Almeida, 2015). The narrow theory of evidentiality, which restricts evidentiality to source of information, is thus arguably better suited to analyse and reflect upon the evidential and epistemic quality of lexical items such as seem and appear. In a narrow theory, inferences signalled by seem and appear would be evidential only when they code for source of evidence. Source of evidence may be both observational evidence and the reasoning process behind a proposition, and this would be different from the speaker's judgement of the likelihood that the proposition is true.

2.6.3. Mixed approach: The categories overlap or are subcategories of one another A third theory of evidentiality suggests that evidentiality and epistemic modality are two distinct categories with an overlapping relation "where modality and evidentiality partly intersect" (Dendale &Tasmowski, 2001 in Brugman 2015). Both evidentiality and modality may thus be equally present in inferences. There are several different takes on how to disentangle the concept of evidentiality from the concept of epistemic modality in a mixed approach.

One take is to see the two as separate concepts in general, but with overlapping meaning and function when it comes to inferences. When seem and appear express inference based on observational evidence, they would express evidentiality and epistemic modality simultaneously, and it would not be possible to separate one from the other. This would be much like epistemic must, which expresses both evidentiality (inference-based on evidence) and epistemic modality (personal assessment and evaluation) (Aijmer, 2009).

Another view, rooted in the non-factuality of modality, distinguishes between direct evidentials that are not modal, and indirect evidentials that are modal. Direct evidentials are evidentials based on direct perception. These are considered too factual to be considered modal. Indirect evidentials, however, are far less reliable and permits a lower degree of commitment to the factuality of the situation, and can therefore be considered genuinely modal (Squartini, 2019: 61). Another version of the mixed approach is to see evidentiality as a subcategory of epistemic modality. This approach sees evidential distinctions as part of the marking of epistemic modality (Willett, 1988 in Brugman, 2015).

2.6.4. Chosen approach to evidentiality For seem and appear, it matters a lot whether we see evidentiality as a category of its own as distinct from epistemic modality or not. In the narrowest definition of evidentiality, seem and appear would not qualify as evidentials, since they are not a closed grammatical class. However, if evidentiality is defined as a semantic category, seem and appear certainly have evidential meaning in some contexts, and would qualify as an evidential. In the broad definition of evidentiality, in which the category of evidentiality is expanded to include epistemic assessment in terms of degree of certainty, they could both qualify as evidentials.

11

For the purposes of this thesis, which is to determine any similarities and differences between seem and appear, and whether that is related to their evidentiality and epistemic modality, it is useful to consider epistemic modality and evidentiality as two distinct categories, allowing for the fact that they sometimes overlap when it comes to inferences. With this approach it is possible to determine if appear is more or less modal than seem, and discover if one of the expressions (for example appear) is primarily contingent on visual evidence and thus an evidential, whereas the other (for example seem) could be either evidential (when referring to evidence) or modal (when signalling personal judgment) or both (signalling personal judgement based on evidence). These differences would reflect a significant difference between the two expressions and influence how they are translated.

In this thesis I take this mixed approach. I define seem and appear as lexical verbs that can have evidential and/or epistemic meaning (one or the other or both in a semantic overlap), and that they can have several pragmatic functions, some of which they share.

Recent studies seem to support such integrated models of evidentiality and epistemic modality, in which "both categories interplay…and intermingle in various degrees" (Squartini, 2019: 64).

2.7. Type and source of evidence The epistemic assessment of a proposition depends on several variables connected to evidence beyond its source, for example mode of knowing, evidence strength, reliability of information and degree of certainty. There are different models for analysing different types of evidence and their relative strength. Two main approaches are presented below: The first approach includes typologies with only an implicit link to epistemic modality. The second approach includes epistemic modal values, such as 'mode of knowing' and 'degree of certainty' as variables (2.7.2). The chosen approach for this thesis is presented in Figure 1.

2.7.1 Typology of evidence excluding epistemic modality In the narrow approach to evidentiality, in which evidentials make up a distinct linguistic category with source of evidence as an obligatory property, source of evidence has the following four values: can be either sensory inputs, another's report, inference and general, cultural knowledge (Brugman, 2015).

A more detailed typology of evidence takes the binary distinction between direct and indirect evidence as its starting point and presents subcategories for each type (Willett in Brugman, 2015). Direct evidence is attested evidence, which can be visual, auditory or other sensory evidence. Indirect evidence has two subcategories: reported evidence and inferred evidence. Reported evidence includes hearsay (second hand or third hand reports) and folklore. Inferences can be based on observable results or one's reasoning, and the reasoning process is thus considered evidence (see 2.6.). The hierarchy of evidence is presented in Figure 1.

12

Figure 1. Evidential hierarchy after Willett (1988), based on adaptation by Faller (Brugman, 2015).

The model incorporates a presupposed hierarchy of strength of evidence. Direct attested evidence is considered stronger than indirect reported evidence, and second-hand sources are more reliable than third hand sources. Folklore is less reliable than other reports. Inferences based on visible results are arguably stronger than inferences based on the speaker's reasoning. The hierarchy of evidential strength in turn affects reliability of information and speaker certainty or uncertainty, but the link between evidentiality and speaker certainty or uncertainty of the truth value of the proposition (i.e. epistemic assessment) is indirect.

2.7.2 Models of evidence-types including epistemic modality Some typologies of evidence include epistemic modality. A multi-dimensional approach with the variables evidence type (sensory, visual, reported), evidence location (location of the speaker vis-à- vis the described situation) and evidence strength (reliability of evidence) has been proposed, for example (Nuyts, 2019).

Another model that includes reliability of knowledge, and thus epistemic assessment, as a variable was presented by Chafe (1986). The variables in this model are:

(i) source of knowledge (sensory evidence, language, a hypothesis). (ii) mode of knowing (one's belief, induction (from physical evidence), sensory evidence, hearsay (from language) or deduction (from a hypothesis). (iii) reliability of knowledge (iv) degree of match between the information or knowledge and the speaker's choice of linguistic device.

The different categories and subcategories in the different models correspond for the most part: 'sensory evidence' in Chafe's model and Nuyts typology corresponds to 'direct attested evidence' (visual, auditory or other sensory) in Willet's model.' Spoken or written language' as source of evidence corresponds to 'Indirect reported evidence', and Chafe's 'hypothesis as a source of knowledge' corresponds to 'indirect inferred evidence' in Willett's model.

With 'mode of knowing' as a dimension on its own, Chafe's model is more specific in its description of how knowledge is acquired. However, the same variables (except 'belief') are incorporated in

13

Willett's classification: Induction from physical evidence in Chafe's model corresponds to knowledge from attested evidence in Willett's model, 'hearsay' is a variable in both models, and what Chafe calls deduction from a hypothesis is covered by Willett's category of inferred evidence. 'Belief' as a source of knowledge is not part of Willett's model. Belief involves personal assessment and epistemic modality and Willett's model pertains only to evidence. It is possible to believe something without evidence. The most significant difference between the different models is thus the variable 'reliability of knowledge' which is explicit in Chafe and Nuyt's models, but implicit in Willett's model.

In my study, I analyse the type and source of evidence in the translation pairs according to Willett's typology. The assessment of the strength of the epistemic modality expressed is thus determined indirectly according to the hierarchy of evidence above. In my analysis, I have included 'no evidence/personal belief' as a category to determine which translation pairs do not refer to evidence at all, only to speaker's judgement, belief and epistemic assessment.

14

3. Grammar, syntax, semantic meaning and pragmatic function of seem and appear as presented in the literature Seem and appear are interesting and complex verbs. They are categorized as lexical verbs with modal meaning, but they are also members of the grammatical category of catenatives and they resemble auxiliaries in many ways. The modal and evidential meanings expressed by seem and appear are notoriously difficult to pin down. As sense verbs in the same vein as see, sound and feel, seem and appear have evidential meaning because they refer to physical evidence as source of knowledge (see 2.5 The concept of evidentiality). As lexical modal verbs they have epistemic meaning and denote uncertainty, non-factivity, and epistemic assessment (2.3.3.) Seem, but not appear, also has additional meanings closer to cognitive verbs like think. (Chafe 1986: 267, Aijmer, 2009: 72).

The difficulties in analysing seem and appear are multiplied by the fact that the two expressions can be used in many different syntactic structures, and that the meaning expressed in fact depends on the syntactic construction they occur in. Furthermore, the distinction is not between modal or not, or evidential or not, but is often a matter of degree and a mix of the two. The analysis of seem and appear will also depend on pragmatic factors such as subjectivity and intersubjectivity, stance, hedging and politeness. Moreover, all these factors can co-occur when we analyse specific examples (Aijmer, 2009).

To analyse seem and appear, it is necessary to analyse them at the levels of grammar, syntax, semantics and pragmatic effect. In the present chapter I present the levels of analysis that are relevant for my study, the distinctions that are made and the terminology I have used.

3.1. The grammar of seem and appear As lexical modal verbs, seem and appear have the same grammatical properties as other lexical verbs. They show person, mood, tense and number. They keep the -s in third person singular present and have do-support. They also belong to the grammatical class of catenatives, or chain verbs as they are sometimes called. Catenatives can link with other verbs to form a chain of verbs. A catenative takes as its complement a non-finite construction, often an infinitive (see section 3.2.).

Seem and appear are not part of the closed class of auxiliaries, nor have they yet found a place among the semi-modals need (to), dare (to), have to, have got to, used to, used to, be supposed to and be going to or other emerging modals are had better, want to, be going to and would rather, which are on the rise, aspiring to become part of the class of auxiliaries through the process of grammaticalization (Bybee et al, 1994, Krug, 2000, Leech, 2004). Grammaticalization is the process of change in languages when lexical verbs with semantic meaning over time become function words, like prepositions or auxiliary verbs; or grammatical markers, like affixes and suffixes. Modal auxiliaries, which developed from the lexical verbs willan, cunnan and motan in Old English into the modal auxiliaries will-would, can-could and must are a prototypical example.

However, even if they do not share the four properties that distinguish auxiliary verbs from other verbs (that auxiliaries alone can be negated, be inverted, show code and will allow a following verb phrase to be ellipted and that they can be emphasized) (Palmer, 2001), they do share many grammatical properties with the auxiliaries (2.2.). They are used in the same position as the

15 auxiliaries, i.e. before a non- form. The difference is that seem and appear are followed by a to-infinitive clause rather than a bare infinitive. They also resemble the auxiliaries in that they are generally semantically independent of the subject (Johansson, 2001), see also section 3.7. Furthermore, seem and appear alone can be negated like the auxiliaries in some syntactic constructions (4), and I will argue that they can be emphasized (5):

(4) It appears/seems not (5) She APPEARS/SEEMS to be okay.

In this thesis I will make the argument that seem (and to a lesser degree appear) in certain constructions may be going through a process of grammaticalization, approaching the modal auxiliaries in form and meaning. An important stage in the process of grammaticalization is the loss or reduction of lexical meaning. This process is called semantic bleaching. A high number of different correspondences and variety among them is a sign of semantic bleaching. As this study will show, seem and appear have a high number of different correspondences, and they have many different senses. This is especially true for seem, and could signify that seem, but also appear, are becoming semantically bleached as part of the process of grammaticalization (6.5.).

3.2. Syntactic structure Seem and appear can be used in many different syntactic structures, but they are not identical with respect to the syntactic structures they can occur in and the structures they favour. Below is a list of possible syntactic structures for seem and appear, in which the similarities and differences between the two are emphasized. The list of syntactic patterns and the examples are adapted from the online edition of the Oxford English Dictionary, The Cambridge Grammar of the English Language (Huddleston and Pullum, 2002) and Hasselgård et al. (2004).

3.2.1. A simple SV-patterned sentence (/ + appear) Appear can occur as a one-place verb in a simple sentence with SV-pattern. It is possible to say

(6) The sun appeared.

However, even though it is possible to make a simple SV-sentence with appear, it is not very common. The less marked use of appear is to add one or more optional adjuncts to the SV-sentence, e.g. a time, place or manner adverbial.

(7) The sun appeared on the horizon before six o'clock in the morning (8) Suddenly, he appeared in the doorway.

This SV-pattern is not possible with seem. Seem is a two-place verb in that it requires a minimum of two arguments. It is not possible to say

(9) She seems*

The obligatory second argument for seem is typically a complement (an , a noun, a prepositional phrase or a clause, see 3.2.6., 3.2.7., 3.2.2., 3.2.4. below) which ascribes a quality to the subject. With seem, a time or place adverbial will not suffice. Both seem and appear are intransitive verbs that do not take an object as the second argument.

16

3.2.2. Catenative constructions (seem/appear + to-infinitive clause) Being catenatives, seem and appear can link with other verbs to form a chain of verbs. They take non-finite verb clauses, as their complement, e.g. to-infinitive clauses with a (10), to- infinitive clauses with to be plus an adjective (11) or to-infinitive clauses with to be plus a present (12). Seem and appear can also be followed by a perfect participle-clause, e.g. the infinitive form of the perfect auxiliary have plus a past participle (13).

(10) They appear/seem to agree. (11) She appeared/seemed to be anxious. (12) She appears/seems to be trying to persuade him to help her redecorate her flat. (13) She appears/seems to have forgotten what he said earlier.

In this thesis, follow Huddleston and Pullum and use the term catenative for both the non-finite complement and the verb that licences it, i.e. the entire construction of verb and complement (Nordquist, 2018). Seem and appear plus any of the to-infinitive clauses above are referred to as catenative constructions in this thesis.

3.2.3. Constructions with existential there (there + seem/appear + to be) Seem and appear also occur in constructions with existential there (14). In these constructions they function as auxiliaries to the verb to be.

(14) There appears/seems to be two buttons missing on my jacket.

3.2.4. Syntactic constructions with a that-clause (seem/appear + that-clause) Both seem and appear can also be followed by complements in the form of that-clauses (15):

(15) It seems/appears that it is raining.

In constructions with a that-clause, the subject is always anticipatory it.

3.2.5. Attributive complement in the form of an adjective phrase or a past participle (seem/appear + adjective and seem/appear + past participle) Both seem and appear can be followed by a complement in the form of an adjective phrase (16). Seem and appear also take a past participle as a complement. Huddleston points out that when the past participle is used with seem and appear, it is always an adjective (17).

(16) She appears/seems capable. (17) It appears/seems broken.

3.2.6. Syntactic constructions with an adjective and a that-clause (appear/seem + adjective + that-clause) Seem can also easily combine with an adjective and a that-clause, e.g. with the adjective possible and a that-clause in (18). The syntactic construction is also possible with appear, but in this pattern, there are significant differences between seem and appear. Appear in combination with some comes across as marked and not idiomatic (19). In combination with other adjectives, e.g. likely and probable, it works just fine (20).

17

(18) It seems possible that it will rain (19) *It appears possible that it will rain (20) It appears likely that it will rain tomorrow

3.2.7 Attributive complement in the form of a noun phrase (appear/seem + noun phrase) Seem, but not appear, can combine with a complement in the form of a noun phrase (21).

(21) She seems a capable person

Any combination of appear with another noun phrase would have to be made by way of a to- infinitive clause (22):

(22) She appears to be a capable person

3.2.8 Attributive complement in the form of a prepositional phrase (seem + prepositional phrase) When a prepositional phrase is used as a predicate, it usually has idiomatic meaning and often resemble adjectives in that they ascribe a quality to the subject. This construction is only possible with seem (23) and (24).

(23) She seems completely at home in politics (24) They all seemed in favour of the change

Again, combining appear with an attributive complement in the form of a prepositional phrase would have to be done with a to-infinitive clause (25).

(25) They all appeared to be in favour of the change.

3.2.9 Syntactic constructions with the conjunctions as if, as though and like (appear/seem + as if/as though) Both seem and appear can occur in syntactic constructions with the conjunctions as if and as though (26). These conjunctions are used to make comparisons, and to talk about an imaginary situation or a situation that may not be true, but that is likely or possible (Cambridge Online Dictionary). When combined with the verb in the past subjunctive or the past perfect subjunctive, this syntactic construction is used to describe an unreal or improbable situation (27).

(26) It appears/seems as if it is raining (27) She seemed as though she were dying.

In informal English, like can be used in a similar way to as if, though it is not always considered correct in formal contexts (Cambridge Online Dictionary). The construction with like is only possible with seem (28).

(28) It seems like it could rain any minute.

18

In this thesis, the constructions with as if, as though and like, and their Norwegian correspondences, syntactic constructions with som and som om, are referred to as comparative constructions and hypotheticals.

3.2.10. Syntactic constructions with like and a noun phrase (seem + like + noun phrase) Seem, but not appear, can also combine with like and a noun phrase:

(29) He seems like a fool

3.2.11. Parentheticals (it + seems/appears) Parentheticals are syntactically unintegrated elements which are separated from the host clause by comma, and in spoken English often by intonation. Because they are syntactically independent, they can occur in medial or clause-final positions. Parenthetical expressions add information but can easily be removed without destroying the meaning or grammar of the main sentence. Parenthetic seem is frequently used (30). It is possible, but less common to use appear as a parenthetical (31).

(30) They're not coming after all, it seems.

(31) The government, it appears, had covered up the scandal.

This construction is also called a comment clause.

3.2.12. Syntactic constructions with the pro-clause so or not (seem/appear + so/not) Another possible construction with seem and appear is in combination with the pro-clause so, or not:

(32) It seems/appears so (33) It seems/appears not.

The combination of seem and appear and the pro-clause so or not is only possible with empty it as the subject, not when the subject of the clause is a noun or a pronoun (34).

(34) She appears/seems not*

3.3 Syntactic function Both seem and appear are multifunctional. This section presents the different functions of seem and appear in the syntactic constructions presented above.

3.3.1. Attributive or equative function Seem and appear are copular verbs, sometimes called linking verbs. These are verbs that connect the subject to a complement. The complement attributes some quality to the subject, e.g. their appearance, their character or their behaviour. As shown in sections 3.2.5., 3.2.6., 3.2.7., 3.2.8., the quality that is ascribed to the subject can be expressed in the form of an adjective phrase or a past participle, an adjective phrase plus a that-clause, a noun phrase, a prepositional phrase or a to- infinitive clause of to be plus a noun phrase. This is the attributive, or ascriptive, function of seem and appear.

19

Another function of linking verbs is to identify a relationship or an existing condition between the subject and the predicate. The relationship can be between , adjectives and clauses. Constructions with existential there are examples of how seem and appear refer to existence. Also, the constructions with empty it are commonly used to describe a condition, typically about the weather, the temperature, time and distance (Hasselgaard), see for example (26) and (28) in section 3.2.9. above.

The choice of appear or seem instead of copular be is a well-established strategy to introduce uncertainty and epistemic modality into an otherwise straightforward relation between the subject and the complement, whether it is an attribute or an existing condition. Replacing be with seem or appear has sometimes been likened to altering an equalizer-sign to an approximately equal-sign.

As linking verbs, seem and appear are also used to make comparisons. Constructions with as if, as though are typically used to make a comparison to an imaginary situation or a situation that may not be true but that is likely or possible (Cambridge Online Dictionary), or an unreal or improbable situation (3.2.9). These are the hypotheticals (3.2.9). Comparative constructions with seem/appear and like (3.2.10.) also typically make comparisons between how something how something looks or is perceived to some pre-established notion. These are the more straight-forward comparisons. However, seem and appear express less reliability than constructions with other verbs like look like, sound like and feel like that also indicate knowledge derived from sensory evidence.

3.3.2. Highlighting the speaker as experiencer I will argue that another function of the verbs seem and appear is to introduce the speaker as the ultimate experiencer. This function is tied to seem and appear as sense verbs. All sense verbs (such as appear) and the verbs of seeming (such as seem) licence a to-phrase where the oblique noun (to me) phrase expresses the experiencer (35). But it can easily be dropped from the clause, in which case the experiencer is only implied (36).

(35) The proposal looks (seems) very promising to me. (36) The proposal seems promising.

With both seem and appear it can be argued that that the ultimate experiencer is always the speaker. This is true, even if the explicit experiencer in the clause is an oblique object in the form of a third person noun or pronoun. For example, in (37), the explicit experiencer is the third person pronoun them. But it is really the speaker's assessment of the situation that is conveyed through the choice of seem or appear. Therefore, the speaker is the ultimate implied experiencer of the situation in the proposition, whereas the experiencer of the different elements in the proposition (e.g. the newness of the idea) is the explicated third party.

(37) The idea appeared/seemed to be new to them.

Of course, the speaker, or the narrator, is always a factor to be reckoned with in every third person narrative, but the effect of using the sense verbs seem and appear is that the speaker is given a stronger presence in the proposition. As sense verbs, they connect the speaker's perception with the

20 proposition. To what extent the speaker's perception is based on evidence, is an observation-based inference or an inference based on logic and reasoning – i.e. whether they convey evidentiality or epistemic modality or both, is a point of debate at the heart of this thesis (see section 6.2.3.).

3.3.3. Changing the subject role and thematic structure Another function of seem and appear can be to play around with the grammatical and semantic subject for certain effects or change the thematic structure of a sentence. Seem and appear are so- called raising verbs, that can displace a noun phrase from a position within an embedded clause to a subject position in a higher clause (Oxford ). Typically, a raised subject is a grammatical subject that carries a semantic role associated with a verb in a lower clause. Thus in

(38) Henriette appears/seems [ _ to like Paul] the subject, Henriette, is related syntactically to the verb seem in the matrix clause, but semantically to the verb like in the lower clause. The subject of the matrix clause (Henriette) is said to have been raised out of the bracketed clause from the position indicated by ‘_’. The effect of raising the subject and placing it in the thematic position, is to emphasize Henriette's role as the experiencer (of like), while keeping the modal meaning of seem. The displaced subject is called a raised subject.

Other syntactic constructions with seem and appear, for example in constructions with anticipatory it or existential there, move the real subject towards the end of the sentence, giving it more importance, according to the principles of end-focus and end-weight (the information principle).

In many ways, it can be argued that it is their function of raising the subject that makes seem and appear auxiliary-like. When the subject is raised form the lower clause the straight-forward relation between the subject and the main verb (that Henriette likes Paul) is interrupted by the modality or evidentiality imbued in seem and appear. In addition, as raising verbs seem and appear are grammatically connected to the subject, but their function is to convey epistemic modality (the source of which is the speaker, not the grammatical subject) in relation to the main verb like. The main verb is thus modified by seem or appear, but seem (or appear) has an independent connection to the subject, similar to epistemic must in Henriette must like Paul.

3.4. Semantic meaning The next level of analysis is semantics. Under his heading, I will summarize what seem and appear mean according to the dictionaries. I will also discuss briefly evidential meaning and epistemic modal meaning, and the topic of vague language.

