In her thought-provoking article, Aneta Pavlenko approaches emotion and emotion-laden words in the bilingual lexicon from an impressive number of different perspectives. This is particularly welcome, since most models of linguistic structure do not account for emotional meanings in a systematic way. One exception worth mentioning, however, is Systemic Functional Linguistics (SFL; Halliday, Reference Halliday1985/1994) with its broad division of functions into the textual, ideational and interpersonal metafunctions. Whereas the textual metafunction is concerned with the flow of information and the ideational with the construction of experience, the interpersonal function is concerned with interaction between people including the sharing of feelings. Recently, this last aspect has been developed more on the basis of SFL within Appraisal theory (Martin and White Reference Martin and White2005), where attitudinal meanings are divided into Affect (the construal of emotion into the frame I feel (very) __), Judgment (social/moral evaluation of people) and Appreciation (aesthetic evaluation of things). The theory has primarily been applied to the development of the expression of feeling and evaluation in student writing in schools, based on qualitative analysis of corpora. The framework has also recently been applied to other languages than English as in Folkeryd (Reference Folkeryd2006), who shows how the expression of affect and other evaluative resources are used in various story genres by Swedish students in grades 5, 8 and 11. The expression of attitudinal meanings turned out to be a problem characteristic of low-achieving students. This study includes bilingual students but they are not focused on in the published work.
One perspective that is rather weakly represented in Pavlenko's article is various types of corpus-based approaches to contrastive and typological analysis of the lexicon. To be sure, the emotional dimension has been rather neglected in such studies but in my opinion corpus-based approaches would enrich the methodology. One of the problems discussed by Pavlenko is the differences between languages with respect to the superordinate term “emotion”, for example, the word émotion in French (Niedenthal et al., Reference Niedenthal, Auxiette, Nugier, Dalle, Bonin and Fayol2004). The verb meaning “feel” which is used to define emotion in several approaches is equally problematic. In Viberg (Reference Viberg2005), the Swedish verb känna “feel” is compared to its closest correspondents in English, German, French and Finnish based on data from a translation corpus. Like English feel, for example, känna can be used with reference to internal perception comprising basic emotions such as happiness and anger, social emotions such as shame and envy as well as with reference to bodily sensations such as hunger and fatigue. With respect to external perception, both verbs basically refer to touch but Swedish känna also refers to the experience of taste (känna smak) and smell (känna lukt). In addition, Swedish känna has several cognitive meanings which are not shared with feel. As shown in an earlier typological study of the verbs of perception (Viberg, Reference Viberg1984), the link between emotion and touch that is characteristic of both feel and känna is characteristic primarily of a number of European languages. In several African languages, for example Setswana, the perception verb that refers to the experience of emotion and touch basically refers to hearing. In Swahili, the verb referring to visual experience (ona “see”) can also be used to refer to the experience of emotions.
Emotion is to a great extent a hidden dimension of the lexicon. For example, the verbs of physical contact such as hit, strike, slap, kick primarily describe the motion of a concrete object (often a body part) that comes into contact with another concrete object (often, but not necessarily, the body of a human being). In this respect, physical contact verbs describe concrete events that can be directly observed with the eyes. However, in many contexts, a simple utterance such as Bill hit Ben strongly motivates the inference “Bill hurt Ben”, a more abstract, psychological meaning. In a study of the Swedish verbs of physical contact (Viberg, Reference Viberg, Allwood and Gärdenfors1999, Reference Viberg, Aijmer and Altenberg2004), an important function of such verbs, especially in novels, turned out to be to express emotional attitudes between people. For some verbs like kyssa “kiss” and smeka “caress”, an emotional component such as “show tenderness” could be said to be completely conventionalized as a part of the lexical meaning. A more interesting case is verbs like stryka “stroke, run the hand along something” that is freely used with both human and inanimate objects. In novels, however, examples such as the following one are so frequent that they almost feel like hackneyed phrases: Hon lutade sig ner och strök honom över huvudet “She bent down and stroked his head” (literally: “She . . . stroked him over head-the”). Concretely, this describes the motion of a hand (which is understood) along a head but the major motivation for including this event in the narration is to describe the display of affection. Most physical contact verbs are related to emotions (or sensations such as pain) along a continuum from weak to strong invited inference to a conventionalized part of the lexical representation. It is also common, especially for the verb meaning “hit, strike” to have extended meanings with an emotional loading. An example like the following one implies an element of surprise: It/The idea struck me that I could be wrong. Especially in the passive, hit and strike can take subjects that describe violent and unpleasant events, e.g. The country was hit by a depression/a cyclone/famine, etc.