3.4.1. Meaning of seem and appear according to the dictionaries Etymologically, appear stems from Latin appārēre, which means "to come into sight, be visible, be evident," and from the combination ad + pārēre, which means "to be visible, be evident, be subject (to), comply (with)," of uncertain origin. Appear is listed with several meanings: It describes how something looks, it is used as a synonym for seem, it can mean "to begin to be seen" or "come into sight", "to perform publicly in a film, play or a show", "to start to exist" or "become available" or "to arrive" (Cambridge English Dictionary).

21

As a sense verb, appear refers to physical experience and knowledge that has been derived from sensory evidence. As this thesis will show, it is frequently used to refer to visual evidence. Lexemes that refer to source of evidence are said to have evidential meaning.

Seem, from Middle English semen which meant "to appear to be" or "to be fitting", is of Scandinavian origin; akin to Old Norse sœma (to honour), sœmr (fitting), and samr (same) according to the Merriam Webster Online Dictionary. In the 13th century, it took on the meaning ‘to have a semblance or appearance’. The meaning of seem today according to dictionaries is 'to appear in a particular way', 'to give the effect of being', 'to be judged to be' (Cambridge Online Dictionary), or 'to give the impression of being' and 'to appear to the observation and understanding' (Merriam Webster). The Oxford Online Dictionary notes that the modern use of seem, in combination with an infinitive, is often equivalent to the finite verb qualified by ‘probably’ or ‘if the evidence may be trusted.’ The Cambridge Dictionary describes the difference between seem and appear this way: Both expressions refer to facts and events, but seem also incorporates the sense of personal judgement and ideas.

3.4.2. Evidential and epistemic meaning For the purpose of describing the different levels of analysis, it is important to point out that the semantic properties of seem and appear, i.e. what they mean, is inextricably entangled with source and type of evidence, (evidentiality) and personal judgement, epistemic assessment and degree of certainty (epistemic modality). The Semantic Mirror-analysis in Chapter 5 will exploit the fact that translations try to incorporate all these levels of meaning and functions, therefore it is possible to use translation data to determine different semantic meanings of seem and appear. The method unveils the different senses of each token, and if and how they share semantic meaning – which they do (5.1.5. and 5.2.5.)

This thesis also explores if and how semantic differences between seem and appear and their correspondences depend on how modal or how evidential they are, i.e. their evidential and epistemic modal meaning. There is also the added difficulty that epistemic assessment is sometimes – if not always – based on evidence. Therefore, seem and appear may have both evidential and epistemic modal meaning. This is especially relevant in inferences. Epistemic modality and evidentiality in inferences are discussed in more detail in 5.6.

As mentioned in the Introduction (Chapter 1), the relationship between syntactic form and epistemic and evidential meaning is also investigated in this study. The findings are presented in chapter 5.This study show that seem and appear have different meanings and functions in different syntactic constructions, and that they are not similar in this regard: seem in catenative constructions is predominantly evidential and seem plus a that-clause is more epistemic, but appear in catenative constructions is more epistemic than appear plus a that-clause, and appear plus a that-clause is predominantly evidential (5.6.).

3.4.3. Vague language Both seem and appear, particularly seem, are so-called vague expressions. Vague expressions are lexical items that are general and vague as opposed to specific. According to the Cambridge Online Dictionary, we use vague language when we don’t know the name of something, to make things sound less factual, or to talk about groups and categories. Vagueness is usually defined as different from ambiguity, which means that a word or an expression has several distinct and unrelated

22 meanings. With vagueness, distinct meanings cannot necessarily be identified. Vague words are also generally perceived as different from fuzzy words: Fuzziness has to do with boundaries between words and meaning and how it is not clear when one word begins and another ends. (Ruzaité, 2006, page 38).

Vague words typically have many different correspondences. This is true of both seem and appear. Previous studies have shown, and this thesis will confirm that neither have one obvious, or prototypical, translation in Norwegian. The ENPC provides a great many different correspondences (see the total number of correspondences in 4.5.), which suggests vagueness that the translator has struggled to find the right correspondence. Vagueness and a wide variety of correspondences could be a sign of semantic bleaching and an ongoing grammaticalization process (6.4.).

3.5. Pragmatic functions The meaning of modal items, especially modal lexical verbs like seem and appear, are often contextually determined (see chapter 2.4.) The most relevant pragmatic functions for seem and appear are discussed below. Examples are from the ENPC, but not exclusively may own data set.

3.5.1 Epistemic stance ‘Stance’ is commonly understood as the expression of a speaker or writer’s attitude, perspective, point of view, standpoint, opinion or position, including in its definition degrees of certainty, knowledge, degrees of commitment to truth of propositions and sources of knowledge, among other epistemic qualities" (Orch, 1996 in Ajmer, 2019). Epistemic stance is distinct from affective stance, which has to do with feelings or emotions, but both types of stance can be expressed in a situation.

This broad definition of epistemic stance entails all the pragmatic functions of epistemic modal markers: the function of construing hypothetical situations and possible worlds, displaying certainty and uncertainty (hedging), signalling of commitment and detachment, defining common ground and establishing 'sources of knowledge'. Consequently, the pragmatic functions of epistemic stance markers include functions of evidentiality, for example the construing of knowledge and the speaker's role in construing the information that is the basis of the modality.

Haddington (2004) points out that the notion is in danger of becoming an "all-embracing and elusive notion…which includes everything and explains nothing" and which is used to describe phenomena that have already been described by using other notions or terminology. In an attempt at unpacking the concept, Haddington distinguishes between 'stance' as a subjective attitude and 'stance taking', which is active use of multiple linguistic resources and practices in an interactional setting.

The starting point for investigating stance as attitude is usually a linguistic form, for example a syntactic unit or a word. This applies to the present study of seem and appear as evidentials or epistemic modal markers. In my analysis (Chapter 5), I have scoped down the concept of epistemic stance so that only the function of marking speaker's commitment and detachment to the proposition is included in epistemic stance. (39) is an example of appear functioning as a marker of detachment. The speaker is distancing himself from the propositional content that Louise shared the contempt for the New World. (40) is an example of the speaker taking ownership of the propositional content. The lack of distance in (40) is of course further emphasized by the oblique object to me, which makes the speaker the explicit experiencer:

23

(39) Many of these relatives still lived in Germany and had the Old World's contempt for the New, a contempt in which Louise appeared to collude. (RF1)

(40) It seemed to me that whatever else was true, it was absolutely the case that Ty deserved to realize some of his wishes. (JSM1) This study investigates similarities and differences between seem and appear with respect to how often and in what context they function as markers of stance and attempts to determine whether their function of marking stance is connected to type of evidence. The findings suggest that there is a difference, and that it is related to the evidential and epistemic modal meaning imbued in appear and seem respectively (6.3.).

3.5.2. Hedging Hedging is generally understood as "a discourse strategy that reduces the force or truth of an utterance and thus reduces the risk a speaker runs when uttering a strong or firm assertion or other speech act" (Kaltenböck, 2010). When the term was first introduced in 1972, the original definition of hedges was that they were "words whose job it is to make things fuzzier or less fuzzy" (Lakoff, 1972: 195 in Fraser, 2010). For some time, they were considered marginal, even redundant, items that contributed little to the communication. Today, they are generally acknowledged as playing a crucial role in both spoken and written discourse, and that understanding a text not only depends on understanding the factual content, but also the writer's intention which is often encoded in hedges (Mkhitaryan, 2015).

Related to the concept of hedging is the mitigation perspective; which is all about reducing the unwelcome effects on the hearer (Caffi in Fraser, 2010). Mitigation introduces the three concepts of bushes, hedges and shields. These concepts are three types of mitigation that reflect which part of the utterance they operate on: the proposition, the illocution and the utterance source. Bushes are lexical items that reduce the commitment to the propositional content, may introduce vagueness in the interpretation of the utterance and affect the truth value of the proposition. Hedges are lexical expressions that lessens the force of the speech act (e.g. a performative or a declaration) and shields are devices that avoid or shift the responsibility of the utterance, e.g. by introducing a different speaker (Fraser, 2010). All three are relevant subcategories for seem and appear, since both expressions clearly can reduce speaker's commitment to the content of the proposition, can lessen the force of a declarative sentence, and can sometimes shift responsibility for an utterance from the speaker to the evidence at the basis for a proposition or introduce the speaker as the ultimate experiencer of a proposition, see sections 3.3.2. and 5.5.1.

Over the years a significant number of subcategories of hedging has been introduced, such as 'adaptors', 'agent avoiders', 'approximators', 'attribution shields', 'bushes', 'committers', 'compromisers', 'diffusers', 'diminishers' and 'down-toners' to name a few from the first section of an alphabetical list (Fraser 2010). However, a main distinction can be made between propositional hedging and speech act hedging. In this view hedging is defined as "a rhetorical strategy, by which the speaker, using a linguistic device, can signal a lack of commitment to either the full semantic membership of an expression (propositional hedging)…or the full commitment to the force of the speech act being conveyed (speech act hedging)" (Fraser, 2010: 22). Hedging thus affects the propositional content of a proposition e.g. by introducing an overall vagueness or questioning the truth value of the proposition. Hedges also reduce the illocutionary force of the utterance and speaker's commitment in general. In this view reinforcements, or boosters, are not part of hedging.

24

Hedging is a particularly relevant concept for the lexical modal verbs seem and appear. Seem is often singled out as a modal expression with the pragmatic effect of signalling that there is some tension between what the speaker thinks and other, more argumentative voices.

Most scholars agree that there is no grammatical class of hedges in English, and that almost any linguistic item and expression can be interpreted as a hedge. According to Fraser, an expression is only recognized as a hedge when it is used in hedging (Fraser, 2010). Typical lexical items that function as hedges are:

• Classical approximators (sort of, kind of, so to say). These hedges affect the truth value of the propositional content. Expressions like sort of and kind of often co-occur with expressions of epistemic modality and with cognitive verbs used as parentheticals, such as I mean, I know, I think, I believe, I feel, it seems (my emphasis), and when they do, they assign some degree of intersubjectivity to the hedge, i.e. they communicate shared knowledge (Ketzner, 2010) When they co-occur with verbs of communication, the hedging attributes a lower degree of speaker commitment. • Epistemic phrases (I think, I guess, it seems - my emphasis). Comment clauses with epistemic meaning can function as hedges. • Hedged performatives (I would suggest). Hedging can lessen the speaker's intention in speech acts such as making a promise or giving a warning and can weaken speaker commitment in general. • Discourse particles. The Norwegian modal particles are examples of how discourse particles can function as hedges. In my material, modal particles are frequent translators of seem, but not appear. • Placeholder nouns (thing). Vague and unspecific nouns can contribute to overall vagueness and non-commitment. As vague expressions, seem and appear can act as hedges. • Approximation of numbers (about half, thousands, fifty-ish). Approximative numerals and numerical expressions are typical hedges. • Adverbs and adverbials (supposedly, presumably, apparently). Hedges were traditionally examined under the header of adverbs and adverbials. They are sometimes called stance adverbs – adverbials that overtly mark a speaker's or writer's attitude to a clause or comment about its content. Stance adverbs are divided in three categories: epistemic, attitude and style. They can express a) certainty and doubt, b) actuality and reality, c) source of knowledge, d) limitation, e) viewpoints or perspective and f) impression (Ketzer, 2010). • Impersonal and passive constructions and if-clauses can also function as hedges.

I include the list of typical hedges, because seem and appear make the list several times, and because the two verbs can be used interchangeably with many of the lexical items listed above, such as modal adverbials and epistemic comment clauses. Furthermore, the Norwegian correspondences of seem and appear manifest themselves in various forms, all found on this list.

The functions of the different linguistic items that functions as hedges that are relevant for seem and appear (Kaltenböck, 2010, Ketzer, 2010 Kärkkainen, 2003) are listed below.

Adaptors indicate that the speaker does not have enough evidence to present a clear-cut case of classification, and therefore provides a rough classification. The classical approximators (roughly speaking, in a way, or something like that, rather, like, kind of, sort of) often co-occur with

25 expressions of epistemic modality and with cognitive verbs used as parentheticals, such as I mean, I know, I think, I believe, I feel, it seems. When they do, they signal that the basis for the shared knowledge may be negotiated, i.e. they assign some degree of intersubjectivity to the hedge. Seem and appear in the comparative constructions (when they combine with like, as if or as though) functions as adaptors:

(41) It seemed as if the whole world was there. (BO1)

Modifiers and boosted modifiers are linguistic items that serve as a signal for loose interpretation of the lexical choice, a signal of indeterminateness and vagueness or they serve as a signal of hesitation and self-correction. When they function as modifiers, seem and appear display uncertainty and doubt and reduce the speaker's commitment to the truth value of the propositional content. In (42), the choice of seem signals that the speaker is hesitant to make a more categorical statement, and seem functions as a hedge in the form of a modifier, signalling uncertainty and doubt and reduced commitment truth of the proposition.

(42) … he seemed to be losing weight. (AT1)

However, the two expressions could also function as boosted modifiers and signal heightened speaker's commitment (speaker assertion). Boosted modifiers, which are used to take on the stance as knowledgeable, are typically adverbs like certainly, or the use of evidentials, i.e. lexical items that communicate about the source of evidence. In (43) appear communicates that the speaker has made relevant observations supporting his statement. Both appear and seem can be evidentials and function as boosted modifiers by referring to type and source of evidence and strengthen the speaker's commitment to truth depending on the strength of the evidence. Seem can also emphasize speaker's subjective judgment independent of evidence. This is particularly true in constructions with the speaker as explicit experiencer (44).

(43) He also appeared to have decided that the examination was over, and started to look round for his hat. (OS1)

(44) They seem to me to rely too much on forms of life we already know. (CSA1)

Adjusters signal that the information is sufficient, even if it is approximate, for the hearer to be able to understand the meaning. This type of hedge can express evidential meaning, in particular imprecision, and thus affect the truth value of the proposition. Approximators and epistemic phrases are examples of lexical items that typically function as adjusters. Constructions with seem and appear in comment clauses (45) are examples of this subcategory of hedges (3.2.11.).

(45) "The children in Little Weirwold have been quite spoilt, it seems," commented the Doctor. (MM1)

3.5.3. Politeness/face-saving function Modal items can also function as politeness markers and have the effect of saving face. Face-saving in pragmatics is to mitigate or avoid face-threatening speech acts, such as requests or insults. The politeness marker can save the face of the speaker or the ones he or she is addressing. A possible reason for modifying a speech act may be, for example, to increase solidarity (positive politeness if

26 the speech act is favourable to the hearer) or to express more distance if the speech act can be experienced as intrusive (negative politeness) (Holmes 1984, Ajmer, 2019). Showing sensitivity to others' feelings and avoiding expert status are two other pragmatic functions related to politeness (Coates, 2013). With seem and appear, the speaker takes on some responsibility for the interpretation of the situation, because of the implied experiencer (it seemed/appeared to me), and the choice of seem or appear thus functions as a face-saving device (46).

(46) "You seem to be unaware," he said, "that this is a private lounge. (AH1) .

A related function is Self-repair. This happens when a lexical item hedges strong opinions by toning down the impact of the statement to reduce the imposition on the hearer.

(47) The Soviet Union seemed to be governed by a gerontocracy, and the Politburo itself might have been an adjunct of the Kremlin hospital. (MAW1)

The function of establishing common ground and providing source of knowledge is sometimes considered a function of hedging. In this thesis, establishing common ground is considered a function of all the subcategories of hedges.

3.5.4 Subjectivity and intersubjectivity Subjectivity and intersubjectivity are central concepts when discussing epistemic modality and evidentiality from a pragmatic perspective. In linguistics, subjectivity is generally defined as the linguistic expression of speaker involvement. A modal item that signals speaker's assessment and that the assessment is the responsibility of the speaker, is subjective. Intersubjectivity is defined "as the sharing of experiential content (e.g. feelings, perceptions, thoughts, and linguistic meanings) among a plurality of subjects"(Zlatev et al., 2008: 1 in Almeida, 2015: 130). A modal evaluation is intersubjective if it is presented as being shared by the assessor and a wide group of people. Intersubjectivity is thus about how the speaker and the audience construe knowledge and define common ground.

Both subjectivity and intersubjectivity are clearly linked to the notions of epistemic modality and evidentiality. Modality is sometimes defined as the grammaticalization of the speaker's subjective attitude. In cognitive linguistics, the link between the different concepts has resulted in a threefold distinction of evidentials, namely explicit (I saw), implicit (epistemic modals) and opaque (it seems)' (Almeida, 2015: 130).

3.5.5. Other pragmatic functions Other pragmatic functions for epistemic modal items are

• The function of achieving certain conversational outcomes in certain sequential slot (e.g. to avoid disagreement in certain slots) • The function of regulating aspects of interaction (like topic transition) • Articulation and rhetorical strategies • Make counter proposals

These functions are not variables in my analysis and are not discussed in detail.

27

3.6. The relationship between form, syntax, meaning and function Studies (of seem) have also shown that there is a is a strong correlation between meanings and functions and the syntactic frame the two tokens occur in. Aijmer (2009) showed that most syntactic constructions with seem were primarily epistemic in meaning, but that the construction seem with a to-infinitive clause were evidential. It follows that appear, too, may have epistemic modal meaning in some syntactic constructions and evidential meaning in others. This will be analysed and discussed in more detail in 5.5.

3.7. Previous studies of seem and appear Since the modal auxiliaries are the only modal expressions which constitute a reasonably well- defined class in English, studies on modal auxiliaries have attracted the most attention, and the number of studies of the modal auxiliaries – and their translations – are exhaustive (Palmer, 2011). Contrastive studies of modal markers like modal auxiliaries and pragmatic markers show that they are translated using a wide variety of forms. Zero-correspondence, i.e. omission instead of translation, is the most frequent translation strategy from English into Norwegian (Johansson 2007, 296-299; Hasselgård 2006: 109–110 in Malmkjær, 2018).

Seem has also been widely studied. Johansson's study from 2001 focused on the fact that seem sometimes disappear without a trace in translations into Norwegian, and likewise might be added, seemingly without any motivation in English translations of Norwegian texts. In a cross-linguistic contrastive analysis of seem, Johansson used data from the ENPC to show that the English verb seem has been involved in a process of grammaticalization, and an accompanying bleaching of meaning. Johansson cites the wide range of correspondences as evidence that seem has become semantically bleached and that the translator struggles to find ways of expressing the meaning of seem in another language. In the same study, he found that the catenative pattern was often translated with a clause in Norwegian.

The high number of zero correspondences also supports the theory of semantic bleaching and grammaticalization. In this, Johansson points out, seem resembles the Norwegian modal particles – a small class of uninflectable lexical items that carry textual or interpersonal meaning. Modal particles, sometimes called pragmatic particles or discourse particles, are found in Norwegian, Swedish, Danish and German, but not in English (Andvik, 1992, Johansson 2001). Modal particles are frequent Norwegian correspondences of seem, which testifies to the modal meaning of seem being similar to that of the modal particles. Johansson and Løken (1997: 164–165) found that the Norwegian modal particles nok, vel, visst and sikkert are more frequent in original texts than in translations. This is most likely due to a lack of equivalents in English. They also found that discourse particles were often added in Norwegian translations where there was no obvious source. Aijmer (1996) showed that the same is true for their Swedish correspondences.

Another empirical, corpus-based study of different syntactic constructions with seem in English and their Swedish correspondences identifies the degree of certainty expressed by seem in various syntactic constructions and its relationship to type of evidence Aijmer (2009). Seem can for instance have the meaning ‘appearance’ and hedge probability or indicate hearsay depending on the construction it occurs in. The different constructions with seem and their functions are explained in a grammaticalization perspective. According to Aijmer, seem to is the most grammaticalized form and can be regarded as an evidential marker which is similar to the modal auxiliaries, and that it seems (that) is similar to an adverbial. Aijmer's findings is in line with previous findings that catenative

28 constructions with seem (seem plus a non-finite verb) are generally thought to show meaning related to aspect and modality (Quirk et al, 1985: in Johansson 2001: 232). These are meanings that are typically expressed by the class of modal auxiliaries in English.

Previous studies of the syntactic construction It seems that has treated it seems that as a hearsay marker. Chafe labelled it a hearsay particle, signalling that knowledge has been acquired through language. However, it seems that can also be used as a hedge on the probability of an assertion. A typical German translation is offenbar (perhaps best glossed as evidently or obviously or as you can see) (Aijmer, 2009). In Norwegian the correspondence åpenbar, which is found in my data as well, is similar in meaning and function.

Both Johansson and Aijmer have shown, and this thesis will corroborate these findings further, that seem in catenative constructions share many semantic and grammatical properties with the modal auxiliaries: The number of non-congruent correspondences in the form of modal particles, modal adjuncts and modal auxiliaries in the Norwegian correspondences of seem testify to the auxiliary-like meaning of the verb. They are used in the same position as the modal auxiliary, i.e. before a non- finite verb form. They also resemble the modal auxiliaries in that they are semantically independent of the subject (6.5.). There is no difference between Sam seemed to realize the importance of the problem and The importance of the problem seemed to be realized by Sam (Johansson, 2001: 232).

3.8. Focus of the present study The conclusion that can be drawn from the present chapter is that while seem has been widely studied, appear has not. And despite the comprehensive catalogue of studies of seem, I have not found a study of seem or appear that takes the all the different concepts of grammar, syntax, semantic meaning and pragmatic function and the interconnectivity between them into account. My analysis in the next chapters is an attempt to offer a comprehensive analysis of seem and appear and their correspondences. For example, both Johansson and Aijmer offer very interesting insight into the evidential and epistemic modal meaning of seem and show how meaning depends on syntax. I wish to expand that insight by including communicative function and pragmatic effect in my analysis. I also wish to analyse appear along the same dimensions that seem has been studied to determine similarities and differences between the two and between their correspondences. I have not been able to find such a study in the literature, and this study is an attempt at offering some new insight on this topic.

29

4. Method and Material The present study is an analysis of the meanings and functions of seem and appear. In order to clarify the differences between seem and appear, I will use a semantic contrastive analysis (Semantic Mirrors) to establish the different senses of these verbs. I will also look at possible interrelationship between the syntactic patterns, the semantic meanings and the pragmatic functions of the two verbs.

The material is corpus data from the English-Norwegian Parallel Corpus (ENPC), which will be used in two separate analysis: The first analysis is to determine different semantic meanings (sense partitions) of the two expression with the method Semantic Mirrors. The second is a corpus-based contrastive analysis of the two expressions and their correspondences. Both methods are based on the assumption that translations are texts that are intended to express the same meanings and have the same discourse functions, and as such can be exploited as sources of information about semantic and pragmatic meaning. In this chapter I present the basis for comparison in both methods and the units used in the analysis (4.1.). I give a presentation of Semantic Mirrors as a method (4.2.), and I present the method of comparing and contrasting seem and appear and their correspondences (4.3.).

4.1. Equivalence and correspondence Two key concepts in translation theory are equivalence and correspondence. Equivalence is a meaning relation between a text in a source language and its translation, that makes the target text as close to the source text as possible. Equivalence exists on many levels – formal equivalence has to do with grammar and syntax, semantic equivalence is concerned with words and their meanings, and functional and pragmatic equivalence is all about recreating the same function and effect in a given context. Non-equivalence can result in misunderstandings and communication breakdown. To avoid it, different strategies can be employed, for example translation by paraphrasing, by using a more general word, by using a less expressive word, translation by omission (Baker 2011: 15). Which strategy to use depends on the problem. Often, the translator must make a trade-off between strategies: change syntax to keep thematic structure or disregard semantic equivalence to create equivalent pragmatic effect, for example. Previous studies have shown that in general, translators have a tendency to explicate, i.e. make what is implicit in the source text explicit in the target text. Another common translation strategy is to reconstruct the syntax with cognitive verbs, which explicates the reasoning process and epistemic assessment, or modal adjuncts that explicate either the presence of visual evidence (for example øyensynlig) or the epistemic modal element (for example etter alt å dømme), see the full list of correspondences in Table 1 in 4.5. and the overview of non-congruent correspondences in 5.3.2. As this thesis will show, explication is a frequently used strategy in the translation of seem and appear, for example by reconstructing the syntax and making the experiencer the subject of the sentence, often making an implicit experiencer the explicit subject. Researchers have also found a general tendency of translations to be simplified and less marked than the source texts; elements that are ambiguous in the source texts are made unambiguous, and the grammar is more conventional in target texts than in source texts (Blum-Kulka, 1986, Malmkjær, 2011 in Grønbeck 2018). Paraphrasing to keep semantic meaning and pragmatic function is another frequently used strategy. Zero-correspondence, i.e. omission instead of translation, is also a common translation strategy (3.7.).

30

There can be equivalence on different levels: between larger chunks of texts or between lexical items.

Correspondences are corresponding, symmetrical units of the translation pairs. The fundamental idea for the analyses in this thesis is that correspondences incorporate syntactic, semantic and pragmatic properties. The translations from the ENPC represent language in use (parole, not langue), and as such the translator of the original will use any translation strategy at his disposal to convey an equivalent text – or as equivalent as possible – on the semantic and pragmatic level. As a consequence, translations can be exploited as sources of information about semantics and pragmatics, and we can compare and contrast the correspondences. The translation relation is the basis for the analysis – the tertium comparationis.

Correspondences are the basis for comparison in my thesis: the corresponding, symmetrical units in the English Norwegian Parallel Corpus, which have been identified manually. In the contrastive analysis a distinction is made between congruent correspondences, non-congruent correspondences and zero correspondences, which each incorporates different translation strategies (5.3.).

4.2. Semantic Mirrors Semantic Mirrors is a method for discovering lexical and functional meaning of words and expressions in one language by studying their correspondences in another language developed by Dyvik (1998, 2004). The method Semantic Mirrors is based on the assumption that translations can be exploited as sources of information about semantic and pragmatic meaning. Semantic and pragmatic meaning is derived from context, purpose and background knowledge, and is encoded in the sign inventories of languages – as such they exist in the expressions in the source language and in the translations. The translation relation between languages and sets of translation pairs is used as a starting point to uncover and describe meaning relations between lexemes, words forms, phrases, sentences and grammatical categories e.g. synonymity, homonymity and vagueness. Dyvik makes use of the translational properties and uses them "as an important window onto the semantic properties of signs, providing an empirical foundation for the way in which we describe them" (Dyvik, 1998: 55). By going back and forth between corresponding units in two or more languages and studying the sets of correspondences (the mirrors), we can discover if a lexeme has different semantic meanings, and find meaning relations between lexemes.

The theory and method are based on several assumptions (Dyvik, 2011), listed here:

• Signs in a language inventory represent semantic and pragmatic meaning. The translational relation is between languages in use (parole, not langue), • Translations represent correct meaning. There is no room for 'bad translations'. • Semantically related words tend to have strongly overlapping sets of translations. • Words with wide meanings tend to have a higher number of translations than words with narrow meaning. • Contrastive ambiguity, i.e. ambiguity between two unrelated senses of a word (for example the two senses of band: 'orchestra ' and 'piece of tape'), tend to be a historically accidental and idiosyncratic property. Therefore, we don't expect to find the same contrastive ambiguity in in other languages.

31

• Words with unrelated meanings will not share translations into another language, except in cases where the shared translations are contrastively ambiguous between the two unrelated meanings.

In the explanation of the method Semantic Mirrors, I will use the Norwegian word tak as an example, simplified for the purpose of explaining the method. The example is from Dyvik (1998).

Creating a semantic mirror is a four-step process, which starts with identifying all translations (correspondences) in a parallel corpus of a word (a lemma) in the original language (L1). This first set translations (L2) make up the first t-image. The first t-image of tak consists of these correspondences:

Figure 2. First t-image of Norwegian tak.

The next step is to translate the words in the first t-image into L1 again. These new sets of words and expressions make up the inverse t-image. In the inverse t-image it is possible to observe that many of the correspondences overlap, which is an early indication of different senses. However, Dyvik does not attempt to distinguish between senses on the basis of the inverse t-image. The inverse t-image of tak looks like this:

Figure 3. The inverse t-image of Norwegian tak.

32

The next step is to find the translations in the corpus of all the words and expressions in the inverse t-image and create a second t-image. The second t-image consists of the t-images of the lexemes in the inverse t-image plus any lexemes in the first t-image that do not re-occur when this last round of translation is done. The original word – the starting point – is excluded in this step. The second t- image will inevitably contain many words and expressions that has nothing to do with the meaning of the original word, because of the ambiguity of the correspondences of the original word (Figure 4).

Figure 4. Second t-image of Norwegian tak.

Dyvik points out that the interest of the second t-image primarily lies in the subset structure it imposes on the first t-image. The idea is to use the second t-image to sort and group the translations that make up the first t-image. The next step in the process is thus to disregard all the words in the second t-image that are not members of the first t-image. In the example with tak, the restricted second t-image consist of the following sets of translations:

33

Figure 5. Restricted second t-image of Norwegian tak.

The final step is to seek out clusters of meanings or groups of translations in the restricted second t- image by identifying how the words and expressions in the restricted second t-image overlap. The restricted second t-image of tak is thus a restructuring of the first t-image, based on overlap patterns in the second t-image (Figure 6). Overlap relations between words mean that they are related. The more overlaps, the stronger the relation. No overlaps between words or groups of words (except the original word) means that they are semantically unrelated and reflect contrastively different senses. The result is a partitioning of the first t-image into sense partitions, where each partition contains semantically related words.

Figure 6. Sense groups of Norwegian tak.

34

The method Semantic Mirrors clearly shows that Norwegian tak has three sense groups: one is cover, the second is grip and hold, the third sense group is ceiling and roof. The conclusion is that Norwegian tak is 3-ways ambiguous with respect to English (Dyvik, 1998: 65).

When the method is applied to seem and appear, three sense groups for seem and three for appear and their degree of overlap are established. A contrastive study of translation pairs is then made to consider any correlation between the syntactic environments, the semantic modal and evidential meanings and the pragmatic functions of the two verbs.

4.3. Corpus linguistics, contrastive analysis and translation correspondences My study is a corpus-based contrastive analysis. Corpus-based studies "typically use corpus data to explore a theory or a hypothesis, typically one established in the current literature, in order to validate it, refute it or refine it." (McEnery and Hardie, 2012: 6). The approach emphasizes that corpus linguistics is a method by itself. It follows from this definition, that the choice of corpus should match the research questions, and that one must take great care in choosing a corpus that is balanced and representative, that it is of a size that can provide sufficient recurrences to generate a valid data set. The pros and cons of the choice of ENPC for this study is discussed I 4.4., and a description of the chosen sample from the corpus and the motivation for the chosen data set is presented in 4.5.

Contrastive Analysis is the systematic comparison of two or more languages, usually with respect to one or more specific features or aspects of language, with the aim of describing their similarities and differences. (Johansson, 2007). Corpus-based contrastive studies, i.e. cross-linguistic research carried out on the basis of a parallel corpus, can reveal both what is general and what is language specific. It can give new insights into the languages compared and illuminate differences between source texts and translations, as well as give insight into the relationship between source and target text (Johansson 1999). The process of contrastive analysis is one of description for both languages, juxtaposition and bidirectional comparison (Hasselgaard, 2018).

For my contrastive analysis, the data has been extracted from the corpus and imported into FileMaker7, a software tool for organising and analysing data. A FileMaker was created for each of the two expressions that are the objects of this study. The file was organised so that each entry showed the sentence in the source language (English) in one window and its translation in Norwegian in another. The translation pairs were analysed with respect to different variables. The variables were chosen because they were presumed to be suited to highlight similarities and differences between the chosen tokens and their translations. For example, type of correspondence, e.g. the number of congruent correspondences, non-congruent correspondences and zero correspondences (4.3.1) is an indicator of the chosen translation strategy, and the variable could be used to compare and contrast the different translation strategies of seem and appear. The Norwegian correspondences could be compared and contrasted as a single variable with respect to seem and appear, or the two verbs, seem and appear, could be analysed with respect to several variables, e.g. type of evidence or pragmatic function. Another variable was the distribution in sense groups determined in the Semantic Mirrors-analysis. By logging how seem and appear placed themselves in the different sense groups, I could calculate the relative frequency of overlapping and diverging semantic meaning for

35 seem and appear and how that was reflected in the translations. Evidentiality and modality was determined by the variable type of evidence (4.3.2, see also 2.7.). The final variable was the pragmatic function (4.5.3., see also chapter 3.5.). The examples in the following sections are from the translation pairs included in the data set for this thesis. When possible, I have included examples with both seem and appear. The relevant parts are in bold.

4.3.1. Translation correspondences and their syntactic form Correspondences in the form of a verb phrase (lexical verbs) with the same overall syntactic and thematic structure as the original have been classified as congruent correspondences (48) and (49).

(48) She seemed reluctant to admit that much. (SG1) Hun virket uvillig til å innrømme så pass. (SG1T) (She seemed unwilling to admit that much.) (49) She appeared to be wearing parts of a dead goat on her feet. (ST1) Hun så ut til å ha tullet føttene inn i deler av en død geit. (ST1T) (She appeared/looked as if to have wrapped her feet in parts of a dead goat.)

Translations that changed the overall structure of the original sentence, but kept the semantic or modal meaning, were categorized as non-congruent correspondences. Since this thesis is not primarily concerned with form, but with modal and evidential meaning conveyed by the chosen tokens in the originals, and their translations, some restructured sentences are categorized as non- congruent correspondences because the modality/evidentiality is retained that would probably have been zero correspondences in studies more focused on form. Non-congruent correspondences found in my material are:

Restructurings that make an implicit or explicit experiencer the subject

(50) It seemed to me that whatever else was true, it was absolutely the case that Ty deserved to realize some of his wishes. (JSM1) Enten det nå var sånn eller slik, syntes jeg så absolutt at Ty fortjente å få noen av ønskene sine oppfylt. (JSM1T) (Whether it was this way or that way, I absolutely thought that Ty deserved to fulfil some of his wishes.)

Restructurings with Norwegian anticipatory or dummy subject det

(51) Mama's stories about Elek were so vivid he almost seemed to be my brother. (TH1) Mammas historier om ham var så levende at det nesten kjentes som om han var min bror. (TH1T) (Mama's stories about him were so vivid that it almost felt as if he were my brother.) (52) I had always thought that dogs were engineered on the principle of four-wheel drive vehicles, with equal propulsion coming from each leg, but the power appears to be concentrated in the back. (PM1)

36

Jeg hadde alltid trodd at hunder var konstruert etter samme prinsipp som firehjuls- drevne kjøretøyer — at hvert ben bidro like mye til fremdriften, men det later til at de er bakbensdrevne. (PM1T) (…., but it appears that they have rear wheel drive.)

Translations with a modal particle

(53) I could n't think of a response and she did n't seem to expect one. (SG1) Jeg klarte ikke å tenke ut noe svar, og hun ventet det visst heller ikke. (SG1T) (I could not think of a response, and she did not expect it [modal particle] either.) (54) Arthur seemed to be climbing down. (RDA1) Arthur var nok på vikende front. (RDA1T) (Arthur was [modal particle] on the receding front.

Translations with a modal adjunct or a prepositional phrase functioning as a modal adjunct

(55) Philip had recently fathered a second son, with his new wife, Cleopatra, with whom he appeared to be giddily infatuated. (JH1) Filip hadde nettopp avlet en sønn til, med sin nye hustru, Kleopatra, som han åpenbart var svimlende forgapt i. (JH1T) (Philip had recently fathered another son, with his new wife Cleopatra, with whom he was obviously giddily infatuated.)

(56) They found that for them both the meaning of life seemed to be contained, if mysteriously, in living useful lives. (NG1) De oppdaget også at for dem begge lå hensikten ved livet i på en eller annen måte å gjøre en innsats. (NG1T) (They also discovered that for them both the purpose of life was in some way or another to make an effort.)

Translations with a modal auxiliary

(57) "I do n't seem to remember your ever giving Molly much of a chance," said Frederick. (DL1) "Jeg kan ikke huske at du noensinne har gitt Molly noen særlig sjanse," sa Frederick. (DL1T) ("I cannot remember that you have ever given Molly any particular chance, said Frederick.")

37

Comparative constructions / hypotheticals

(58) He had clean, good-natured lines in his movements which seemed to say to everyone I 'm here and ai n't complaining about it, so why are you? (GN1) Bevegelsene hans var rene og blide, det var som om de sa til alle og enhver — her er jeg og klager ikke, så hvorfor gjør du? (GN1T) (His movements were pure and friendly, it was as if they told everyone – here I am, and [I am] not complaining, so why are you?)

The non-congruent correspondences are indicators of how modal or evidential seem and appear are. For example, a high number of correspondences with modal meaning suggest that the original expression is modal, rather than evidential in nature.

Omissions, as this thesis will show, is a frequently used translation strategy for seem and appear in some contexts. In my thesis, zero correspondences are omissions, see (59) below, or significant restructuring of clauses with a clear loss of evidential and modal meaning (60):

(59) He seemed to stiffen. (RD1) Han stivnet til. (RD1T)

(He stiffened)

(60) Andrew was ascending the hospital front steps when, ahead of him, the main door slammed open and Dr. Overton, the resident, appeared to hurl himself out. (AH1) Andrew gikk opp trappen da hoveddøren gikk opp og dr. Overton kom settende, meget opphisset, med bustet hår. (AH1T) (…, the main door opened, and Dr. Overton came hurling out, very agitated, with unkept hair.) The number of zero correspondences may thus be an indicator of how modal or evidential the expression is. For example, a high number of zero correspondences will testify to a low degree of modality in the original expression.

4.3.2. Evidential and epistemic modal meaning Seem and appear are also categorized with respect to various sense groups as given by the Semantic Mirrors-analysis (5.1.5. and 5.2.5.). This makes it possible to provide a quantitative measure of the extent to which they share semantic meaning, and how often they are semantically distinct.

The other variable that provides insight into how modal or evidential seem and appear are, is 'Type and source of evidence'. Each sentence pair is analysed with respect to this variable in both languages.

The values for this variable are: • Direct Attested Visual • Direct Attested Auditory • Direct Attested Other Sensory • Indirect Reported Hearsay (third and second hand) • Indirect Reported Folklore

38

• Indirect Inferring Result • Indirect Inferring Reasoning • No evidence/Personal judgement and belief • Irretrievable from context.

The frequencies of direct attested evidence and indirect evidence were used as an indicator of how evidential the original expression was. When seem and appear refer to direct and indirect evidence, they are interpreted to have evidential meaning. When the evidence is inferred, they are epistemic and evidential at the same time. With respect to the two types of inferential evidence, the categorization was a qualitative judgement of whether the epistemic or the evidential component was the strongest. If the inference was based on some observable result, the evidence was categorized as 'Indirect Inferring Result', if the reasoning process was the main source for the inference, the chosen evidence type was 'Indirect Inferring Reasoning'.

If there was no reference to evidence, seem and appear were interpreted to have epistemic modal meaning. (A subsequent manual analysis of the last category showed that there was only one translation pair where appear had no modal or evidential meaning, this example was excluded from any discussions about evidentiality and epistemic modality. See 5.2.3.). This interpretation of the variable 'Type of evidence' as an indicator of modality or evidentiality is in line with the chosen theoretical approach to evidentiality and epistemic modality, which sees them as separate categories, but with overlapping meaning and function in inferences (see section 2.6.4).

4.5.3. Pragmatic functions Lastly, seem and appear and their Norwegian correspondences were categorized with respect to their pragmatic functions (3.5.). The pragmatic functions were:

• Hedging • Stance • Politeness marker • Undetermined

I subsequently did a manual count to determine if there were any difference in the functions of seem and appear as adaptors, modifiers, boosted modifiers, adjusters or construers of common ground, and I counted manually how the different functions co-occurred with different types of evidence. I also did an analysis of the different correspondences with respect to the different functions. The contrastive analysis offers insight into the functions of the two expressions and show similarities and differences in pragmatic function and effect.

To properly determine the relationship between form, meaning and function, I also counted manually how catenative constructions and constructions with a that-clause correlated with the source and type of evidence to determine whether they were primarily evidential, epistemic or both. I also counted manually the frequencies with which seem and appear in different syntactic constructions had different pragmatic functions. The result of this three-way analysis is presented in 5.5.

39

4.4. About the chosen corpus Both analyses in the thesis uses data from the English-Norwegian Parallel Corpus (ENPC), which is a sub-corpus of the Oslo Multilingual Corpus at University of Oslo. The ENPC consists of text extracts from 30 fictional texts in each language and their translations, and 20 original, non-fictional texts and their translations. Each text consists of 10,000 – 15,000 words. The sentences are aligned at sentence level. The alignment at word level for this thesis is done manually, which means that corresponding, symmetrical units were identified manually.

The ENPC is designed so that it can be used both as a comparable corpus (one can compare and contrast linguistic phenomena in non-fictional texts in English and Norwegian, for example) and a parallel corpus, since it contains original texts in both languages and can be combined both ways. For the Semantic Mirrors-method, which depends on the translation relation between sentence pairs and lexical and syntactic units, and go back and forth between source language and target language, a parallel corpus is the obvious choice.

Furthermore, the ENPC ticks all the boxes of what to look for in a corpus to perform a contrastive analysis of a lexical phenomenon such as seem and appear. It is a well-balanced corpus even though fictional texts are dominant. Moreover, the language is representative for written English and Norwegian in the genres that are sampled in the corpus. The corpus is small, but with almost 600 hits for seem in English originals (raw frequency), the corpus generates sufficient recurrences of most syntactic constructions with seem. The number of hits for appear are fewer, with a raw frequency of about a quarter of that number, but I will nevertheless argue that there are enough translation pairs with appear to make a valid analysis of the token and its correspondences (see 4.4.1).

4.5. The data set The data set for this study consists of a total of 312 translation pairs from the fictional and non- fictional part of the English Norwegian Parallel Corpus. There are 269 translation pairs with seem and 43 translation pairs with appear. The selection of translation pairs was done in the following way: A search for seem/seems/seemed/seeming returns a total of 589 results; 393 hits in the fictional part of the ENPC and 196 results in the non-fictional part of the ENPC. All the possible syntactic constructions with seem (see section 2.2.) are found in the ENPC. The Norwegian correspondences include verb phrases, adverbials, modal particles and significant structural changes of the clause, and a number of translations with zero correspondences. A search for all verb forms of appear (search string: appear/appears/appeared/appearing) returns a total of 154 results, 70 results in the fictional part of the ENPC and 84 hits in the non-fictional part. At first glance, appear occurs in grammatical structures and sentences very similar to seem, and the correspondences are for the most part very similar to those of seem. To achieve a manageable scope, I restricted the analysis to instances where seem and appear are followed by an infinitive clause or a that-clause. A search in the English originals in the ENPC for seem/seems/seemed/seeming AND +1 to produces 182 hits in the fictional part of the ENPC and 91 hits in the non-fictional part. Hits for seem/seems/seemed/seeming + to + a noun or a pronoun (the experiencer-construction) were excluded, unless they were followed by an infinitive clause later on in the sentence, see example (62). 41 translation pairs were excluded. A search for seem/seems/seemed/seeming followed by a that-clause results in 37 relevant hits. All in all, a total of 269 translation pairs with seem were included in the Semantic Mirrors-analysis and the contrastive analysis. For appear, I manually edited the list generated by the search string for

40 appear/appears/appeared/appearing. The washed list, which includes appear plus an infinitive or a that-clause, consists of a total of 43 translation pairs in the fictional and non-fictional part of the ENPC.

This is an example of a translation pair that was excluded: (61) Far better, it seemed to me, was the simple strategy of waiting for nature to come to you, rather than going clumsily to look for it. (DM1) Jeg skjønte snart at det var en mye enklere strategi å vente på at naturen skulle komme til deg, fremfor å klampe rundt på jakt etter den. (DM1T)

This is an example of a translation pair that was kept in the material: (62) They seem to me to rely too much on forms of life we already know. (CSA1) Samtlige synes å legge for stor vekt på former for liv som vi allerede kjenner. (CSA1T)

The full list of the Norwegian correspondences of seem and appear in combination with a to- infinitive clause or a that-clause and the frequency with which they occur in the extracted data, is presented in Table 1 (next page).

Note that the total number of correspondences does not match the number of translation pairs. The reason is that some correspondences combine several of the items on this list in one correspondence, for example a modal auxiliary and a lexical verb (63).

(63) If they appear to be two-dimensional they suddenly seem less "animal-like" and may avoid the unwelcome attentions of a hungry predator. (DM1) Dersom de kan fremstå som todimensjonale, virker de straks mindre "dyreliknende" og kan unngå uønsket oppmerksomhet fra sultne rovdyr. (DM1T)

Zero correspondences and significant restructuring of the clause without the use of any of the lexical items above are obviously also not on in this overview of correspondences.