Among verbs, what is broadly referred to as manner in many cases can include both inferences about the affective state of the agent and evaluative judgements by the speaker. There are many examples of that among the manner of speaking verbs such as moan which is defined by the Longman dictionary of contemporary English (moan 2. BrE Informal; 3rd edition, 2001) “to complain in an annoying way, especially in an unhappy voice and without good reason”. One of the examples is: He's always moaning that we use too much electricity. The modifications “in an annoying way” and “without good reason” represent (current) speaker evaluations, whereas “in an unhappy voice” is more of an objective description. The verb “complain” can be unpacked as “say that one is dissatisfied with something” and refers to the feelings of the subject of the verb moan (the speaker being reported about in distinction to the current speaker). Verbs with a high degree of emotional or evaluative loading are found at the bottom of the hierarchies of troponyms in semantic dictionaries of the wordnet type (see Global WordNet, http://www.globalwordnet.org/). For example, verbs meaning “talk”, “walk” and “look” have many such troponyms in several European languages. There is, however, crosslinguistic variation on this point. As is well known by now, languages differ typologically with respect to the degree to which manner of motion is encoded in verbs. As Slobin (Reference Slobin, Gentner and Goldin-Meadow2003) has shown, more elaborate and vivid descriptions of motion will also increase the number of evaluative conclusions that are drawn in reception.
Within translation studies, similar aspects of word meaning have been referred to as verb descriptivity. According to Snell-Hornby (Reference Snell-Hornby1983, p. 43) a descriptive verb can be divided into a core of verbal action and a descriptive modifying complex which can be paraphrased with one or more adjectives or manner adverbs. The modification can apply either to the verbal core (direct descriptivity) or characterize one or more of the participants or circumstances (indirect descriptivity). It is the latter that is particularly interesting in this context since it typically expresses a speaker-evaluation that is gradable in terms of some norm set by the speaker or the social environment. A verb like stalk has “walk” as its verbal core, whereas “in a stiff or erect way” objectively characterizes the manner of walking (direct descriptivity) and “with affective dignity” represents the speaker's emotion-laden subjective evaluation (i. e. indirect descriptivity). Snell-Hornby presents an extensive contrastive comparison of around 600 English and around 500 German descriptive verbs classified into a number of semantic fields grouped into four major classes Human Behaviour, Movement and Position, Sounds, and Light and Facial Expression. Verbs that appear to be rather close equivalents across languages may turn out to contrast with respect to emotional aspects of meaning. For example, English stare in addition to directly describing the act (“look hard”) also includes an indirect description of the speaker's negative value judgment (“in a stupid or impertinently curious way”). According to Snell-Hornby (Reference Snell-Hornby1983, p. 29), the German correspondent starren only shares the objective part of the description and does not include any value judgment.
Within corpus-based analysis, attention has been drawn to the tendency for certain words to appear in contexts that span a stretch of words that share a certain value. This phenomenon is referred to as semantic prosody (see McEnery, Xiao & Tono, Reference McEnery, Xiao and Tono2006, pp. 83f. for a recent overview) or discourse prosody (Stubbs, Reference Stubbs2001, pp. 105–108). Typically, such prosodies have a negative value but they can also be positive. One of Stubbs's major examples is the verb cause which is typical as it has a negative prosody and tends to collocate with words such as accident, concern, damage, death and trouble. It can be compared with provide, a word with a positive prosody, which typically collocates with nouns referring to care, food, help, money and shelter. It is especially remarkable that cause which is often used as a semantic primitive with neutral value should be so emotion-laden. The value is partly conventionalized since an expression like causing work usually has a negative value whereas providing work is positive.