41

Table 1. The Norwegian correspondences of seem and appear with a to-infinitive clause or a that-clause.

seem appear

Form Correspondence Frequency Form Correspondence Frequency

Verb arte seg som 1 Verb phrase phrase fortone seg som 3 fremgå 1 forekomme 1 fremstå som 2 føles som 1 gi inntrykk av 1 gi inntrykk av 2 gi uttrykk for 1 kjennes som (om) 1 late til (å/at) 34 late til (å/at) 9 mene 2 minne om 3 se ut som (om)/til 48 se ut som/til 8 synes (å) 36 synes (å) 6 tyde på 1 virke (som/som om) 40 virke (som) 12 være som om 7 være nesten som om 2 være tydelig at 1 være en illustrasjon på 1

Modal Kan/kunne 4 Modal Kan/kunne 2 auxiliary skulle 2 auxiliary

Adverbial etter alt å dømme 4 Adverbial tilsynelatende 2 på en eller annen måte 3 øyensynlig 1 åpenbart 2 åpenbart 1 tydeligvis 2 så vidt 1 forresten 1 særlig 4 sannsynlig 2 opplagt 1

Modal visst 11 Modal 0 particle liksom 6 particle nok 2 vel 1 jo 1

42

Moreover, It is important to point out at this point that the Norwegian correspondence synes (a cognitive verb, best glossed as 'think') occurs in two different Norwegian constructions with two different meanings: In an active sentence, in which the subject is the agent, it is a cognitive verb with epistemic meaning, best glossed as I think, my opinion is. In a passive construction, with no agent, the sentence would be det synes (meg) å være, best glossed as It looks like/appears/seems (to me) to be. In this construction the Norwegian correspondence has evidential meaning. In all six translation pairs where synes is a correspondence for appear (see Table 1 above), synes is used in a passive construction (64):

(64) Many of these relatives still lived in Germany and had the Old World's contempt for the New, a contempt in which Louise appeared to collude. (RF1) Mange av disse slektningene bodde fremdeles i Tyskland og hadde Den Gamle Verdens forakt for Den Nye, en forakt Louise syntes å dele. (RF1T) (…a contempt Louise appeared to share.)

As a correspondence for seem, synes occurs in a passive construction (65) in half of the translation pairs with synes (18 pairs) and as an active verb (66) in the other half.

(65) But none of my questioners seems to have any teeth. (FW1) Men ingen av mine utspørrere synes å ha noen tenner. (FW1T) (But none of my questioners seems to have any teeth.)

(66) The smell was almost shut out, but to Alice it seemed that an invisible film of stench clung to everything, and she would feel it slippery on her fingers if she touched. (DL2) Stanken kjente de nesten ikke, men Alice syntes en usynlig hinne av vond lukt lå over alt, som om fingrene hennes ville bli klissete hvis hun rørte ved noe. (DL2T) (They barely noticed the stench, but Alice thought [that] an invisible film of bad smell clung to everything…)

4.7.1. Some further restrictions of the data set for the Semantic Mirrors-analysis The Semantic Mirror-method involves creating semantic mirrors from a great number of lexemes in both languages. It starts with one lexeme in the original language (English in this case), but the number of semantic mirrors increases quickly when all the correspondences in Norwegian is translated back again into English. To keep the number of translations and backtranslations at a manageable level, I limited the correspondences to verbs, so that only verbs were included in the first t-image, and in the inverse t-image. Adverbials, modal particles and restructured sentences, were excluded. The choice of restricting the Semantic Mirrors-analysis to only include verbs and to exclude modal adjuncts like øyensynlig (possibly evidential) and prepositional phrases like etter mitt syn (epistemic) in the early stages of the process of the Semantic Mirrors analysis, might have skewed the results. In the second t-image, I kept all the lexemes that came out of the final round of translating.

Also, lexemes in the inverse t-image that were clearly from a semantic field wholly unrelated the original expression were not translated from the inverse t-image into the second t-image (e.g. work, have effect, promote and shape, which are translations of virke and virke som that are unrelated to

43 the modal and evidential meaning of seem). For expressions consisting of two words or more, e.g. the prepositional verb late til or expressions with a like in virke som, the search strings included til or som in up to 5 words after the main verb to allow for expressions with adverbials between the verb and preposition, like in this example

(67) I virkeligheten later det motsatte til å ha vært tilfelle (DM1T)

When the number of hits in the corpus exceeded 200 (i.e. the search string for the verb see in different forms returned more than 1300 hits), I included only the first 100 results from the fictional part of the corpus and the first 100 results from the non-fictional part in the analysis.

I considered extending the material to include results from seem and appear plus an adjective (a verbless clause), in order to expand the data set for appear. However, by scanning the unedited list for appear in various forms in the ENPC, I find only 15 translation pairs with appear plus an adjective. None of them produce new correspondences, but share the same correspondences as appear plus a to-infinitive clause and appear plus a that-clause. I therefore conclude that although the corpus is small, and the data relatively sparse, it is sufficient to provide a valid analysis of the chosen expressions and their correspondences.

44

5. Analysis. In this chapter I present the semantic mirror analysis of the two verbs, and the sense partitions it engenders (5.1. and 5.2.). The analysis is summed up with a comparison of the sense partitions for the two in view of their degree of semantic overlap. In 5.3. I present the results of the contrastive analysis. The analysis offers insight into the similarities and differences between seem and appear beyond the semantic meaning and different sense partitions discovered in the Semantic Mirror- analysis: Similarities and differences with respect to their syntactic form and that of their correspondences, semantic meaning and pragmatic function (5.5). Finally, I consider any correlation between the syntactic environments, the semantic modal/evidential meanings and the pragmatic functions of the two verbs.

5.1. The senses of seem I apply the method of Semantic Mirrors to seem and appear, with the purpose of describing different semantic meanings of the two tokens, and determine what the different sense partitions, or sense groups are for each of them. Seem is a vague lexeme that covers several meanings, depending on context. Vague lexemes will typically generate a rapid increase in the number of lexemes that is the result of every round of translation. For seem, the number goes from 19 lexemes in the first t-image, 57 in the inverse t-image and 364 in the second t-image. The Venn-diagram is therefore not suited to visualize the union and overlap relations between sets. Instead, the members of a union are shown in brackets, like this: {member1, member 2, member 3}.

5.1.1. First t-images of seem Since only verb phrases are included in the first steps of Semantic Mirrors (see 4.4.1), the first t- image of seem is the set of verb phrases from table 1, listed here in alphabetical order:

{arte seg som, fortone seg som, forekomme, føles som, gi inntrykk av, gi uttrykk for, kjennes som, kan/kunne, late til, mene, minne om, se ut til, se ut som, skulle, synes, tyde på, virke/virke som, være som om, være nesten som om, være tydelig at, være en illustrasjon på}*

* mene and skulle are not part of the analysis in the next chapters.

5.1.2. Inverse t-image of seem The inverse t-image of seem consists of the translations of all the lexemes in the first t-image. Technically, the inverse t-image is a union of all the t-images of each of the lexemes in the first t- image, listed here in alphabetical order:

{act, appear, be/be like, be able to, be capable, be familiar with, be likely, be possible, be reminiscent of, be visible, bear signs of, become capable, believe, can, claim, consider, could, express/give expression, fancy, feel/feel like/feel as though, find, function, give hint of, give inclination of, give impression/produce impression/convey impression/get impression, have effect, have in common, indicate, imply, know, look, may, might, occur, operate, pretend, probably, promote, recall, resemble, remind, result, reveal, show up, see, seem, shape, sound like, strike as, suggest, tend to, think, work, will, would}

45

5.1.3. Second t-image of seem The second t-image is created by translating the lexemes in the inverse t-image once more into Norwegian. The second t-image consists of the t-images of the lexemes in the inverse t-image plus any lexemes in the first t-image that does not reoccur when this last round of translation is done. The original word, seem, is not included in this step.

Only the words in bold in the inverse t-image above were translated (see section 4.5. for a description of the data set included in the Semantic Mirrors-analysis.) The sets of correspondences that make up the second t-image are thus:

{{arbeide,delta, fungere som, gjøre tjeneste som, gå frem, handle, leve ut, oppføre seg, opptre, te seg, tjene som, treffe beslutning, virke som} {fremstå som, gi inntrykk av, late til, se ut til , se ut som, synes, synes å, tilsynelatende, virke, virke som, åpenbart} {kjennes som, ligne, være lik, sammenligne, være som, være som om} {hende, kan, kan tenkes, kanskje, kunne, la seg gjøre, sannsynligvis, være mulig} {anse som sannsynlig, fortone seg, forventes, sannsynligvis, som oftest, være nærliggende, være sannsynlig, være sikkert, være tenkelig, være trolig, være (lite) som taler for} {minne om} {komme til syne, se, skjelne, tre frem, vise, være synlig} { } {anse, begripe, få inn i hodet, håpe, mene, se, tenke seg, tro, tro på , ha tiltro til, virke, være sikker, stole på, ta for, overbevise} {greie, få, la seg, kan, klare, kunne, være mulig} {erklære, forlange, gi seg ut som, gjøre krav på, gå ut på, hevde, insistere, kreve, legge beslag på, , påstå, si, synes, søke} {anse, betrakte, drøfte, forstå som, føle seg, grunne over, lure på, mene, oppfatte, regne som, se på, synes, tenke, tro, reflektere over, ta hensyn til} {greie, ha vanskelig for, kan, klare, kunne, vil, ville, være mulig} {formidle, gi uttrykk for, komme til utrykk, omskrive, uttrykke, synes, si, uttale} {drømme om, forestille seg, like, vil, ville} {ha lyst til, fortone seg, føle, føle seg, ha inntrykk av, kjenne, mene, merke, oppleve, synes, tro} {betrakte, finne, få, få høre, få tak i, få vite, gi, gripe seg i, oppfatte, oppleve, se, sjekke, synes, tro, undersøke, virke, være, være av den mening} {angi, anføre, antyde, bety, gi tegn til, gjøre tegn, indikere, markere, nikke mot, peke på, referere til, røpe, tyde på, vise, være tegn på} {antyde, bety, forutsette, innebære, kreve, uttrykke, være, være et tegn på} {gi inntrykk av, gjøre inntrykk, minne om, virke} {late til, ligne, minne om, se ut som, se ut til, tyde på} {behøve, få, kan, kan hende, kan tenkes, kan være, kanskje, muligens, må, måtte, skal, sikkert, tør, være mulig} {faktisk, kan, kan hende, kanskje, kunne, kunne godt, minne om, muligens, måtte, se ut som, skal, skulle, ville, være mulig} {bli, falle inn, foregå, forekomme, foresveve en, inntreffe, komme, komme på, late til, melde seg, oppstå, skje, slå en, stå ved, tenke på} {drive, fungere som, operere, sette ut i livet, virke, være basert på} {drømme, gi inntrykk av, innbille seg, late som, leke at, prøve, utgi seg for} {erindre, gjenkalle, huske, komme til å tenke på, minne om, minnes, rykke tilbake, ta tilbake, tenke,

46 tillate} {ha likhet med, ligne, ligne på, minne om} {assosiere, gjøre oppmerksom på, forsterke inntrykket av, minne om, si, slå en at, tenke, være en påminnelse} {blottlegge, gi seg utslag, dukke opp, komme, komme til syne, stille opp, virke som} {betrakte, besøke, forstå, finne grunn til, forestille seg, få glimt av, få øye på, kjenne til, konstatere, legge merke til, mene, oppleve, oppdage, ordne, registrere, synes, treffe, være sikker på, være vitne til, øyne} {forekomme meg, høres som, høres ut som, ligne, lyde, lyde som, se ut som, tyde på} {anbefale, anslå, anse, antyde, bety, dukke opp, foreslå, fremsette hypotese, formode, gi vink om, indikere, mene, synes, lede tanken hen på, tyde på, tilsi, snakke om, stå (i en bok)} {tendens til, ha tilbøyelighet til, ofte, pleie, synes, som oftest, som regel, stort sett, vanligvis} {betrakte, gruble, huske, lure på, mene, synes, tenke, tro, visst} {begynne, bli, kan, komme til, skal, vil, ville} {akte å, bli, pleie, kan, kunne, skulle, vil, ville}}

The empty subset happens because there are no English originals that have used the expression bear signs of, therefore there are no correspondences to this term in Norwegian. The term was only used once as a translation for se ut som in my material (it is used four times as a translation in the corpus in total).

5.1.4 Restricted t-image of seem The interest in the second t-image lies primarily in "the subset structure it imposes on the first t- image" (Dyvik, 1998, 65). The idea is to use the second t-image to sort and group the original translations of seem. The restricted second t-image, is created by sorting the sets above the following way: If a set includes both words and expressions from the original t-image and words and expression not found in the first t-image, the set is kept, but reduced so that it only includes words and expression that also occur in the first t-image. If a set consists entirely of lexemes not found in the first t-image, the set is discarded. The restricted second t-image is thus:

{{arte seg som} {fortone seg, føle, ha inntrykk av, synes} {fortone seg} {forekomme, late til} {forekomme, tyde på} {gi inntrykk av, late til, se ut til, se ut som, synes, synes å, virke, virke som} {gi inntrykk av, gjøre inntrykk, minne om, virke} {gi uttrykk for, komme til utrykk, utrykke, synes} {gi inntrykk av} {kjennes som, være som, være som om} {kunne} {kunne, minne om, se ut som} {late til, se ut som, se ut til, minne om, tyde på} {late til} {minne om}} {se ut som} {synes} {synes, føle}

47

{synes, virke} {tyde på} {virke, virke som} {uttrykke}}

5.1.5 Sense partitions of seem The restricted t-image can be partitioned into groups of sets. Each group consists of sets with words and expressions that overlap. When there is no overlap between groups, each group represents a distinct sense of the original lexeme. For seem, there are three distinct groups:

Group 1S {arte seg som}

Group 2S {{forekomme, late til} {forekomme, tyde på} {gi inntrykk av, late til, se ut som, synes, synes å, virke, virke som} {gi inntrykk av, gjøre inntrykk, minne om, virke} {gi uttrykk for, komme til utrykk, utrykke, synes} {gi inntrykk av} {fortone seg, føle, ha inntrykk av, synes} {fortone seg} {late til, se ut som, se ut til, minne om, tyde på} {late til} {kunne, minne om, se ut som} {kunne} {minne om} {se ut som} {synes} {synes, føle} {synes, virke} {tyde på} {virke/virke som} {uttrykke}}

Group 3S

{kjennes som, være som, være som om}

The method Semantic Mirrors shows that seem has three distinct senses: The first, arte seg som, is a special case with the semantic meaning of acting or pretending (Group 1S). The second sense partition represents a broad semantic meaning indicating that something superficially looks like something else and gives the same impression (Group 2S). The third sense partition is that of a more deeply rooted similarity: something does not merely look like something else, it is like something or feels like something (Group 3S). In terms of evidentiality and epistemic modality, the sense of seem in Group 2S may be said to be more evidence based, and thus have a higher degree evidentiality, given that the semantic sense has to do with visible similarities. The third sense partition of seem has

48 a higher degree of personal judgement as the basis for its modal meaning, and the sense of similarity is encoded in the semantic meaning of the word.

5.1.6. Translation hierarchy and prototypical translations of seem By ranking the different lexemes within each sense group based on the number of subsets in the restricted t-image they occur in, it is possible to provide a hierarchy of the translations of seem: The words and expressions that occur in the most subsets represent the most 'general 'or 'prototypical' meaning (Dyvik, 1998, p. 73).

Synes, which occurs in six subsets, is the most general translation of seem in the first sense partition (Group 1S). Next is se ut som, ha inntrykk av/gi intrykk av and virke/virke som, which occur in four subsets. The expression tyde på is present in three subsets, and the others, such as late til, minne om and føle, occur in two subsets. For the second sense partition (Group 2S), both være som and kjennes som occur in only on subset. Arte seg som (Group 3S) also occurs only in the one subset. It is at best a marginal sense partition: It occurs only once as a correspondence for seem in the ENPC and is only part of the second t-image because it is transferred from the first image, in line with Dyvik's method to ensure rare translations are not overlooked. However, it can be argued whether this is really a distinct sense partition. The expression is semantically very close to the sense partition of Group 2S: that something gives the impression of being something or having a quality. On the other hand, the expression arte seg does indicate agency of some sort, e.g. that someone is actively putting on a show of some appearance or is pretending to be something.

5.2. The senses of appear The process of creating semantic mirrors for appear involves fewer lexemes in the first t-image, the inverse t-image and fewer subsets in the second t-image than seem, and many lexemes and subsets are the same for appear and seem.

5.2.1. The first t-image of appear The first t-image of appear consists of the following correspondences:

{gi inntrykk av, fremgå, fremstå som, kunne, late til, se ut til, se ut som, synes, virke (som)}

It is worth pointing out that all but two of the correspondences of appear are also correspondences of seem. The exceptions are fremstå som and fremgå. It follows that the inverse t-image and the second t-image for appear have many of the same words and expressions as the inverse and second t-image for seem, particularly since both fremstå som and fremgå are rare expressions in the ENPC. There is, for example, only one hit in the entire fictional part of ENPC for fremgå, and five hits the non-fictional part. A search for gå plus frem returns five hits in the fictional part of the corpus, but only one with the same sense as fremgå. Interestingly for a paper on evidentiality and modality, the translation of this one is to be evident. Fremstå is also rare, with only five hits in the Norwegian originals in the non-fictional section, none in the fictional part.

49

5.2.2. Inverse t-image appear The inverse t-image of appear looks like this (lexemes in bold are translated into a second t-image in the next step):

{appear, be, be born, be evident, be possible, be visible, be able to, be capable, be familiar with, bear signs of, believe, can, could, consider, convey impression/give impression/produce impression, emerge, fancy, find, form, look, may, might, pretend, know, tend to, see, seem, stipulate, strike as, show, think, will, would.

5.2.3. Second t-image of appear The second t-image of appear consists of the union of the sets of translations of the lexemes in the inverse t-image plus any lexemes from the first t-image that does not occur in the final translation:

{{kjennes som, ligne, være lik, sammenligne, være som, være som om} {bli født/være født/fødes, komme fra, komme av} {være åpenbart, være klart} {hende, kan, kan tenkes, kanskje, kunne, la seg gjøre, sannsynligvis, være mulig} {komme til syne, se, skjelne, tre frem, vise, være synlig} { } {anse, begripe, få inn i hodet, håpe, mene, se, tenke seg, tro, tro på , ha tiltro til, virke, være sikker, stole på, ta for, overbevise} {greie, få, la seg, kan, klare, kunne, være mulig} {greie, ha vanskelig, kan, klare, kunne, vil, ville, være mulig} {anse, betrakte, drøfte, forstå som, føle seg, grunne over, lure på, mene, oppfatte, regne som, se på, synes, tenke, tro, reflektere over, ta hensyn til} {gi inntrykk av, gjøre inntrykk, minne om, virke} {avtegne, bevege seg, dukke opp, dukke frem, forme seg, gå ut fra, komme frem, komme til overflaten, komme ut, komme for en dag, komme inn, oppstå, opptre, spre seg, stige frem, stå frem, tre frem, utvikle seg, vise seg} {drømme om, forestille seg, like, vil, ville} {betrakte, finne, få, få høre, få tak i, få vite, gi, gripe seg i, oppfatte, oppleve, se, sjekke, synes, tro, undersøke, virke, være, være av den mening} {bli, bli til, danne, forme, lage, stille, skape, stå som, vokse frem, være, utgjøre utvikle} {gi inntrykk av, gjøre inntrykk, minne om, virke} {behøve, få, kan, kan hende, kan tenkes, kan være, kanskje, muligens, må, måtte, skal, sikkert, tør, være mulig} {faktisk, kan, kan hende, kanskje, kunne, kunne godt, minne om, muligens, måtte, se ut som, skal, skulle, ville, være mulig} {drømme, gi inntrykk av, innbille seg, late som, leke at, prøve, utgi seg for} {tendens til, ha tilbøyelighet til, ofte, pleie, synes, som oftest, som regel, stort sett, vanligvis} {betrakte, besøke, forstå, finne grunn til, forestille seg, få glimt av, få øye på, kjenne til, konstatere, legge merke til, mene, oppleve, oppdage, ordne, registrere, se, synes, være sikker på, være vitne til, øyne} {arte seg som, fortone seg som, føles som, gi inntrykk av, gi uttrykk for, kjennes som (om), kunne, late til (å/at), minne om, se ut til (å/at), se ut som (om), synes (å), være som om, virke (som/som om), være nesten som om, være tydelig at, være en illustrasjon på} {anse, falle inn, slå en, synes} {blottlegge, dukke opp, forestille, få, gi utslag, gå frem, forestille, fremheve, ha, holde frem, komme

50 til syne, oppvise, påvise, røpe, se ut som, skinne igjennom, skilte med, stå, synes på noe(n), være, være fremtredende, være synlig} {betrakte, gruble, huske, lure på, mene, synes, tenke, tro, visst} {begynne, bli, kan, komme til, skal, vil, ville} {akte å, bli, pleie, kan, kunne, skulle, vil, ville} {fremstå som} {fremgå}}

Note that the singleton sets fremstå som and fremgå from the first t-image are included in the second t-image. Singleton sets from the first t-image that has not reoccurred as translations of the inverse t-image in the second t-image are included to ensure that special cases, e.g. technical terminology, are not lost in in the process of translating and back-translating the correspondences (Dyvik, 1998, 59).

5.2.4. Restricted t-image of appear After eliminating words in the second t-image that are not members of the first t-image, the restricted second t-image of appear consists of the following sets of translations:

{{kunne} {synes} {gi inntrykk av} {synes, virke/virke som} {virke/virke som} {kunne, se ut som} {se ut som} {se ut som, synes} {gi inntrykk av, kunne, late til, se ut som, synes, virke/virke som} {fremstå som} {femgå}}

5.2.5. Sense partitions of appear Except for fremgå and fremstå, both of which are special cases, all the subsets in the restricted t- image are part of the largest subset {gi inntrykk av, kunne, late til, se ut som, synes, virke/virke som}. We can therefore conclude that appear has three sense groups:

Group 1A {gi inntrykk av, kunne, late til, se ut som, synes, virke/virke som}

Group 2A {fremstå som}

Group 3A {fremgå}

The first and largest sense group of appear (Group 1A) has overlapping semantic meaning with the second sense group of seem (Group 2S), i.e. that something superficially looks like something else and gives the same impression. The second sense partition – fremstå som in Group 2A – is similarly

51 concerned with how something looks or is perceived. Fremstå som communicates that even though the speaker may commit to how things are presented or present themselves, the speaker will not fully commit to saying that that is how things are. Thus, the correspondence fremstå som arguably has much the same meaning as the expressions in Group 1A, but is an even stronger hedge. In Bokmålsordboka, fremstå is used in two examples, both of which are associated with visual evidence: 'to look and behave as a new person' (fremstå i ny skikkelse) or 'to look like the winner' (fremstå som seiersherre). The latter example could possibly be glossed as 'emerge as victorious', which would entail not only visual evidence, but some other evidence, perhaps a body count in medieval times, but I will argue that the expression in Norwegian is strongly associated with how someone carries themselves, i.e. how they look or are perceived.

The last sense partition of appear, fremgå in Group 3A, differs significantly from the other two. Fremgå signals that an inference has been made on the basis of some observable evidence. In Bokmålsordboka fremgå is explained with the verb vise seg ('turn out to be' or 'show') or by splitting the compound fremgå into two separate verbs; gå frem. The examples in Bokmålsordboka emphasizes the imbued inferential meaning: Det fremgår ikke klart av teksten ('It does not come clearly across from this text') and Det fremgår av denne statistikken at… ('It can be deduced from these statistics that…'). Fremgå has both evidential and epistemic in meaning.

In terms of epistemic modality and evidentiality, all three sense partitions of appear are strongly connected to evidence. The first two sense groups are concerned with how things look or present themselves and have evidential meaning. The last sense group also has evidential meaning encoded into its semantic meaning. Appear thus has very little epistemic modality encoded in any of its sense groups. With a small data set like the one extracted for appear from the ENPC, it may be bold to draw that conclusion. However, as I have discussed above (4.4. and 4.7.1.), even though the corpus is small and the data set restricted, expanding the data set yields no new information, and I have argued that the data set is sufficient for my analysis. Therefore, the conclusion holds.

5.2.6. Translation hierarchy and prototypical translations of appear The analysis of overlaps between the sense groups, i.e. in how many subsets in the restricted t-image an expression occurs, shows that se ut som and synes are the most prototypical meanings of appear. They occur in four subsets. Next is virke (som) which occurs in three subsets and gi inntrykk av and kunne which occur in two subsets each. In other words, the most prototypical translations of appear are strongly associated with visible evidence.

5.2.7. Overlapping sense partitions of seem and appear There are two additional noticeable findings from the Semantic Mirrors-analysis. One is that the largest sense partition for appear (Group 1A) is a subset of the largest sense partition of seem (Group 2S), as shown in Figure 6. The other is that the sense partitions of seem and appear can be placed on a scale, ranging from senses relating to visual evidence, and thus displaying a high degree of evidentiality, to senses that convey epistemic assessment and thus have a high degree of epistemic modal meaning (Figure 7.). The relationship between the sense partitions and the type of evidence will be analysed further in 5.4.

52

Figure 7. Sense groups of seem and appear.

Note that there are several additional subsets and overlap relations between subsets within the large sense group for seem that are not shown in this illustration.

5.3. Grammatical form and syntactic structures of seem and appear and their correspondences This section presents the findings from the contrastive analysis of the correspondences of seem and appear and explores what the grammatical constructions and syntactic forms in both the English originals and the correspondences reveal about the similarities and differences between seem and appear. The distribution of congruent, non-congruent and zero correspondences is presented in Table 2 below.

Table 2. Congruent, non-congruent and zero correspondences of appear and seem, raw number and percentage in relation to the total number.

Appear Seem

Congruence 26 60 % 174 65 %

Non-congruence 9 23 % 57 21 %

Zero correspondence 8 16 % 38 14 %

Total 43 100 % 269 100 %

The overview of the correspondences in Table 2 (4.5) shows that congruent correspondences, i.e. a lexical verb corresponding to seem or appear with the same overall structure as the English sentence, are the most frequent types of correspondences for both seem and appear. Non-congruent correspondences, i.e. translations that changed the overall structure of the original sentence, but

53 kept the semantic or modal meaning, make up about a quarter (23 %) of the correspondences for appear, and about a fifth (21 %) of the correspondences for seem. Zero correspondences, i.e. omissions or significant restructuring of clauses with a clear loss of evidential and modal meaning, are the least frequent correspondences. Even so, in more than one in ten sentences (16 % for seem and 14 % for appear) the modal or evidential element disappears in the translation.

5.3.1. Congruent correspondences More than half the sentences with appear are translated with a congruent correspondence (60 %). For seem the number of congruent correspondences is even higher: 65 % of the translations are done by way of a congruent expression. For both seem and appear, approximately two thirds of the congruent correspondences are lexical verbs. They include simple verbs (synes, virke), complex verbs (late til) and phrasal verbs (gi inntrykk av). The most frequent congruent correspondences are late til, se ut som/til, synes and virke for seem and virke som, se ut som/til, and synes for appear, which suggest that if we go by frequency alone, the two expressions would be practically synonyms. (68) and (69) are examples of congruent correspondences of seem and appear in which the original construction with anticipatory it plus a that clause is kept in the translation (68) and the construction with a nominal subject in thematic position is retained in the translation (69).

(68) It was an uncertain profession, of course, and nothing was guaranteed, but it seemed that he had the gift. (AB1) Det var et usikkert yrke, selvfølgelig, og ingenting var garantert, men det virket som han hadde talent. (AB1T) (It was an uncertain profession, of course, and nothing was guaranteed, but it seemed that he had talent) (69) But he was lonely, polite, and he appeared to be fond of the little girl, and Martine, thinking of the room she had vacated in the rue Washington, took courage and ended her connection with France. (AB1) Men han var ensom, høflig, han syntes å være glad i den lille piken, og Martine, som tenkte på rommet hun hadde forlatt i Rue Washington, tok mot til seg og brøt forbindelsen med Frankrike. (AB1T) (But he was lonely, polite, he appeared to be fond of the little girl…..)

5.3.2. Non-congruent correspondences About a quarter of the sentences with appear (23 %) and a fifth of the sentences with seem (21 %) are non-congruent correspondences. A further analysis of the correspondences shows that seem and appear differ when it comes to which non-congruent expressions they co-occur with. Seem is found to have a wider range of non-congruent correspondences, which indicates that seem is more vague than appear, and possibly more difficult to translate. Below are listed the non-congruent correspondences found in my material.

Restructuring to make the experiencer (oblique object) the subject

A common translation strategy is to restructure the clause and make the explicit or implicit experiencer of the original sentence the subject in the translation. The subject is typically combined with a perception verb of a mental process: tenke ('think'), føle ('feel'), synes ('have an opinion') and mene ('think', 'have an opinion'). In the translation pairs included in this study, there are several examples with syens (70) and one example with mene (71).

54

(70) It seemed to her to be too white, too soft, too spotty. (FW1) Hun syntes den var for hvit, for bløt, med for mange føflekker. (FW1T) (She thought it was too white, too soft, with too many spots.) (71) It seemed obvious to Aristotle that Plato had brought abstruse philosophical theories into his Phaedo, Symposium, and Timaeus that could not have been held by the unprepossessing figure of the Socrates he had dramatized in these works and his Republic and that they had to have originated with Plato himself. (JH1) Aristoteles mente det var åpenbart at Platon hadde trukket dunkle filosofiske teorier inn i sin Faidon, Symposion og Timaios som ikke kunne vært tatt med av den lite vinnende Sokrates-skikkelsen som han hadde dramatisert i de verkene og i Republikken, og at de måtte hatt sitt utspring hos Platon selv. (JH1T) (Aristotle thought/had the opinion that it was obvious that Plato had brought questionable philosophical theories into his Phaedo, Symposium, and Timaeus……)

In the Norwegian translations, the effect of making the oblique object the grammatical subject is similar to the effect of a raised subject (a subject that has been raised from a subordinate clause to a subject position in a matrix clause). The restructuring has the effect of highlighting the experiencer by placing the experiencer in a thematic position (3.3.3.)

In my material, this construction does not occur as a correspondence for appear. This corroborates the findings from Semantic Mirrors that this sense of personal opinion and judgement of a similarity is unique to seem. Moreover, it strengthens the hypothesis that appear is used less frequently to convey opinions and personal judgement (epistemic modality) than seem.

It is worth pointing out at this point that the congruent expression in (69) hand syntes å være, may be the most literal and formal equivalent translation of seem and appear in catenative constructions. However, this translation does come across as a more unconventional grammatical form than the non-congruent forms above (even more so if the experiencer is explicated, e.g. in han syntes meg å være, a rather archaic form in Norwegian, of which there are no examples in my data). Restructuring the clause to avoid a marked form in the target text is a common translation strategy. Making the experiencer the grammatical subject of the clause keeps the semantic and pragmatic meaning, and makes the Norwegian correspondence not as marked grammatically as it would otherwise be.

Restructuring with dummy subject det (dummy it)

Another frequent type of non-congruent correspondence is to restructure a clause with a nominal subject into a clause with dummy det as subject in thematic position and the real subject in a lower position in the clause (72) and (73). These non-congruent correspondences are examples of how a raised subject in the English original becomes a non-raised subject in the Norwegian correspondence. See 3.3.3.

(72) "Nothing that I know of," he replied with a smile, but people seem to think there 's something wrong with my eyes." (OS1) "Ikke noe som jeg vet om", svarte han med et smil, "men det virker som folk tror det er noe i veien med synet mitt." (OS1T) (Not something that I know of, he replied with a smile, but it seems as if people believe there is something wrong with my sight)

55

(73) The law can appear to be broken when some metastable store of internal energy is tapped, as when a match is struck, or a piece of plutonium experiences nuclear fission, but once used up the energy cannot be recovered. (JL1) Det kan se ut som om denne loven blir brutt når man kan tappe et metastabilt forråd av indre energi, som for eksempel når man tenner en fyrstikk eller la et stykke plutonium gjennomgå en kjernefysisk fisjon. Men når energien først er oppbrukt, kan den ikke gjenvinnes.(JL1T) (It can look as if this law is broken when one can tap a metastable store of inner energy…)

The effect of changing the syntactic structure from a nominal subject to a dummy subject is mostly pragmatic and has to do with stance (5.2.2.). Dummy det-constructions are more impersonal and signal more distance between the speaker's modal assessment of the propositional content. The structural change can also affect epistemicity. In the original English sentence above (72), the speaker clearly offers an opinion (albeit tentatively) of what he believes people think, and seem signals epistemic modality. In the Norwegian translation with a dummy subject det, it is less clear whether the speaker's assessment of what people think is primarily based on evidence, and therefore evidential, or on speaker's opinion, and thus epistemic. Non-congruent correspondences, in which the English original clauses are restructured to become clauses with anticipatory or dummy it, are relatively more frequent with appear than seem. In three of the nine non-congruent correspondences for appear (33 %) and seven of the 56 non-congruent expressions for seem (13 %), these are the changes that are made.

Translations with a modal particle

Only seem is translated with a modal particle (74). Translations in which the epistemic modal or evidential meaning is conveyed in Norwegian by modal particles visst, nok, vel or liksom are found in 21 translation pairs (8 %) with seem, see Table 1 in section 4.5.

(74) "You seem to be unaware," he said, "that this is a private lounge. (AH1) "De vet visst ikke at dette er et privat oppholdsrom for leger", sa han. (AH1T) ("You know [modal particle] not that this is a private lounge for doctors", said he.)

The fact that modal particles are correspondences for seem testifies to the modal quality of seem. It is worth pointing out that the modal particles have different meanings and cannot be used interchangeably. The modal particle visst signals inference based on some source of evidence. The core meaning of liksom has to do with similarity and comparison, and functions as an approximator like ' in a way' or 'sort of', and the mode of knowing is perception (Johansson, 2001). Vel, nok (and jo) signal speaker's commitment with no connection to evidence and are arguably more epistemic. The function of all the particles is to reduce the speaker's commitment to the truth of the proposition. In other words, they show epistemic modality, but it can be argued that visst is more evidential in nature. The findings form my analysis thus support the findings from the Semantic Mirrors-analysis in this thesis: that seem has several distinct senses, one of which is more evidential than the other, and

56 two of which are more epistemic than those of appear, as shown by the difference in how many correspondences are translations with a modal particle.

Modal adjunct In some of the non-congruent correspondences, seem and appear are translated with an adjective phrase with modal or evidential meaning (75) or a prepositional phrase functioning as a modal adjunct (76).

(75) So a snarling, spitting, clawing prey is given rather more respect than it may appear to deserve. (DM1) Derfor vises et snerrende, frådende og klorende bytte større respekt enn det tilsynelatende fortjener. (DM1T) (Therefore a snarling, frothing, clawing prey is shown greater respect than it apparently deserves.) (76) The answer to this old dilemma seems to lie in the nature of Darwin's world. (SJG1) Løsningen på dette gamle problemet ligger etter alt å dømme innebygd i den darwinistiske verdens spesielle egenart. (SJG1T) (The solution to this old problem lies – judging by all signals – built into the Darwinian world's special characteristics that are unique to it.)

The Norwegian correspondences with modal adjuncts are both epistemic and evidential. They weaken the speaker's commitment to the truth value of the proposition, but they also clearly refer to visual evidence (e.g. tydelig, klart, åpenbart, which can all be glossed as 'clearly' or 'evidently') or an inference (judgement based on some kind of evidence). In my material, evidential adjuncts occur as correspondences for both appear and seem, whereas the expression etter alt å dømme, which is more epistemic, only occurs as a correspondence of seem.

Modal auxiliaries

Both seem and appear are sometimes translated with a modal auxiliary. In most of the translation pairs with a modal auxiliary, the modal auxiliary co-occurs with another frequent or prototypical correspondence e.g. se ut som (77). The modal auxiliary gives an added effect of hedging, but the evidential meaning is conveyed by the lexical verb. In five of the translation pairs with seem (9 % of the non-congruent correspondences) the modal auxiliary was the only correspondence of the verb (78). A modal auxiliary on its own is never a correspondence for appear. (79) is an example of a non- congruent correspondence with a modal auxiliary and a cognitive verb in a significant restructuring of the syntax, from a passive construction in the English original to an active, but impersonal pronoun in the translation.

(77) The law can appear to be broken when some metastable store of internal energy is tapped, as when a match is struck, or a piece of plutonium experiences nuclear fission, but once used up the energy cannot be recovered. (JL1) Det kan se ut som om denne loven blir brutt når man kan tappe et metastabilt forråd av indre energi, som for eksempel når man tenner en fyrstikk eller la et stykke

57

plutonium gjennomgå en kjernefysisk fisjon. Men når energien først er oppbrukt, kan den ikke gjenvinnes. (JL1T) (It can look as if this law is broken when …..) (78) "I do n't seem to remember your ever giving Molly much of a chance," said Frederick. (DL1) "Jeg kan ikke huske at du noensinne har gitt Molly noen særlig sjanse," sa Frederick. (DL1T) ("I cannot remember that you have ever given Molly any particular chance, said Frederick.") (79) The answer seems to be that humans made the face. (CSA1) Rent umiddelbart skulle man tro at det må være mennesker som har tegnet dette ansiktet. (CSA1T) (At first glance, one would assume that people must have drawn this face.)

Modal auxiliaries as correspondences to seem and appear are rare, both as an independent correspondence and in combination with a prototypical correspondence. There are six in all for seem and two in all for appear (see frequency list in 4.4.). This suggests that that the modal auxiliaries are not well suited to convey the same type of modality as seem and appear. It also suggests that the epistemic modal or evidential meaning and pragmatic function of seem and appear are unique, and that seem and appear fill a gap in the modal system.

Comparative expressions and hypotheticals

As discussed in Chapter 3, seem and appear are both used to make comparisons. They can relate the propositional content to an imaginary situation or a situation that may not be true but that is likely or possible (3.3.1.) or an unreal or improbable situation (3.2.9). Both verbs are used as expressions to relate the proposition to an imagined or possible truth (hypotheticals). Comparisons are also made between how something how something looks or is perceived to some established notion, like a rumour, a documented truth or an image. These are the more straight-forward comparisons (3.2.10).

While translations of appear frequently use an evidence-based verb (like the Norwegian correspondence se ut som ('look like') in (81), only translations of seem are done with copular verbs (80). Norwegian correspondences in the form of comparative expressions include the copular verbs være ('be') and bli ('become') and the som ('as' or 'like').

(80) She seemed to know that neither crying nor sulking ever got anyone anywhere. (RD1) Det var som om hun skjønte at det ikke nyttet verken å gråte eller sutre. (RD1T)

(It was as if she realized that it was no use to cry or sulk.)

These findings support the findings from the Semantic Mirrors-analysis that the sense partition of not just looking like something else, but being like something else is unique for seem. Comparative

58 constructions with copular være ('be') or bli ('become'), are predominantly expressions of personal opinion based on one's own perception, the evidence they are based on is less clear.

Both seem and appear have correspondences that are comparative expressions involving the sensory verb se ut som ('look like') which clearly indicates/denotes knowledge derived from sensory evidence (81) and (82). The use of a sensory verb plus a conjunction is an established strategy for expressing lesser reliability (Chafe, 1986: 267).

(81) They have special glands around their eyes for getting rid of excess salt, which is why they appear to cry when out of water. (ML1) Rundt øynene har de spesielle kjertler som skiller ut overskudd av salt. Det er derfor de ser ut som om de gråter når de kommer opp av vannet. (ML1T) (…. That is why they look as if they cry when they come out of the water.) (82) David seemed to wince and suffer, but he had to face it: what mattered was the house and the life that would be lived in it. (DL1) David så ut som om han krympet seg og led, men han måtte innrømme det: Det som betydde noe, var huset og det liv som skulle leves i det. (DL1T) (David looked as though he cowered and suffered, but he had to face it…)

5.3.3. Zero correspondences Appear disappears from the translations in 16 % of the translation pairs, through omissions or restructuring of the clause with significant loss of evidential meaning (see example (60) in 4.3.1). In one translation pair categorized as a zero correspondence, appear is a lexical verb whose main communicative function is to convey the semantic meaning of the word: 'to become visible' or 'to show signs of something' (83) with no evidential or modal meaning, but this use of appear is a marginal case is my analysis.

(83) Evelyn kept calm and refused to appear to worry: Ted was far from calm, and would declaim his anxieties to anyone who would listen. (MD1) Evelyn tok det rolig og nektet å vise at hun var bekymret. Ted var langt fra rolig, han snakket om sine bekymringer til alle og enhver som ville høre på ham. (MD1T) (Evelyn kept calm and refused to show that she was worried.)

14 % of the correspondences of seem in my material are zero correspondences. This is in line with Johansson's (2001: 221) observation, that "the English word seem sometimes seems to disappear without a trace in translations into Norwegian and likewise might be added, seemingly without any motivation, by English translators in rendering Norwegian original texts".

Zero correspondences with seem are omissions or significant restructuring of clauses with a clear loss of evidential and modal meaning (see example (59) in 4.3.1). However, in some of the translations pairs with seem categorized as zero correspondences, the modality or evidentiality conveyed by seem is lost, but the translation retains some of the modality through other lexical items or contextual clues. For example, in (84), seem is chosen as copular verb in the original English sentence to convey epistemic modality. In addition to seem, the adjective probable conveys epistemic

59 modality and introduces uncertainty about the propositional content. In the translation, the Norwegian equivalence of copular be (være) and the adjective sannsynlig ('probably') are used, and some epistemic modality is retained because of the adjective. However, there is no doubt that the degree of uncertainty conveyed when we say that something seems probable is greater that when we say that something is probable. Because of this loss of epistemic meaning, the construction is classified as a zero correspondence, even though the epistemicity is not completely gone.

(84) Accurate numbers are difficult to estimate, but it seems probable that the Library contained half a million volumes, each a handwritten papyrus scroll. (CSA1) Det er vanskelig å angi nøyaktige tall, men det er sannsynlig at biblioteket omfattet en halv million titler, det vil si håndskrevne papyrusmanuskripter. (CSA1T)

(It is difficult to provide exact numbers, but it is probable that the library contained half a million titles…)

In other examples, zero correspondence with seem does not necessarily result in loss of meaning. For example, in (85), the verbs seem and hear are both perception verbs, and what they convey is sensory evidence and the experience of the experiencer: a dizzying feeling and echoing voices. In the Norwegian translation, seem is dropped completely. However, it is clear from the context and the narrative style, that what is stated in the proposition is the experience of the experiencer.

(85) The room seemed to swim and he heard both his and Tom's voice echoing. (MM1) Alt rundt ham fløt, og ekkoet fra hans egen og Toms stemme ringte i ørene. (MM1T) (Everything around him was floating, and the echo from his own [voice] and Tom's voice rang in the ears.)

The fact that appear has a higher number of zero correspondences than seem may be an indicator that it is less modal that seem, and therefore more easily dropped in translation.

5.4. Evidential and epistemic meaning There are significant differences between the type and source of evidence signalled by seem and appear.

5.4.1. Type and source of evidence An overview of the type and source of evidence (4.3.2.) that seem and appear refer to in the translation pairs included in my data set, and the distribution of type of evidence for each is presented in Table 3 below.

60

Table 3. Type and source of evidence in sentences with appear and seem, raw frequencies and percentages.

Appear Seem

Raw Percentage Raw Percentage frequencies frequencies

Direct attested visual 25 60 % 85 32 %

Direct attested auditory 2 4 % 4 2 %

Direct attested other 1 2 % 11 4 %

Indirect reported hearsay 2 4 % 11 4 %

Indirect reported folklore 4 2 %

Indirect inferred result 6 14 % 54 20 %

Indirect inferred reasoning 6 14 % 67 25 %

No evidence/personal judgement 1 2 % 24 9 %

Irretrievable from context 7 3 %

Total 43 100 % 269 ~100 %

Direct attested evidence

Seem and appear refer to direct attested evidence (visual, auditory or other) in about two thirds (66 %) of the translation pairs with appear, and more than a third (38 %) of the translation pairs with seem. In the most straightforward examples, appear refers to visible evidence and how something would look or be perceived by the onlooker (86). In other examples sensory evidence is implied (87). (88) is an example with appear, in which visual or auditory evidence is merely implied.

(86) A completely plain, fawn-coloured antelope standing in a fawn-coloured landscape would appear to be solid-bodied because the sun, beating down on its back, would illuminate this area more than its sides and much more than its belly. (DM1) En fullstendig ensfarget, lysebrun antilope som står i et lysebrunt landskap vil virke massiv og tredimensjonal fordi solen skinner ovenfra og belyser ryggpartiet mer enn flankene og mye mer enn buken. (DM1T) (A completely fawn-coloured antelope which stands in a fawn-coloured landscape will appear solid-bodied and three-dimensional because the sun shines from above and illuminates the back more than the sides and much more than its belly.) (87) And Roberta, putting her arm swiftly around her friend's shaking shoulders, said softly, "Faye, Faye darling, Faye, Faye," until the girl suddenly shuddered and seemed to go limp, and collapsed into her arms. (DL2) Og Roberta la armen om venninnens skuldre og sa lavt: "Faye, kjæreste Faye, Faye, Faye," til jenta plutselig skalv voldsomt og liksom falt sammen i armene hennes. (DL2T)

61

(And Roberta put her arm around her friend's shoulders and said softly: "Faye, Faye darling, Faye, Faye," until the girl suddenly shivered violently and [modal particle] collapsed in her arms.)

(88) He appears to be talking more to himself now than to her. (ABR1) Nå virker det som om han snakker mer med seg selv enn med henne. (ABR1T) (Now it appears as if he is talking more to himself than with her.)