The expression of emotion is a fundamental part of human communication. This is reflected also in the lexical resources which are the major focus in this commentary. However, basic emotion words such as “happy” and “angry” to a great extent are used to describe and not to express emotions. In a more indirect but perhaps more important way, emotional meaning is interwoven with other aspects of meaning in the semantic representation of words and larger units such as idioms. Finer shades of descriptive meaning in Snell-Hornby's sense appear to be acquired rather slowly both in first and second language acquisition. This is reflected in a longitudinal study (Viberg Reference Viberg, Verhoeven and Strömqvist2001) of children acquiring Swedish L2 compared to a group of age-mates with Swedish as L1. One important source of data consisted in oral renarrations of video-clips. There was a general trend in the data (similar to other studies) of an age-related development towards the emergence of an evaluative stance in the renarrations and towards increased lexical variability and idiomaticity, which could most clearly be seen in the L1 children. The L2 children first used simplifications and overextensions, and then, after finding a correct basic expression, slowly introduced more expressive lexical alternatives. One of the reasons for this slow development is that the mastery of those parts of the vocabulary that are loaded with emotion and evaluation involves a very large number of lexical items with low token frequency and very fine-grained meaning distinctions. Just inserting a large number of words simply meaning “good” or “bad” would not have the same effect.
To sum up, corpus-based analysis could be used to as a complement to the impressive range of experimental techniques described by Pavlenko. The experimental studies tend to use isolated words, which misses the prosodic nature of emotionality. There is also a need to include lexical items where emotion represents a hidden dimension of meaning.
You have
Access
In her thought-provoking article, Aneta Pavlenko approaches emotion and emotion-laden words in the bilingual lexicon from an impressive number of different perspectives. This is particularly welcome, since most models of linguistic structure do not account for emotional meanings in a systematic way. One exception worth mentioning, however, is Systemic Functional Linguistics (SFL; Halliday, Reference Halliday1985/1994) with its broad division of functions into the textual, ideational and interpersonal metafunctions. Whereas the textual metafunction is concerned with the flow of information and the ideational with the construction of experience, the interpersonal function is concerned with interaction between people including the sharing of feelings. Recently, this last aspect has been developed more on the basis of SFL within Appraisal theory (Martin and White Reference Martin and White2005), where attitudinal meanings are divided into Affect (the construal of emotion into the frame I feel (very) __), Judgment (social/moral evaluation of people) and Appreciation (aesthetic evaluation of things). The theory has primarily been applied to the development of the expression of feeling and evaluation in student writing in schools, based on qualitative analysis of corpora. The framework has also recently been applied to other languages than English as in Folkeryd (Reference Folkeryd2006), who shows how the expression of affect and other evaluative resources are used in various story genres by Swedish students in grades 5, 8 and 11. The expression of attitudinal meanings turned out to be a problem characteristic of low-achieving students. This study includes bilingual students but they are not focused on in the published work.