The findings also show that appear is more frequently used to convey visual evidence than seem. Appear refers to direct attested visual evidence in more than half the translation pairs (60 %), compared to seem, which refers to direct attested evidence in about a third of sentences (32 %).

The most frequent Norwegian correspondences for appear when referring to direct attested evidence are virke som, se ut som/til and late til. The Norwegian correspondence se ut som has an obvious link to visible evidence. The correspondence virke som for seem and appear is more difficult to gloss – the closest meaning is perhaps 'to give the impression of something' or 'to invite a certain interpretation' both of which convey that there must be some attested evidence that triggers a comparison either with how something looks (the given impression) or how something is interpreted (evidence that the interpretation is based on.) The correspondence late til has a similar link to direct evidence: late til is best glossed as 'give impression of being', 'pretend to be' or 'look as if', which all emphasize the presence of attested evidence.

Moreover, both instances of fremstå in my material are connected to visual evidence and how someone carries themselves, i.e. how they look or are perceived. (5.2.5).

Another finding is that all but one of the zero correspondences for appear happens when appear refers to direct attested visual evidence. This indicates that when something is seen, there is little room for doubt and no reason to hedge, therefore the epistemic modality or evidentiality may be dropped in the translation.

Indirect evidence

Indirect evidence is referred to more frequently with seem than with appear. This applies to all categories of indirect evidence – reported or inferred. In total, seem communicates indirect evidence in more than half of the sentences in my material (51 %) compared to appear (34 %).

Inferred evidence is the most common type of indirect evidence for both seem and appear. Inferences are notoriously tricky to analyse with respect to modality and evidentiality because they combine evidence and personal judgement, and it is sometimes difficult to determine which is which (see chapter 2.6.). The variables in this thesis distinguish between inferences based on observable results which are more evidential (89), and inferences based on logic and personal reasoning, which are more epistemic (90). Type of evidence is presented in 4.3.2. In (89) and (90) the source of the modality in the proposition is the reasoning of the speaker based on his own observations, in (91) the evidence is the speaker's comparison of his own observations to reported evidence.

(89) It seemed that bingo afternoons left her so exhausted both physically and emotionally that she never had enough energy left to cook an evening meal. (RD1) Bingospillingen var tydeligvis så slitsom at hun ikke hadde krefter igjen til å lage en skikkelig middag. (RD1T)

62

(The bingo was evidently so tiring that she did not have the strength left to make a proper dinner.)

(90) My mother appeared to be coping well in fact, she seemed almost exhilarated. (JB1) Moren min lot til å ta det bra — faktisk virket hun nesten oppstemt, men så fikk hu kreft, overalt. (JB1T) (My mother appeared to take it well – in fact she seemed almost exhilarated, but then she got cancer, everywhere.) (91) The tales he had heard about evacuees did n't seem to fit Willie. (MM1) Historiene Tom hadde hørt om evakuerte unger stemte visst ikke i Willies tilfelle. (MM1T) (The stories Tom had heard about evacuated kids did [modal particle] not fit in Willie's case.)

A total of 45 % of translation pairs with seem are inferences compared to 28 % for appear. For seem, inferences based on logic and reasoning are more frequent than inferences based on some observable result (25 % to 20 %). For translation pairs with appear, there is an even split between the two subtypes of inferences (14 % and 14 %). These findings indicate that seem is used to convey personal judgement and epistemic modality more often than appear, and vice versa; that appear is used more frequently to convey evidentiality.

The Norwegian correspondences for seem in inferences based on reasoning are very diverse. The most frequent ones are virke som, synes, se ut som/til and late til. Not surprisingly, correspondences with cognitive verbs are also found in this category; correspondences like føles som, kjennes som, minne om and etter alt å dømme. When appear is used in inferences based on reasoning, synes is the preferred Norwegian correspondence (once expressed as tilsynelatende). The only other correspondence is forekomme, which occurs only in this one instance in my material.

5.4.2. No evidence/personal belief as source of modality In about a tenth (9 %) of the translation pairs with seem, there is little or no manifestation of evidence in the original language. For example, in (92) the proposition is a formulation of speaker's opinion, and no evidence is presented. Even though there may be some evidence that has triggered the opinion, the personal judgement is present in the sentences, the evidence is not. Translation pairs in this category thus convey epistemic modality.

(92) She seemed reluctant to admit that much. (SG1) Hun virket uvillig til å innrømme såpass. (SG1T) (She seemed unwilling to admit that much)

The only example in which appear does not refer to any evidence in my material, is a special case of interpreting the letter of the law, making it both epistemic and evidential (93) with a zero correspondence as its translation:

(93) 2. If it appears to the Contracting Parties that such an association should be extended to other committees which present similar characteristics, the EEA Joint Committee may amend Protocol 37. (AEEA1)

63

2. Dersom avtalepartene kommer til at tilknytningen burde utvides til andre komitéer med lignende kjennetegn, kan EØS-komitéen endre protokoll 37. (AEEA1T)

(2. If the Contracting Parties arrive at the conclusion that the association should be extended to other committees with similar characteristics, the EEA Joint Committee may amend Protocol 37.)

The fact that only translation pairs with seem – bar the one exception above - fall into this category, supports the hypothesis that seem has a stronger component of epistemic assessment than appear.

The most frequent Norwegian correspondences in translation pairs with no evidence are virke or virke som, which are almost twice as frequent as synes and være som.

5.4.3. Distribution of sense groups The contrastive analysis shows that the sense partition that seem and appear have in common (Group 2s for seem and Group 1A for appear) is the most frequent meaning of both seem and appear. 63 % of the translation pairs with seem and 65 % of the translation pairs with appear are categorized as members of these sense groups (Table 4).

Table 4. Distribution of sense-groups of seem and appear, raw frequencies and percentages.

seem appear

Sense groups Raw Percentage Raw Percentage frequencies frequencies

Group 1S (arte seg som) 1 ~ N/A N/A

Group 2S (virke, se ut som) 170 63 % N/A N/A

Group 3S (være som om) 14 5 % N/A N/A

Group 1A (virke, se ut som) N/A N/A 28 65 %

Group 2A (fremstå som) N/A N/A 2 5 %

Group 3A (fremgå) N/A N/A 1 2 %

Other 46 17 % 3 9 %

Zero correspondences 38 14 % 8 19 %

Total 269 ~100 % 42 ~100 %

The other sense partitions are far less frequent. The second most frequent sense partition for seem (Group 3S) is that of a more deeply rooted similarity (to be or feel like something). This sense partition is unique to seem and make up five percent of the translation pairs. The second most frequent for appear (Group 2A) is the sense partition that adds emphasize to the given impression (fremstå).

64

The correlation between the sense partitions and type of evidence support the findings from the Semantic Mirrors-Analysis that the different senses of seem and appear can be ranked according to how evidential or how epistemic they are (5.2.6.). They range from referring only to direct evidence and showing a high degree of evidentiality on the left of the scale, to indicating personal judgement and a high degree of epistemic modality on the far right of the scale, as illustrated in Figure 8.

Figure 8. Sense groups and degree of evidentiality and epistemic modality.

Both translation pairs with fremstå as a correspondence for appear (Group 2A) refer to direct attested visual evidence and the sense group is thus placed on the left-hand side of the scale. The sense group with arte seg som (glossed as 'take form as' or 'pretend to be') as a correspondence for seem (Group 2S) also denotes direct attested visual evidence and is placed on the left, with high degree of evidentiality.

The largest sense group for seem (Group 2S), of which the largest sense group for appear (Group 1A) is a subset, refers to direct, indirect and inferred evidence. The correspondence fremgå, which was determined by the semantic Mirror analysis as being inferential based on observable evidence, has ' Indirect inferring reasoning' as its source of evidence according to the contrastive analysis. This source of evidence has a strong component of epistemic modality (2.7.), and the sense group is on the far end of the scale. The third sense group for seem (Group 3S) is somewhat heterogeneous with respect to its inferential source, in that four translation pairs have no evidence/personal judgement, four translation pairs are inferences based on reasoning, and two translation pairs are inferences based on result, which means that 70 % of the members of this sense group indicate epistemic assessment, placing it on the far end of the scale.

5.5. Pragmatic functions of seem and appear Seem and appear also vary with respect to their pragmatic functions (4.5.3.). The distribution of the different pragmatic functions of seem and appear plus a to-infinitive clause and a that-clause is presented in Table 5.

65

Table 5. Pragmatic functions of appear and seem, raw frequencies and percentages.

Appear Seem

Hedging 28 67 % 192 71 %

Stance 4 9 % 69 26 %

Politeness marker - - 4 1 %

Other 10 24 % 4 1 %

Total 42 100 % 269 ~100 %

5.5.1. Hedging The most common function of seem and appear is hedging. To hedge is to communicate a lack of commitment to the truth value of the proposition and reduce the illocutionary force of the utterance (2.6.3.). In 67 % of the sentences with appear and 71 % of the sentences with seem, hedging is the pragmatic function of the two verbs. A closer look at the data shows that hedging in translation pairs with appear is for the most part associated with direct attested evidence (68 % refer to direct attested evidence). Hedges with seem refers to direct attested evidence in less than a third (32 %) of the sentences.

Different types of hedges are present in my material. Examples of adaptors (94), (95), )96) adaptors (52) and (53) and modifiers are discussed below. One subtype of hedging that is not specifically mentioned in the literature, but is quite frequent in my material, is when seem and appear are used to signal that what follows is an exaggeration or a metaphor and should not be taken literally.

Adaptors

In (94), (95) and (96) below, seem and appear signal that the speaker does not have enough evidence to make a categorical statement, and that the shared knowledge in the proposition can be negotiated. Prototypical adaptors are comment clauses with seem, which are excluded from my material, and constructions with seem and appear plus like or as if/as though, which are also excluded, since these construction types do not make up my core data (see section 4.5.) This means that although they are interesting, they will not be commented on any further. The examples below are other constructions in which seem and appear function as adaptors:

(94) She fell silent for a moment, then seemed to make up her mind to something. (MW1)

Hun ble stille et øyeblikk, så lot det til at hun bestemte seg for noe. (MW1T)

(She became silent for a moment, then it seemed that she made up her mind about something.)

(95) This analysis seemed to cheer her, and she stirred her coffee with a new access of energy. (MD1) Det så ut til at denne analysen muntret henne opp, og hun rørte i kaffen sin med ny energi. (MD1T)

(It looked as if this analysis cheered her up, and she stirred her coffee with new energy.)

66

(96) It appears to have" — he hesitated — "five outpouchings, if this is the word." (OS1) Det virker som at den har" — han nølte "fem utposninger, om det er ordet." (OS1T) ("It appears that it has" – he hesitated – "five outpouchings, if that is the word."

In (94) the speaker is hesitant to state with 100 percent certainty what is on someone else's mind, hence the hedging. The evidence is present in the English sentence ('she fell quiet') and the scope of the hedge is the speaker's interpretation of the evidence. In (95) the evidence is also present in the English original sentence (the observation of someone stirring the coffee with new energy) and the hedge applies to the speaker's interpretation of the evidence. In the latter case, the uncertainty applies to several factors, one is whether she really was cheered up, the other questions whether it was the analysis that cheered her. The shared knowledge can be negotiated.

(96) is an interesting example, because appear in the English original clearly hedges the otherwise categorical statement of the number of outpouchings. As such it would be a hedge on the speech act to weaken the illocutionary force of the speech act, in this case the declaration (see section 3.5.2.) It may also be a hedge on the choice of expression (outpouchings), in which case it would be a propositional hedge. The choice of appear implies that the statement is backed up by observations (visual evidence) but the speaker is still not willing to communicate absolute certainty. The reference to evidence construes the shared knowledge. At the same time, in the English original at least, the speaker displays a lack of commitment to the truth of the proposition. The Norwegian correspondences for these kinds of hedges are typically the correspondences: late til, se ut som/til, virke som, which all emphasize the evidential aspect of this type of hedge. In other words, this type of hedge seems to be more frequently associated with seem and appear as evidentials.

Modifiers

There are a few examples in my material where seem or appear signal loose interpretations, but this does not seem to be a frequent type of hedge when seem and appear occur with a to-infinitive clause or a that-clause, which form the basis for my analysis. (97) might be an example of how seem functions as a hedge and lessens the speaker's commitment to the truth value of the proposition by being vague, but the sentence itself consists of so many vague expressions ('no one', 'be around' and 'thing') that the vagueness of the sentence does not necessarily have to do with seem.

(97) No one else seemed to be around to notice a thing. (DF1) Det var visst ingen andre som hadde sett opptrinnet. (DF1T) (There was [modal particle] no one else who had seen the incident.)

Hedges that signal indeterminateness and loose interpretations to reduce the speaker's commitment to the truth of the propositional content only occur with seem in my material.

Other modifiers typically display uncertainty and doubt (98) and (99). In (98) the speaker displays uncertainty and doubt as to whether he has interpreted the mood of his fellow dinner guest correctly. The hedge pertains to the accuracy of his interpretation. In (99), observed evidence is implied (the speaker signals that he has heard first names being widely used), but he still conveys uncertainty with respect to the conclusion drawn on the basis of the observation.

67

(98) Mme Wyatt, with whom I was à côté, seemed to enjoy it, or at least to relish the salmon. (JB1) Ikke for det, det smakte helt utmerket, litt fantasiløst, bare ... mme Wyatt, som jeg satt à cote med, lot til å like maten, i det minste laksen. (JB1T)

(…Mme Wyatt, with whom I sat a á côté, seemed to like the food, at least the salmon.)

(99) In spite of the exalted status, in anyone's estimation, of her employers, it appeared that they were all on Christian name terms here, Davina and Harvey and Naomi and Brenda. (RR1) Til tross for den høye status som alle og enhver tilla herskapet, lot det til at alle var på fornavn — Davina og Harry og Naomi og Brenda. (RR1T) (Despite the elevated status that everyone credited the employers with, it appeared that everyone was on Christian name terms – Davina and Harry and Naomi and Brenda.)

Both examples above are inferences. With seem, there are 45 hedges that co-occur with the evidence type 'inferred reasoning', 41 sentences with seem in which seem is a hedge have the evidence type 'inferred result', and 3 do not refer to any type of evidence at all, only epistemic assessment, making the total number of hedges with a component of epistemic assessment 46 %. For appear, the total number of hedges are 28, and only 6 of these (21%) co-occur with the evidence type 'indirect inferred result' and only 4 (14 %) with the evidence type 'indirect inferred result', and none with no evidence, only belief and epistemic assessment as its source. This shows that inferences with seem are more epistemic than inferences with appear.

Boosted modifiers

(100) and (101) are examples of boosted modifiers. In (100) the speaker's commitment to the truth value of the propositional content (that 'the foetus seemed to be trying to tear its way out of her stomach') is strengthened because seem in this case describes how something feels, i.e. sensory evidence. This particular example is also an example of how seem and appear can signal that what comes next is not an accurate comparison, but an exaggeration, (see below). Also, even though (100) is not a straightforward comparative construction, the function of seem in this case is to make a comparison between sensory evidence and the language expressing the experience. (101) is an example of appear functioning as a boosted modifier, in which the reference to evidence strengthens the reliability of the proposition. This is a typical hedge in academic writing. In my material it is – not unexpectedly – more frequent in the non-fictional texts.

(100) She went back to Dr. Brett, for she could not sleep or rest because of the energy of the foetus, which seemed to be trying to tear its way out of her stomach. (DL1) Hun gikk til doktor Brett igjen, for hun fikk hverken sove eller hvile på grunn av det energiske fosteret. Det var som om det prøvde å sprenge seg ut av maven hennes. (DL1T)

(She went to Dr. Brett, for she could neither sleep nor rest because of the energetic foetus. It was as if it was trying to blow itself out of her stomach.)

68

(101) The marsupial mole, like the placental moles it so closely resembles, appears to feed on earthworms and insects. (ML1) Pungmoldvarpen ligner mye på den placentale moldvarpen. Den lever av meitemark og insekter. (ML1T)

(The marsupial mole is very similar to the placental mole. It feeds off earth worms and insects.)

The Norwegian correspondences in the translation pairs where seem functions as a boosted modifier are typically the third sense partition from Semantic Mirrors (group 3S): være som/om and kjennes, which are unique to seem. Note that the Norwegian correspondence of appear as a boosted modifier can be a zero correspondence, indicating that the evidence is so strong that it is redundant and the sentence can be straight forward proposition without a hedge. The function of seem as a boosted modifier thus emphasizes the epistemic quality of seem, whereas the function of appear as a boosted modifier is to strengthen the evidence.

Adjusters

There are examples of adjusters in the translation pairs. When seem and appear function as adjusters they signal that the information is sufficient, even if it is inaccurate. Prototypical examples are approximators (2.4.2). Seem and appear as copular verbs in equative constructions that mark the relationship between the verb and the predicate as approximately equative (3.3.1.) are other examples of seem and appear as adjusters. Adjusters can also signal evidential meaning in the form of imprecision. Adjusters are rare in my material, but (102) might be an example of this type of hedge:

(102) Twice, he 'd been put on probation, but nothing seemed to have affected the nature of his behavior, which appeared nearly pathological in its thrust. (SG1) To ganger var han blitt løslatt på prøve, men ingenting lot til å forandre atferdsmønsteret hans, som virket nærmest patologisk i sin hardnakkethet. (Twice he had been put on probation, but nothing seemed to change his behaviour, which appeared almost pathological in its stubbornness.)

In (102) the scope of the hedge is that nothing has affected the nature of the behaviour. The fact that he is still on probation is not negotiable, and the observation that the behaviour shows signs of being pathological is not negotiable in this context (this is actually negotiated by appear). The information that his behaviour is unchanged, or that it has not changed enough to alter any of these facts, means that even though the information about the unchanged behaviour may not be totally accurate, it is sufficient.

Exaggerations and metaphors

There are several examples in my material where the pragmatic function of seem and appear is to signal that what comes next is not an accurate comparison, but an exaggeration or a metaphor (103), (104). This pragmatic function of seem and appear is exclusive to the fictional text samples.

(103) And this issue he had been hard put to it to find a couple of intelligent letters for the correspondence page; sometimes it seemed that every crackbrain in north-east Norfolk read the PANUP newsletter but that no one else did. (PDJ3)

69

Og til dette nummeret hadde han hatt store vanskeligheter med å få inn et par intelligente bidrag til brevspalten. Noen ganger virket det som om hver eneste skrulling nordøst i Norfolk leste PANUP-bladet, men ingen andre. (PDJ3T)

(…sometimes it seemed that every madman in Norfolk read the PANUP Magazine, but no one else.)

(104) She appeared to be wearing parts of a dead goat on her feet (ST1)

Hun så ut til å ha tullet føttene inn i deler av en død geit. (ST1T)

(She looked as if she had wrapped her feet in parts of a dead goat.)

14 % of the hedges with appear fall into this subcategory of hedges, compared to about 10 % of the hedges with seem. The fact that appear is more frequently used to signal a metaphor or an exaggeration may be because the exaggerations and metaphors are made by way of comparisons between what is attested and the lexical choices for what one is trying to express, which would be another manifestation of evidentiality.

5.5.2. Stance In my analysis 'stance' is defined as a marker of commitment or detachment, i.e. as a marker of how seem and appear function as a way of distancing the speaker from the propositional content (105) or - the opposite - taking a stronger ownership to the propositional content on the speaker's behalf (106) or on a third party's behalf (107).

(105) Recently, the World Bank has come under fire from several quarters — academics, grassroots US environmental pressure groups and the US Congress — for funding projects which appear to be destroying both the local environment and local cultures. (LT1) I det siste er Verdensbanken blitt kritisert fra flere kanter. Akademikere, amerikanske miljøvernforkjempere på grasrotplan og den amerikanske kongressen har kritisert banken for at den har finansiert prosjekter som ser ut til å ødelegge både det lokale miljøet og den lokale kulturen. (LT1T) (…and the American Congress has Criticized the bank for having financed projects that appears to be [lit: looks to be] destroying both the local environment and the local culture.) (106) If anything, Elizabeth was slightly superior in breeding to Yvette, whose touching childhood, the half-remembered episode of her mother urging her to eat in the train to Bordeaux — an episode which so moved him when he first heard it — seemed to clothe her in a vulnerability of which she remained unaccountably unaware. (AB1) Elizabeth hadde nok litt mer dannelse enn Yvette; hans kones rørende barndom, den halvglemte episoden med moren som inntrengende ber henne om å spise på toget til Bordeaux — en episode som gjorde ham svært beveget første gang han hørte den — alt syntes å gi henne en slags sårbarhet som hun på uforklarlig måte selv var uvitende om. (AB1T)

70

(…everything showed signs that gave her some kind of vulnerability, which she in an inexplicable way was unknowing of herself.)

(107) The smell was almost shut out, but to Alice it seemed that an invisible film of stench clung to everything, and she would feel it slippery on her fingers if she touched. (DL2) Stanken kjente de nesten ikke, men Alice syntes en usynlig hinne av vond lukt lå over alt, som om fingrene hennes ville bli klissete hvis hun rørte ved noe. (DL2T)

(They barely noticed the stench, but Alice thought [that] an invisible film of bad smell clung to everything…)

Stance as markers of detachment or commitment are more frequent with seem than appear. In a quarter of the sentences with seem (26 %), the main pragmatic function is to communicate how close – or distant – the speaker is to the propositional content. For appear, the number of sentences where this is the main pragmatic function is one in ten (9 %). In my material, stance is usually connected to Indirect inferred evidence: three of four translation pairs in which appear functions as a marker of stance refer to inferred evidence, and in about half of the translation pairs in which seem is a stance marker (34 of 69), there is reference to inferred evidence.

In translation pairs in which seem and appear signal detachment, there is often an explicit third-party experiencer (107) and the Norwegian correspondence is typically a restructuring of the sentence so that the experiencer becomes the subject (5.3.2.). Furthermore, the source of evidence determines how invested the speaker is. If the evidence is experienced second hand or third hand, seem and appear signal distance (105).

Less distance to the propositional content – or stronger commitment – is the stance most frequently communicated when the speaker is the source of the evidence, by way of his if her perceptions (direct evidence) or his or her logical reasoning (indirect evidence). In (107) seem functions as a perception verb in a sentence with a great many other lexemes that have to do with sensory evidence; smell, stench, and feel. The evidence is categorized as 'other sensory', and the stance is one of commitment. The commitment is somewhat weaker when the type of evidence is the speaker's interpretation of the facts (inferences made on the basis of some observed result). (107) is an example of how seem signals that the propositional content is the speaker's interpretation – he is the third person narrator in this text sample after all. In fictional texts, the voice of the narrator will matter: When it is a first-person narrator, seem frequently signals that the speaker takes ownership of the propositional content.

5.5.3. Politeness marker In my material seem is the only expression with the pragmatic function of a politeness marker. Seem functions as a pragmatic marker in only four translation pairs (1 %), and the correspondence is usually a zero correspondence (108) or the pragmatic particle visst (109):

(108) Now, what seems to be the problem?" (FF1)

Nå, hva er problemet (FF1T)

71

(109) "You do n't seem to understand. (ST1)

"Du Schinner visst ikke helt dette her. (ST1T)

The politeness in all the translation pairs in my data set is negative politeness. The function of seem in the translation pairs included in my study is soften a blunt and categorical statement, for example that there is a problem (108) or a strong opinion that would cause the hearer to lose face, for example that he does not understand (109). Seem functions as a politeness marker by toning down the impact of the statement to reduce the imposition on the hearer (3.1.3.).

5.6. The relationship between syntactic form, evidentiality and epistemic modality and function When combined, the findings from sections 5.4. and 5.5. show the relationship between syntactic form, evidentiality and epistemic modality and pragmatic function. It turns out that specific syntactic constructions co-occur more frequently with direct and indirect evidence and others with epistemic assessment (no evidence), and that some syntactic constructions are more frequently used in inferences. In short, seem appears to be is more evidential in catenative constructions than in constructions with a that-clause, and appear is evidential in both catenative constructions and constructions with a that-clause.

An overview is presented below in Table 6.

Table 6. Syntactic structures with seem and appear, source of evidence and pragmatic function, raw frequencies.

Evidential Epistemic Epistemic Irretri- Hedging Epistemic Politeness Other (direct (no and evable stance marker and evidence) evidential from indirect (inferences) context evidence)

Seem plus 1 14 18 - 19 13 - 1 that-clause

Seem as 115 10 104 7 173 56 4 3 catenative

Appear 3 1 1 - 3 - - 2 plus that- clause

Appear as 27 1 10 - 25 4 - 9 catenative

72

Seem plus a that-clause

The table shows that seem is more epistemic in syntactic constructions with a that-clause than in catenative constructions. Combined with a that-clause, seem entails direct or indirect evidence only once. The table also shows that in 14 translation pairs (more than 40 % of the translation pairs with seem and a that-clause), seem entails only to personal judgement and epistemic assessment. About half of the translations pairs with seem and a that-clause are inferences (55 %), in which evidentiality and epistemic modality overlap (see section 2.6.4). There are inferences which are based predominantly on the speaker's reasoning (evidence type 'Indirect Inferred reasoning', and there are inferences in which the observed evidence that triggers the reasoning is the dominating component (evidence type 'Indirect Inferred result') (5.4.1.). The numbers indicate that inferences with seem have a higher degree of epistemic modality. The fact that that the syntactic construction with seem in a that-clause is either full on epistemic or conveys a mix of evidential and epistemic meaning, demonstrates the seem plus a that-clause is predominantly epistemic. The syntactic construction is used to hedge in more than half (58 %) of the that-clauses with seem that were possible to analyse with respect to evidence (19 of the 33 that-clauses) or to mark stance in 40 % of the translation pairs with seem and a that-clause.

The numbers also show that inferences with seem and a that-clause are far more frequent than inferences with seem as a catenative (18 instances vs. 104).

Seem as a catenative

Seem as a catenative refers to evidence in half of the translation pairs with seem and a to-infinitive clause that were possible to analyse with respect to evidence (114 translation pairs of 229, see Table 5 above). In contrast, the number of translation pairs with seem plus a to-infinitive clause that were categorized as epistemic are insignificant in comparison, with only 10 translation pairs (4 % of the translation pairs that were categorized with respect to evidence). Seem in a catenative construction is both epistemic and evidential in 45 % of the translation pairs in (104 of 229). In other words, seem as a catenative is predominantly evidential.

Seem as a catenative is used to hedge in seven out of ten sentences with this construction (73 %), and to mark epistemic stance in almost a quarter of the sentences with this construction (24 %). Only catenative seem is used as a politeness marker.

Appear plus a that-clause

Inferences with appear are almost exclusively expressed with appear and a that-clause, not appear in the catenative constructions (ten translation pairs vs. one). This pattern is thus the exact opposite of that of seem in inferences – as mentioned above, seem to is the preferred syntactic construction for inferences signalled by seem. Moreover, appear plus a that-clause is predominantly evidential. The numbers are small, but they do show that the construction with appear plus a that-clause refers to evidence and is thus evidential three times as often as any other category. The pragmatic function of appear in a syntactic construction with a that-clause is to hedge, never to mark epistemic stance.

Appear as a catenative

The syntactic construction with appear as a catenative entails direct or indirect evidence in more than 70 % of the translation pairs with this syntactic construction (27 translation pairs of 33), see

73

Table 6 above. In this syntactic construction, appear signals inference in about 30 % of the translation pairs. Appear is categorized as epistemic only once. These findings indicate that appear in the catenative construction has the highest degree of evidentiality of the two verbs in the two constructions.

Again, the most frequent function is to hedge (66 % of the translation pairs with appear and a that- clause have this pragmatic function).

74

6. Discussion and conclusion The discussion in this chapter is organized to answer the research question presented in Chapter 1:

1. How different or similar are seem and appear with respect to grammar, syntax, semantic meaning and pragmatic functions? 2. How does the interaction between form, meaning and function determine the meanings and functions of the seem and appear? 3. How are these differences reflected in the translation?

The findings from the Semantic Mirrors-analysis and the contrastive analysis are combined and discussed below in order to reveal how different or similar the two expressions are with respect to grammar and syntax (6.1.), semantic meaning (6.2. and 6.3.), pragmatic function (6.4.) and how form, meaning and function are interconnected (6.5.). The suggestion that they might be in the process of becoming auxiliaries is discussed in 6.6.

6.1. Seem and appear differ with respect to grammar and syntax The overview of the grammatical properties of seem and appear (3.1.) and the different syntactic constructions they can occur in (3.2. and 3.3) shows that although seem and appear are both lexical modal verbs, catenatives and perception verbs, with the same grammatical properties, there are some significant differences between them. The major points on which they differ are summarized here:

- Only appear can occur as a one-place verb in a simple sentence with SV-pattern. See section 3.2.1. - Both seem and appear can occur in syntactic constructions with an adjective and a that- clause, but there is a limited number of adjectives that appear can combine with to make an idiomatic sentence. See section 3.2.6. - Seem, but not appear, can combine with a complement in the form of a noun phrase. See section 3.2.7. - Only seem can take an attributive complement in the form of a prepositional phrase. See section 3.2.8. - While both seem and appear can combine with as if or as though to make a comparison, possibly to an imaginary situation or a situation that may not be true but that is likely or possible (3.2.9), Only seem can combine with the more informal like and a noun phrase to make a comparison. See section 3.2.10. - Both seem and appear can be used in parentheticals, but the use of the comment clause it appears is not idiomatic in all contexts, whereas it seems can be used more widely. See section 3.2.11. - Seem appears to be more vague and more semantically bleached than appear, indicated by the greater number and diversity of correspondences. See Table 1 in 4.5.

In short, the differences add up. To answer the first part of the first research question, it is clear from the above, that seem and appear are not that similar with respect to grammar and syntax.

6.2. Seem and appear have semantic meanings that are unique to them. Both analyses in his thesis confirm that seem and appear have overlapping semantic meaning, and thus may be used as synonyms in some contexts, but they also find that each expression has

75 semantic meanings that are unique to them. These nuances in semantic meaning can easily be missed in an analysis that only considers the raw frequencies in the corpus. The study demonstrates that the Semantic Mirrors-method, which takes language in use (parole, not langue) as its starting point, and thus incorporates semantic and pragmatic meaning, uncovers nuances in the semantic meaning that might otherwise have remained undetected. At the same time, the contrastive analysis confirms and corroborates findings from the Semantic Mirrors-analysis and offers insight into other aspects, for example frequency and interdependency between variables like form and type of evidence.

The overview of the correspondences (Table 1 in 4.5.) shows that the most frequent correspondences of appear are also amongst the most frequent ones for seem. These are virke som, late til and se ut som for appear and se ut som, virke som, synes, late til for seem). In other words, if we go by frequencies alone, the two expressions come across as near-synonyms, albeit with seem having a wider range of correspondences, a sign that seem is possibly more vague than appear (3.4.3.).

Moreover, the contrastive analysis shows that the shared semantic meaning identified by the Semantic Mirrors-method – the sense of being perceived as similar to something based on visual evidence – is the most frequent use of both expressions. In about two thirds of the translation pairs (63 % for seem, 65 % for appear) this is the sense of seem and appear in the original text. The combined findings from the Semantic Mirrors-analysis and the contrastive analysis thus confirm that the shared sense partition is the dominating sense of both expressions.

The sense partition of seem indicated by the Semantic Mirrors-analysis that something not only looks like something, but that the speaker makes a personal judgement that something is like something else or feels like something else – testify to the epistemic modality encoded in seem. This result from the Semantic Mirrors-analysis thus support the idea that was introduced at the outset of thesis, that seem can be both evidential and epistemic. For something to look like something else, one must first have made an observation (manifested evidentiality), then made a personal judgement of the similarity and encoded that epistemic assessment of how true the likeness really is into the proposition (epistemic modality). Since this sense partition is unique to seem, the results of the Semantic Mirrors also support the hypothesis that seem is more epistemic than appear (more on this in section 6.3.2. below).

The other sense partitions of seem and appear found in the Semantic Mirrors-analysis may be outliers and special cases, but they seem to fall on either side of the demarcation between evidentiality and epistemicity: Seem's third sense partition of 'pretending to be' or 'acting like something', does have a strong component of personal judgement of something being an act or pretence and not a true similarity (5.1.5.). Again, this emphasizes the epistemic meaning of seem. The other two sense partitions of appear are connected to observed evidence. Based on the Norwegian correspondences, the sense partition 'to be presented as similar' (fremstå) is predominantly concerned with how something looks or is (visually) perceived, less with how it is inferentially judged and assessed. The last sense partition for appear – that an inference has been made from observed evidence (fremgå) – is also predominantly evidential (5.2.5.).

It is interesting that the most frequent correspondences are similar, but not identical, to the most prototypical translations for seem found in by the Semantic Mirrors method. The prototypical

76 correspondences, which are calculated on the basis on their frequency and overlap relations are synes, gi/ha inntrykk av, virke and tyde på for seem and synes, virke/virke som and kunne, before gi inntrykk av and se ut som for appear. I find it particularly interesting that tyde på, which clearly has epistemic meaning, is among the prototypical correspondences for seem, and that se ut som, which is clearly evidential and based on visual evidence, is prototypical for appear, but not for seem, even though se ut som was a high-frequent correspondence for both seem and appear.

6.3. Seem and appear differ with respect to evidentiality and epistemic modality The contrastive analysis confirms that appear is predominantly evidence-based, and that the most frequent correspondences refer to visual evidence (6.3.1.). The findings also show that seem is more epistemic and convey personal judgement and commitment to a greater extent than appear (6.3.2.). The findings suggest that it might be useful to consider modality and evidentiality as a matter of degrees rather than to operate with a strict distinction between the two – especially with respect to inferential meaning (6.3.3.).

6.3.1. Appear is more evidential than seem The findings from the contrastive analysis confirm that appear in combination with a to-infinitive clause and a that-clause is more evidential than seem. This is supported by the following results:

- There are more zero correspondences for appear than seem, indicating that the evidence is strong enough for there to be no doubt or uncertainty about the propositional content, and that the modal element can be dropped in translation. See section 5.3.3. - Appear refers to direct or indirect evidence far more frequently than seem (70 % vs. 44 % of the translation pairs, inferences excluded). See section 5.4.1. - Appear refers to direct attested visual evidence in more than half the translation pairs (60 %). In comparison, seem refers to direct attested visual evidence in about a third of sentences (32 %). See section 5.4.1. - Almost three times as many translation pairs with seem than appear have no manifestation of evidence (9 % vs. 2 %). See section 5.4.1. - Seem and appear have overlapping pragmatic functions. Both are predominantly used for hedging. However, the hedging is overwhelmingly connected to direct attested visual evidence in translation pairs with appear (68 % of the hedges). See section 5.5.1. - The functions of appear as a boosted modifier (i.e. strengthening the illocutionary force rather than weakening it), tend to emphasize the reference to evidence. See section 5.5.1. - The non-congruent correspondences in the form of adverbials refer to visual evidence (tilsynelatende, åpenbart). See Table 1 in section 4.5.

The following findings from the Semantic Mirrors analysis support the claim that appear is more evidential than seem:

- The second most frequent sense partition of appear is fremstå, which in Norwegian is strongly associated with how someone carries themselves, i.e. how they look or are perceived. See section 5.2.5. - One of the sense partitions of appear (fremstå) has the highest degree of evidentiality of all the sense partitions discovered.

77

- The prototypical meaning of appear include the Norwegian correspondence se ut som, which is clearly evidential. This prototypical meaning is unique to appear. See section 5.2.6.

In other words, there is substantial evidence backing up the claim that appear, in the syntactic constructions with a to-infinitive clause (a catenative) and with a that-clause, could qualify as an evidential. Both analyses in in my study confirm that appear has semantic meaning overwhelmingly pertaining to evidence, especially visual evidence. Even if I adhere to the narrowest definition of evidentiality, which restricts evidentials to members of a grammatical class (see chapter 2.6.2.), it is possible to argue that appear could qualify as an evidential in catenative constructions and constructions with appear and a that-clause.

6.3.2. Seem is more epistemic than appear The findings from the contrastive analysis confirms independently that seem in combination with a to-infinitive clause and a that-clause is more epistemic than appear. The hypothesis is supported by the following results:

- Seem is translated with a modal particle, a modal complement or a modal adjunct more frequently than appear. The Norwegian correspondences thus emphasize the epistemic modality of seem. See section 5.3.2. - Only seem is translated with an explicit experiencer and a verb of cognition e.g. mene. See Table 1 in section 4.5. - A fair number of the modal adjuncts that are correspondences of seem (but not of appear) have epistemic meaning, e.g. etter alt å dømme, tydeligvis. See Table 1 in section 4.5. - Hedges that signal indeterminacy and loose expressions (i.e. weak evidence) only occur with seem in my material. See section 5.5.1. - The function of seem as a boosted modifier (when its function is to strengthen the force or the truth of an utterance) tends to emphasize the personal assessment and epistemic quality of seem. In contrast, appear as a boosted modifier has the opposite effect and tends to emphasize the reference to evidence and thus the evidential meaning and function of appear. See section 5.5.1. - Seem is used more frequently than appear to signal stance (26 % compared to 9 %). When the speaker is the source of the evidence, e.g. by way of his if her perceptions (direct attested evidence) or his or her logic reasoning (indirect evidence), the stance signals stronger commitment to the propositional content. See Table 4 in section 5.5. and section 5.5.2. - Seem as an epistemic marker is more frequent in syntactic constructions with seem and a that-clause, than when it is a catenative. See section 5.6.

The following findings from the Semantic Mirrors-Analysis support the claim that seem is more epistemic than appear:

- The second most frequent sense partition of seem is the sense of a more deeply rooted similarity than just looking like something else, but' being like something' or 'feeing like something' (Group 3S). This sense partition has the highest degree of epistemic modality of all the sense groups. See section 5.1.5. - The prototypical meaning of seem includes the Norwegian correspondence tyde på (best glossed as 'imply') which has epistemic meaning. See section 5.1.6.

78

The findings above thus uphold the established notion that seem is a lexical with epistemic assessment and personal evaluation imbued in the semantic meaning, and show that seem in comparison with appear is more epistemic (3.4.1.).

There is also an argument to be made that even though seem as a perception verb is used to relate attested evidence, there is a stronger element of personal evaluation and assessment when we talk about how something feels or is perceived than when we refer to observed evidence. After all, an observation can be verified by others and can be objective, how one feels is entirely subjective. Consequently, even if seem as a perception verb were to qualify as an evidential, it comes across as more epistemic than appear.

Parallel to the argument that appear could qualify as an evidential, there is evidence to support the claim that seem, in the syntactic construction with a to-infinitive clause (a catenative) and a that- clause, could qualify as a modal. Both analyses in in my study confirm that seem has semantic meaning overwhelmingly pertaining to epistemic modality, and seem as a catenative and in combination with a that-clause could qualify as a modal, at least if modality is defined as a semantic concept (2.3).

6.3.3. Seem is both epistemic and modal The findings from both the contrastive analysis and the Semantic Mirrors-analysis show that seem can be either epistemic (with no reference to evidence, only personal judgement of the truth of the proposition) or evidential (evidence is entailed) or it can be both (e.g. in inferences). This is confirmed by the following findings in the contrastive analysis:

- About a tenth (9 %) of the sentences with seem have the speaker's opinion as source of modality – no evidence is mentioned. In these translation pairs seem conveys epistemic modality. See Table 3 in section 5.4.1. - In about four out of ten sentences seem refers to direct or indirect evidence (44 %) and is thus evidential. See Table 3 in section 5.4.1. - In almost half the translation pairs seem is used to signal an inference (45 %). Inferences are both epistemic and evidential: Inferences based on speaker's reasoning make up 25 % of the translation pairs, and inferences in which the speaker makes an assumption or deduction based on some observable evidence make up 20 %. In comparison, 28 % of translation pairs with appear are inferences. See Table 3 in section 5.4.1.

The findings in the Semantic Mirrors-analysis that support the claim that seem is both epistemic and evidential are:

- Sense partitions of seem are found on both ends of the scale measuring degree of evidentiality and epistemic modality, indicating that seem can have a high degree of epistemic modality and a low degree of evidentiality in some contexts and a high degree of evidentiality and low degree of epistemic modality in others. See section 5.4.3.

Even though my approach to evidentiality and epistemic modality is to consider evidentiality and epistemic modality as two separate categories, it is clear from the above that in inferences, evidentiality and epistemic modality co-exist in inferences (2.6.3): However strong the evidence, the inference signals that a reasoning process has taken place, and vice versa: logical reasoning and personal judgement typically has some kind of evidence as its starting point.

79

In inferences, epistemic modality and evidentiality thus become a matter of degrees. The interesting question is not whether seem is evidential or epistemic, but whether the evidentiality is stronger than the epistemic modality, or the epistemic modality is stronger than the evidentiality. In the contrastive analysis, translation pairs where seem signalled inferences are evenly distributed between being based on reasoning and results, but the balance tips slightly in favour of inferences based on reasoning (25 % of the translation pairs) compared to inferences based on results (20 % of the translation pairs). The numbers indicate that inferences with seem have a higher degree of epistemicity, but not by much.

In short, my study shows that seem can be either evidential, epistemic and it can be both in inferences, in which case seem has a higher degree of epistemic modality than appear.

6.4. Seem and appear have different pragmatic functions that reflect their epistemicity and evidentiality As discussed in 5.5. seem and appear have various pragmatic functions. To answer the question of how the pragmatic functions of seem and appear differ and how that is reflected in the translation of the two expressions, it is useful to first establish that seem and appear have many pragmatic functions in common. Both expressions are used primarily to hedge, both are predominantly referring to direct attested evidence, and both are frequently used to signal that an exaggeration or a metaphor follows and function as a signal to the reader not to take what follows too literally. I therefore conclude that seem and appear not only have overlapping semantic meaning, they also have similar and overlapping pragmatic functions. The shared correspondences and the fact that the largest and most frequently used sense partition of appear is a subset of the largest and most frequently used sense partition of seem testify to this.

There are, however, some functions that are unique to seem. For example, only seem (in a catenative construction) functions as a politeness marker in my material (see section 5.5.3. and Table 5 in section 5.6.). In my opinion, it is the epistemic qualities of seem that makes it effective as a politeness marker: With seem, the speaker assumes responsibility of having assessed the situation (perhaps wrongly) to save one's own face or the face of the addressee. When the speaker is always the implied experiencer (3.3.2.), seem mitigates the force of an otherwise strong or uncompromising statement by allowing the speaker to take responsibility for the statement (3.5.3).

The function of signalling loose meanings is also unique to seem. This too, is likely because of the epistemic meaning of seem, which to a greater extent than appear communicates the speaker's assessment and take on the situation, see section 6.3.2. above. Imprecision and loose meanings are thus expressions of the speaker's interpretation and opinions, and that particular type of modifier (5.5.1.), which is unique to seem, is thus a consequence of the epistemic modality imbued in seem.

Another finding is that epistemic stance as a marker of detachment or commitment is more frequent with seem than appear. The contrastive analysis shows that hedges in the form of boosted modifiers emphasize the epistemic quality of seem and the evidential quality of appear respectively. The consequence is that seem functions as linguistic means to signal closeness to and personal investment in the propositional content, and appear as a boosted modifier is used to focus on the evidence and thus signal speaker's detachment.

80

To conclude, this study finds that seem and appear share many pragmatic functions, but when they differ, the pragmatic functions of seem can be traced back to the epistemic qualities of seem, and the pragmatic functions of appear are connected to the evidentiality of appear. The most deeply rooted difference between the two is thus their respective epistemic modality and evidentiality.

6.5. Form, meaning and function As discussed in section 5.6. the evidential and/or epistemic meaning of seem and appear, and their pragmatic functions correlate with the syntactic constructions they occur in. Table 6 in section 5.6. shows that seem is more epistemic in a syntactic construction with a that-clause, and its pragmatic function in this construction is predominantly to hedge, to a slightly lesser degree to mark stance. Seem in a catenative construction is more evidential than epistemic, and its function is primarily to hedge, less frequently to mark stance compared to seem and a that-clause. Appear in a syntactic construction with a that-clause is more evidential than epistemic, and functions only as a hedge. Appear in a catenative construction is also far more evidential than epistemic and is mostly used to hedge.

Another observation from Table 6 is that evidentiality correlates more strongly with the catenative construction than with the that-construction. This is true for both verbs. This may imply that when choosing the appropriate expression to express a certain level of evidentiality and epistemic modality, the choice is not only between the verbs seem and appear, the syntactic form should also be considered. For example, on the scale ranging from a high degree of evidentiality on the one side and a high degree of epistemic modality on the other side, appear in sense group 2A (fremstå) may well be the most epistemic choice (5.4.3.), but it may be even more evidential in a syntactic construction with a that-clause. Or, in the same vein, the largest sense group of seem (Group 2S) might be nuanced and made slightly more evidential by choosing the catenative construction over the syntactic construction with a that-clause. If that is the case, the scale (Figure 8.) could be even more nuanced by adding the dimension of syntax.

6.6. Seem and appear are becoming like the modal auxiliaries In the established literature, seem and appear are not part of the grammatical class of modal auxiliaries, nor are they yet considered part of the marginal modal auxiliaries like need (to), dare (to), have to, have got to, used to, used to, be supposed to and be going to, or the other emerging modals like had better, want to, be going to and would rather, that some claim are developing into proper modal auxiliaries through a process of grammaticalization (2.2.).