One perspective that is rather weakly represented in Pavlenko's article is various types of corpus-based approaches to contrastive and typological analysis of the lexicon. To be sure, the emotional dimension has been rather neglected in such studies but in my opinion corpus-based approaches would enrich the methodology. One of the problems discussed by Pavlenko is the differences between languages with respect to the superordinate term “emotion”, for example, the word émotion in French (Niedenthal et al., Reference Niedenthal, Auxiette, Nugier, Dalle, Bonin and Fayol2004). The verb meaning “feel” which is used to define emotion in several approaches is equally problematic. In Viberg (Reference Viberg2005), the Swedish verb känna “feel” is compared to its closest correspondents in English, German, French and Finnish based on data from a translation corpus. Like English feel, for example, känna can be used with reference to internal perception comprising basic emotions such as happiness and anger, social emotions such as shame and envy as well as with reference to bodily sensations such as hunger and fatigue. With respect to external perception, both verbs basically refer to touch but Swedish känna also refers to the experience of taste (känna smak) and smell (känna lukt). In addition, Swedish känna has several cognitive meanings which are not shared with feel. As shown in an earlier typological study of the verbs of perception (Viberg, Reference Viberg1984), the link between emotion and touch that is characteristic of both feel and känna is characteristic primarily of a number of European languages. In several African languages, for example Setswana, the perception verb that refers to the experience of emotion and touch basically refers to hearing. In Swahili, the verb referring to visual experience (ona “see”) can also be used to refer to the experience of emotions.
Emotion is to a great extent a hidden dimension of the lexicon. For example, the verbs of physical contact such as hit, strike, slap, kick primarily describe the motion of a concrete object (often a body part) that comes into contact with another concrete object (often, but not necessarily, the body of a human being). In this respect, physical contact verbs describe concrete events that can be directly observed with the eyes. However, in many contexts, a simple utterance such as Bill hit Ben strongly motivates the inference “Bill hurt Ben”, a more abstract, psychological meaning. In a study of the Swedish verbs of physical contact (Viberg, Reference Viberg, Allwood and Gärdenfors1999, Reference Viberg, Aijmer and Altenberg2004), an important function of such verbs, especially in novels, turned out to be to express emotional attitudes between people. For some verbs like kyssa “kiss” and smeka “caress”, an emotional component such as “show tenderness” could be said to be completely conventionalized as a part of the lexical meaning. A more interesting case is verbs like stryka “stroke, run the hand along something” that is freely used with both human and inanimate objects. In novels, however, examples such as the following one are so frequent that they almost feel like hackneyed phrases: Hon lutade sig ner och strök honom över huvudet “She bent down and stroked his head” (literally: “She . . . stroked him over head-the”). Concretely, this describes the motion of a hand (which is understood) along a head but the major motivation for including this event in the narration is to describe the display of affection. Most physical contact verbs are related to emotions (or sensations such as pain) along a continuum from weak to strong invited inference to a conventionalized part of the lexical representation. It is also common, especially for the verb meaning “hit, strike” to have extended meanings with an emotional loading. An example like the following one implies an element of surprise: It/The idea struck me that I could be wrong. Especially in the passive, hit and strike can take subjects that describe violent and unpleasant events, e.g. The country was hit by a depression/a cyclone/famine, etc.
Among verbs, what is broadly referred to as manner in many cases can include both inferences about the affective state of the agent and evaluative judgements by the speaker. There are many examples of that among the manner of speaking verbs such as moan which is defined by the Longman dictionary of contemporary English (moan 2. BrE Informal; 3rd edition, 2001) “to complain in an annoying way, especially in an unhappy voice and without good reason”. One of the examples is: He's always moaning that we use too much electricity. The modifications “in an annoying way” and “without good reason” represent (current) speaker evaluations, whereas “in an unhappy voice” is more of an objective description. The verb “complain” can be unpacked as “say that one is dissatisfied with something” and refers to the feelings of the subject of the verb moan (the speaker being reported about in distinction to the current speaker). Verbs with a high degree of emotional or evaluative loading are found at the bottom of the hierarchies of troponyms in semantic dictionaries of the wordnet type (see Global WordNet, http://www.globalwordnet.org/). For example, verbs meaning “talk”, “walk” and “look” have many such troponyms in several European languages. There is, however, crosslinguistic variation on this point. As is well known by now, languages differ typologically with respect to the degree to which manner of motion is encoded in verbs. As Slobin (Reference Slobin, Gentner and Goldin-Meadow2003) has shown, more elaborate and vivid descriptions of motion will also increase the number of evaluative conclusions that are drawn in reception.