However, the findings in the present study show that there is some evidence to support the claim that seem and appear are going through a process of grammaticalization and are becoming very much like the modal auxiliaries in form and meaning. There is more evidence for seem becoming grammaticalized into an auxiliary than for appear. These are the results from my study that support the claim:

- The overview of correspondences of seem and appear shows that seem and appear have a large number of correspondences. This is particularly true for seem, which has a much wider set of correspondences, indicating perhaps that the expression is the most semantically bleached and is further along in the grammaticalization process. A wide range of

81

correspondences may suggest that seem and appear are vague or semantically bleached and support the theory that seem and appear are part of an ongoing process of grammaticalization. See Table 1 in section 4.5. - The Semantic Mirrors-analysis shows that both verbs have several sense partitions each, which could be a sign of semantic bleaching. See sections 5.1.1 and 5.2.5. - The high number of zero correspondences (omissions in particular) could be evidence that the meaning of seem and appear has been weakened and that it contributes less to the meaning of the sentence. Omission can therefore indicate loss of lexical meaning, and thus support the hypothesis of grammaticalization. See section 3.1. - The number of non-congruent correspondences like modal particles, modal complements and modal auxiliaries in the Norwegian correspondences testify to the auxiliary-like epistemic modal meaning of seem. See section 5.3.2. - When seem and appear are used as raising verbs (verbs that raise the subject up from a clause in a lower position to become the subject of the matrix clause, see section 3.3.3.) they are auxiliary-like. In these constructions, the roles of seem and appear are primarily to modify the main verb, even though they have a grammatical connection to the subject. The subject has a semantic connection to the verb in the lower clause, and seem or appear become more like function words, modifying the main verb. See section 3.3.3. - Seem and appear share many grammatical properties with the modal auxiliaries: They are used in the same position as the modal auxiliary, i.e. before a non-finite verb form. See section 3.1. - They have some, but not all, of the NICE-properties that are obligatory for the traditional modal auxiliaries. See section 3.1. - They resemble the modal auxiliaries in that they are semantically independent of the subject, i.e. the object of an active sentence can made into the subject of a passive sentence without it influencing the verb phrase. See section 3.7.

It may seem like a contradiction to claim that seem and appear are in the process of developing into modal auxiliaries, yet this study shows that they are rarely translated with a modal auxiliary. Instead, when a modal auxiliary is used as a correspondence of seem or appear, its function is to emphasize the epistemic modality (or evidentiality) already conveyed by seem or appear (5.3.2.). This suggests that seem and appear fill a gap in the range of modal auxiliaries. And it is quite possible that this lexical gap has to do with evidentiality and epistemic modality, which this study has shown is what makes these two verbs special and is the major point of difference between them.

6.7. Conclusion In conclusion, I find that seem and appear are indeed similar in form, meaning and function, and they share many of the same Norwegian correspondences. However, results from the two analyses I have performed, show that although they are similar in many ways, they are also very different from each other in certain contexts, and the verbs themselves and their Norwegian correspondences cannot be used interchangeably as synonyms, or even as near synonyms in all contexts.

My study shows that seem and appear are different with respect to grammar, syntax, semantic meaning and pragmatic function, and that the interaction between form, meaning and function determines the meaning and function of the two verbs. The most important differences are highlighted in the present chapter.

82

Appear is more evidential than appear seem and refers to visual evidence far more frequently than seem. Appear is also used to signal inferences and has a sense partition that is strongly related to deduced evidence (fremstå), implying epistemic assessment based on an observation. Seem is more epistemic than appear, made evident by the choice of Norwegian correspondences (for example the high frequency of Norwegian modal particles and modal adjuncts) and its reference to inferred evidence based on logical reasoning.

Seem and appear have different pragmatic functions that reflect their epistemicity and evidentiality. This is especially obvious when seem and appear are hedges and function as boosted modifiers. In this capacity, appear tends to increase the focus on the evidence, thereby reducing uncertainty and doubt, whereas seem emphasizes the speaker's epistemic assessment and strengthens the force or the truth of the proposition. The same is true for seem and appear as markers of epistemic stance: Appear signals distance to the propositional content more often than seem, which is a consequence of attested evidence being more objective, while seem signals epistemic assessment and personal judgement, i.e. stronger ownership to the propositional content. The epistemic qualities of seem may also be a reason for its function as a face-saving or politeness marker, which is unique to seem. The present chapter concludes that seem is more evidential in catenative constructions than in constructions with a that-clause. Appear is evidential in both catenative constructions and constructions with a that-clause. Inferences, in which evidentiality and epistemic modality co-occur, are far more frequent with appear plus a that-clause than appear as a catenative, indicating perhaps that appear in combination with a that-clause may have a strong epistemic component as well.

Finally, the present chapter summarizes all the evidence the two analyses have found to support the claim that seem and appear are going through a process of semantic bleaching and grammaticalization and are becoming very similar to the semi-modals and the emerging modal auxiliaries in English, possibly filling a lexical gap in the range of the modal auxiliaries pertaining to evidential meaning.

83

7. Summary In this thesis, I use translation data from English-Norwegian Parallel Corpus (ENPC) and two different methods to show that even though seem and appear are very similar and have many of the same correspondences, they differ with respect to grammar, syntax, semantic meaning and pragmatic function. I also show that the interaction between form, meaning and function determines the meaning and function of the two verbs.

The concepts of modality, epistemic modality and evidentiality and the relationship between hem are central to my study, and the theoretical framework for the thesis takes as its starting point the fact that in the established literature, modality and evidentiality are defined and discussed as grammatical concepts, semantic concepts and pragmatic concepts. In my thesis I attempt to describe different approaches to the concepts and the demarcations between them. I have chosen a theoretical framework that allows for a discussion of epistemic modality and evidentiality in terms of grammar, syntax, semantics and pragmatics. These are the levels of analysis in this thesis. The analysis of seem and appear depends on factors from all these levels: grammatical properties, such as the syntactic constructions seem and appear can occur in, evidential and epistemic meaning and how they in turn depend on source of evidence, mode of knowing and reliability, and pragmatic factors such as subjectivity and intersubjectivity, stance, hedging and politeness. The chosen approach is to regard epistemic modality and evidentiality as two different categories, with overlapping functions and meanings in inferences.

The present study uses the method of Semantic Mirrors, to determine that seem has three distinct sense partitions: one with the semantic meaning of pretending or acting a certain way, one of being similar to something based on visual evidence, and a third sense which is to make a personal judgement that something is like something or feels like something. Appear has three distinct semantic senses as well. The most frequently used is a subset of seem's sense of looking similar. The other two are possible outliers, one is the sense of being presented as similar, the other has the meaning that something has been deduced.

The second analysis is a contrastive analysis of seem and appear and their correspondences. The analysis provides evidence that appear is more evidential than seem, and that appear usually refers to visible and physical existence, and vice versa: that seem is more epistemic than appear conveying subjective judgement and no evidence, or conveying subjective judgement of some evidence to a greater extent than appear. The contrastive analysis also shows that while appear is almost exclusively evidential, seem can be either evidential, epistemic, or both, i.e. in inferences. In inferences where evidentiality and epistemic modality co-exist, the question of how modal or how evidential an expression is, becomes a matter of degrees.

Findings from the contrastive analysis also confirm that seem and appear have similar pragmatic functions, and that they are most commonly used to hedge. But the two verbs also differ with respect to pragmatic functions, and when they do, the pragmatic function typically reflects the evidential or epistemic qualities of appear and seem respectively. For example, only seem is used as politeness markers, because only seem has that face-saving effect of the speaker assuming responsibility for the statement. Seem signals loose interpretations and epistemic stance in the form of closeness to the proposition more often than appear, which also testifies to the epistemic qualities

84 of seem. Appear signals distance to the propositional content more often than seem, which is a consequence of its reference to attested evidence, which is objective.

A three-way analysis of the variables syntactic form, sematic meaning and pragmatic function demonstrates that seem is more evidential in catenative constructions than in constructions with a that-clause. Appear is evidential in both catenative constructions and constructions with a that- clause. Moreover, the findings suggest that generally, catenative constructions co-occur more frequently with direct and indirect evidence, whereas the construction with a that-clause is associated with epistemic assessment. This opens up a possibility of a new set of combinations with a more (or less) evidential expression with a more (or less) evidential or epistemic form. This is a line of research I think should be pursued further: In this thesis I have only looked at two syntactic constructions (catenatives and constructions with a that-clause). It would be very interesting to study seem and appear in other syntactic constructions, and analyse the interconnectivity between syntax, evidential and epistemic meanings and pragmatic functions further. I also think a study with more fine-grained variables for pragmatic functions and a systematic link to the various correspondences could be productive – I fear I have only touched upon some of the most obvious differences with respect to pragmatic function, and with the design of the present study, I have not been able to demonstrate the need for nuance and variety in translation strategies that I believe are necessary in order to translate well the exact meaning of the two verbs in the various syntactic constructions.

Finally, the study provides cautious support for the claim that seem and appear are going through a process of semantic bleaching and grammaticalization and are becoming very similar to the semi- modals and the emerging modal auxiliaries in English, possibly filling a lexical gap in the range of the modal auxiliaries pertaining to evidential meaning.

85

8. References Aijmer, Karin. 2009. "Seem and Evidentiality." In Functions of Language 16, no. 1, 63-88.

Aijmer, Karin. 2019. "Chapter 21. Modality and Mood in Functional Linguistic Approaches." In Oxford Handbook of Modality and Mood, edited by Jan Nuyts and Johan van der Auwera. Oxford: Oxford University Press, 495 – 513.

Altenberg, Bengt. 1999. "Adverbial Connectors in English and Swedish: Semantics and Lexical Correspondences." In Out of corpora. Studies in Honour of Stig Johansson, edited by Signe Oksefjell and Hilde Hasselgård, 249-268. Amsterdam: Rodopi B.V.

Altenberg, Bengt and Sylviane Granger. 2002. "Recent Trends in Cross-Linguistic Lexical Studies." In Lexis in Contrast. Corpus-Based Approaches, edited by Bengt Altenberg and Sylviane Granger, 3-47. John Benjamins Publishing Company.

Almeida, Francisco Alonso. 2015. "The functions of seem and parecer in early medical writing". In Discourse Studies, Vol. 17, No. 2, Special issue: Evidential and epistemic strategies in discourse, 121- 140. Sage Publications, Ltd. https://www.jstor.org/stable/24815149 Accessed 13 July 2019.

Andvik, Erik E. 1992. A Pragmatic Analysis of the Norwegian Modal Particle. SIL International Publications.

Baker, Mona. 1992. In other words. A course book of translation. London &New York: Routledge.

Baker, Mona and Gill Francis, John Sinclair and Elena Tognini-Bonelli.1993. Text and Technology: In Honour of John Sinclair. Philadelphia: J. Benjamins.

Biber et al, 1999. The Longman Grammar of Spoken and Written English. London: Longman.

Bokmålsordboka og Nynorskordboka. Online edition. Språkrådet og Universitetet i Bergen. www.ordbok.uib.no/. Accessed 11 November 2019.

Brugman, Claudia M. and Monica Macaulay. 2015. "Characterizing evidentiality". In Linguistic Typology 2+15; 19 (2); 201 – 237. De Gruyter Moutun. Accessed Online 17 October 2018.

Bybee, Joan, Revere Perkins and William Pagliuca. 1994. The Evolution of Grammar. Tense, Aspect and Modality in the Languages of the World. Chicago: The University of Chicago Press.

Chafe, Wallace and Johanna Nichols (eds). 1986. Evidentiality. The Linguistic Coding of Epistemology. Advances in Discourse processes, vol xx. New Jersey: Ablex Publishing Corporation.

Cornillie, Bernt and Paola Pietranda. 2012. "Modality at work. Cognitive, Interactional and Textual Function of Modal Markers." Journal of Pragmatics 44, no 15, 21099-2115. Accessed Online 30 March 2019.

Coates, Jennifer. 2013. "The role of epistemic modality in women's talk" in Modality in Contemporary English edited by Roberta Facchinetti, Manfred Krug and Frank Palmer. Berlin/New York: Mouton de Gruyter, 332-348.

Dyvik, Helge Julius. 1998. "A Translational Basis for Semantics." In Corpora and Cross-linguistic Research. Theory, Method and Case Studies, edited by Stig Johansson and Signe Oksefjell. Amsterdam: Rodolophi, 51-86.

86

Dyvik, Helge Julius. 2004. "Translations as Semantic Mirrors: From Parallel Corpus to Wordnet". In Language and Computers, vol. 49. https://folk.uib.no/hfohd/SLF/Dyvik/ICAMEpaper.pdf Accessed 12 September 2018.

Dyvik, Helge Julius. 2011. "Semantic Mirrors". In Common Language Resources and their Application (CLARA), University of Bergen. https://clara.w.uib.no/files/2011/06/semmirrors.pdf Accessed 13 September 2018.

Egan, Andy, and Brian Weatherson. 2011. Epistemic Modality. Oxford: Oxford University Press.

Ebeling, Signe Oksefjell and Jarle Ebeling. 2013. "Chapter 2: Contrastive analysis." In Patterns in Contrast, 13-33. John Benjamins.

Fagerlund, Martin. 2010. Computing Word Senses by Semantic Mirroring and Spectral Graph Partitioning. Exam paper at Department of Mathematics, University of Linköping. http://liu.diva- portal.org/smash/get/diva2:330160/FULLTEXT01.pdf Accessed 11 July 2019.

Facchinetti, Roberta, Manfred Krug and Frank Palmer (eds). 2003. Modality in Contemporary English. Topics in English Linguistics, vol. 44. Berlin: Mouton de Gruyter.

Fraser, Bruce. 2010. "Pragmatic Competence: The case of hedging." In New approaches to hedging. Studies in Pragmatics 9, edited by Kaltenböck et al. Bingley: Emerald Group Publishing.

Gries, Stefan Thomas. 2009. "Chapter 2: The Three Central Corpus Linguistic methods". In Quantitative Corpus Linguistics with R. A Practical Introduction, 7-18. New York: Routledge.

Grønbeck, Silje Kathrin, 2018. Egentlig, actually. A corpus-based study of the translation of Norwegian egentlig and its English correspondences. Master's Thesis at University of Oslo.

Haddington, Pentti. 2004. Stance Taking in News Interviews. SKY Journal of Linguistics 17. http://www.ling.helsinki.fi/sky/julkaisut/SKY2004/Haddington.pdf?q=stance Accessed online 23 June 2019.

Halliday, M.A.K. & Matthiessen, C.M.I.M., 2004. An introduction to functional grammar. 3rd ed., London: Arnold.

Hasselgård, Hilde. 2007. "Using the ENPC and the ESPC as a Parallel Translation Corpus: Adverbs of Frequency and Usuality". Nordic Journal of English Studies Vol.6(1).

Hasselgård, Hilde, Stig Johansson and Per Lysvåg. 2014. English Grammar: Theory and use. Oslo: Universitetsforlaget.

Hasselgård, Hilde. 2018. Lecture notes. Corpus Linguistics. University of Oslo.

HU, Guangwei and Cao Feng. 2011. "Hedging and boosting in abstracts of applied linguistics articles: A comparative study of English- and Chinese-medium journals". In Journal of Pragmatics no 43. 2795-2809. https://www.sciencedirect.com/science/article/abs/pii/S0378216611001135 Accessed 10 July 2019.

Huddleston, Rodney, and Geoffrey K. Pullum. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press, 2002.

Ivir, Vladimir. 1981. “Formal Correspondence vs. Translation Equivalence Revisited.” In Poetics Today, vol. 2, no. 4, 51–59. www.jstor.org/stable/1772485.

Kastrone, Inga. 2008. "Need to and the Modality of Obligation: A Corpus-based Approach". Master's thesis, University of Oslo.

87

Kranich, Svenja. 2016. Contrastive Pragmatics and Translation. Evaluation, Epistemic Modality and Communicative Styles in English and German. Amsterdam: John Benjamins Publishing Company.

Kranich, Svenja. 2011. "To Hedge or Not to Hedge: The Use of Epistemic Modal Expressions in Popular Science in English Texts, English-German Translations, and German Original Texts." In Text & Talk: An Interdisciplinary Journal of Language, Discourse & Communication Studies 31, no. 1, 77-99.

Krug, Manfred G. 2000. Emerging English Modals: A Corpus-Based Study of Grammaticalization. Berlin: Mouton de Gruyter

Johansson, Stig and Berit Løken. 1997. "Some Norwegian Discourse Particles and their English Correspondences". In Sounds, Structures and Senses. Essays presented to Davidsen Nielsen on the Occasion of his sixtieth Birthday, edited by Carl Bache and Alex Klinge, 149-170. Odense: Odense University Press.

Johansson, Stig. 1999. "On the role of Corpora in Cross-Linguistic research" in Corpora and Cross- linguistic research. Theory, Method and Case studies, edited by Stig Johansson and Signe Oksefjell. Rodopi B. V. Editions.

Johansson, Stig. 2001. "The English verb seem and its correspondences in Norwegian: What seems to be the problem?" In Karin Aijmer (ed.), A wealth of English: Studies in honour of Göran Kjellmer (Gothenburg Studies in English 81), 221–245. Göteborg: Acta Universitatis Gothoburgensis.

Johansson, Stig. 2007. Seeing Through Multilingual Corpora. On the Use of Corpora in Contrastive Studies. Amsterdam/Philadelphia: John Benjamins Publishing Company.

Kaltenböck, Gunther, Wiltrud Mihatsch and Stefan Schneider (eds), 2010. New approaches to hedging. Studies in Pragmatics 9. Bingley: Emerald Group Publishing.

Kärkkäinen, Elise. 2003. Epistemic stance in English Conversation. A description of its internal functions, with focus on I think. Pragmatics and Beyond Series, vol. 115. Amsterdam: John Benjamins Publishing Company.

Ketzer, Anita. 2010. "Hedges in Context: Form and function of sort of and kind of. In New approaches to Hedging, edited by Gunther Kaltenböck, Wiltrud Mihatsch and Stefan Schneider. Bingley: Emerald Group Publishing.

Kranich, Svenja. 2011. "To Hedge or Not to Hedge: The Use of Epistemic Modal Expressions in Popular Science in English Texts, English–German Translations, and German Original Texts." In Text & Talk - An Interdisciplinary Journal of Language, Discourse & Communication Studies 31, no. 1, 77-99.

Krug, Manfred G, 2000. Emerging English Modals: A corpus-based Study of Grammaticalization. Berlin/New York: Mouton de Gruyter.

Leech, Geoffrey, 2003: "Modality on the move: the English modal auxiliaries 1961-1992" in Modality in Contemporary English edited by Roberta Facchinetti, Manfred Krug and Frank Palmer. Berlin/New York: Mouton de Gruyter, 223-240.

Lewis, Diana M. 2015. "A comparable-corpus based approach to the expression of obligation across English and French" in Nordic Journal of English studies 14 (1), 152-173

Lid, Kristian. 2010. "Future-referring Expressions in English and Norwegian: A Contrastive Study Based on the English-Norwegian Parallel Corpus." Master's thesis, University of Oslo.

88

Løken, Berit H. 2007. "Beyond Modals: A Corpus-based Study of English and Norwegian Expressions of Possibility." No. 296: X, 335. PhD dissertation, University of Oslo.

Løken, Berit H. 1997. "Expressing Possibility in English and Norwegian: Can, Could, May, Might and Kunne and Their Correspondences." Master's thesis, University of Oslo.

Merriam-Webster Online Dictionary. 2015. https://www.merriam-webster.com Accessed 20 October 2019.

McArthur, Tom, Jacqueline Lam-McArthur, and Lise Fontaine (eds). 2018. The Oxford Companion to the English Language (2 ed.) Oxford University Press. Online edition. Accessed 29. December 2018.

McEnery, Tony and Andrew Hardie. 2012. Corpus Linguistics. Methods, Theory and Practice. Cambridge: Cambridge University Press.

Mkhitaryan, Yelena and Sona Tumanyan. 2015. "On Differences in the Use of Hedging in English and Armenian Academic Discourse". In Procedia - Social and Behavioral Sciences, volume 197, 2506-2511. Accessed online 12 July 2019. https://www.sciencedirect.com/science/article/pii/S1877042815043256?via%3Dihub

Nordquist, Richard. 2018. "What is a Catenative Verb." https://www.thoughtco.com/what-is- catenative-verb-1689832. Accessed 20 October 2019.

Nuyts, Jan. 2019. "Chapter 3. Analysis of the Modal Meanings." In Oxford Handbook of Modality and Mood, edited by Jan Nuyts and Johan van der Auwera. Oxford: Oxford University Press, 33-49.

Nuyts, Jan and Johan van der Auwera (eds). 2019. Oxford Handbook of Modality and Mood. Oxford: Oxford University Press.

Oxford English Dictionary (online edition). www.oed.

Palmer, Frank, 2001. Mood and Modality, 2nd ed. Cambridge: Cambridge University Press.

Perkins, Michael R. 1983. Modal Expressions in English. London: Frances Pinter Publishers.

Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1976. A Grammar of Contemporary English. 6th. ed. London: Longman.

Ruzaité, Juraté. 2006. Vague language in academic discourse: approximators and Quantifiers in British and American English. Doctorate Thesis at Department of English at the University of Bergen. Bergen: University of Bergen.

Soliene, Audrone. 2010. "Choice of Strategies in Realizations of Epistemic Possibility in English and Lithuanian. A Corpus-based Study." International Journal of Corpus Linguistics 15, no. 2: 291-316.

Squartini, Mario. 2019. "Chapter 4: Interactions between modality and other semantic categories". In Oxford Handbook of Modality and Mood, edited by Jan Nuyts and Johan van der Auwera. Oxford: Oxford University Press, 50-67.

Thompson, Geoff and Susan Hunston. 2007. System and Corpus: Exploring Connections. Functional Linguistics. London: Equinox.

Vold, Eva Thue. 2006. "Epistemic Modality Markers in Research Articles: A Cross‐linguistic and Cross disciplinary Study." International Journal of Applied Linguistics 16, no. 1, 61-87.

89

Online sources https://dictionary.cambridge.org/ Accessed 20 October 2019. https://www.grammaring.com/ Accessed 20 October 2019. https://grammar.yourdictionary.com/ Accessed 20 October 2019.

Primary data The English-Norwegian Parallel Corpus. Part of Oslo Multilingual Corpus. 1999-2008. The Faculty of Humanities, University of Oslo. The Oslo Multilingual Corpus is a product of the interdisciplinary research project Languages in Contrast (SPRIK), directed by Stig Johansson and Cathrine Fabricius- Hansen, and compiled by the OMC corpus team. http://www.hf.uio.no/ilos/english/services/omc/

90