Within translation studies, similar aspects of word meaning have been referred to as verb descriptivity. According to Snell-Hornby (Reference Snell-Hornby1983, p. 43) a descriptive verb can be divided into a core of verbal action and a descriptive modifying complex which can be paraphrased with one or more adjectives or manner adverbs. The modification can apply either to the verbal core (direct descriptivity) or characterize one or more of the participants or circumstances (indirect descriptivity). It is the latter that is particularly interesting in this context since it typically expresses a speaker-evaluation that is gradable in terms of some norm set by the speaker or the social environment. A verb like stalk has “walk” as its verbal core, whereas “in a stiff or erect way” objectively characterizes the manner of walking (direct descriptivity) and “with affective dignity” represents the speaker's emotion-laden subjective evaluation (i. e. indirect descriptivity). Snell-Hornby presents an extensive contrastive comparison of around 600 English and around 500 German descriptive verbs classified into a number of semantic fields grouped into four major classes Human Behaviour, Movement and Position, Sounds, and Light and Facial Expression. Verbs that appear to be rather close equivalents across languages may turn out to contrast with respect to emotional aspects of meaning. For example, English stare in addition to directly describing the act (“look hard”) also includes an indirect description of the speaker's negative value judgment (“in a stupid or impertinently curious way”). According to Snell-Hornby (Reference Snell-Hornby1983, p. 29), the German correspondent starren only shares the objective part of the description and does not include any value judgment.
Within corpus-based analysis, attention has been drawn to the tendency for certain words to appear in contexts that span a stretch of words that share a certain value. This phenomenon is referred to as semantic prosody (see McEnery, Xiao & Tono, Reference McEnery, Xiao and Tono2006, pp. 83f. for a recent overview) or discourse prosody (Stubbs, Reference Stubbs2001, pp. 105–108). Typically, such prosodies have a negative value but they can also be positive. One of Stubbs's major examples is the verb cause which is typical as it has a negative prosody and tends to collocate with words such as accident, concern, damage, death and trouble. It can be compared with provide, a word with a positive prosody, which typically collocates with nouns referring to care, food, help, money and shelter. It is especially remarkable that cause which is often used as a semantic primitive with neutral value should be so emotion-laden. The value is partly conventionalized since an expression like causing work usually has a negative value whereas providing work is positive.
The expression of emotion is a fundamental part of human communication. This is reflected also in the lexical resources which are the major focus in this commentary. However, basic emotion words such as “happy” and “angry” to a great extent are used to describe and not to express emotions. In a more indirect but perhaps more important way, emotional meaning is interwoven with other aspects of meaning in the semantic representation of words and larger units such as idioms. Finer shades of descriptive meaning in Snell-Hornby's sense appear to be acquired rather slowly both in first and second language acquisition. This is reflected in a longitudinal study (Viberg Reference Viberg, Verhoeven and Strömqvist2001) of children acquiring Swedish L2 compared to a group of age-mates with Swedish as L1. One important source of data consisted in oral renarrations of video-clips. There was a general trend in the data (similar to other studies) of an age-related development towards the emergence of an evaluative stance in the renarrations and towards increased lexical variability and idiomaticity, which could most clearly be seen in the L1 children. The L2 children first used simplifications and overextensions, and then, after finding a correct basic expression, slowly introduced more expressive lexical alternatives. One of the reasons for this slow development is that the mastery of those parts of the vocabulary that are loaded with emotion and evaluation involves a very large number of lexical items with low token frequency and very fine-grained meaning distinctions. Just inserting a large number of words simply meaning “good” or “bad” would not have the same effect.
To sum up, corpus-based analysis could be used to as a complement to the impressive range of experimental techniques described by Pavlenko. The experimental studies tend to use isolated words, which misses the prosodic nature of emotionality. There is also a need to include lexical items where emotion represents a hidden dimension of meaning.