Versionen vergleichen von : A Survey on Sentiment and Emotion Analysis for Computational Literary Studies

2 2
3 Evgeny Kim 4 Evgeny Kim
4 Kontakt: Universität Stuttgart, Institut für Maschinelle Sprachverarbeitung GND: 1193672481ORCID: 0000-0001-6822-6709 5 Kontakt: Universität Stuttgart, Institut für Maschinelle
6 Sprachverarbeitung GND: 1193672481ORCID: 0000-0001-6822-6709
5 Roman Klinger 7 Roman Klinger
6 Kontakt: roman.klinger@ims.uni-stuttgart.deInstitution: Universität Stuttgart, Institut für Maschinelle Sprachverarbeitung GND: 173873820ORCID: 0000-0002-2014-6619 8 Kontakt: roman.klinger@ims.uni-stuttgart.deInstitution: Universität Stuttgart, Institut für Maschinelle
9 Sprachverarbeitung GND: 173873820ORCID: 0000-0002-2014-6619
7 10
9 12
10 DOI: 10.17175/2019_008 13 DOI: 10.17175/2019_008_v2
11 Nachweis im OPAC der Herzog August Bibliothek: 167855300X 14 Nachweis im OPAC der Herzog August Bibliothek: 176443949X
12 Erstveröffentlichung: 16.12.2019 15 Erstveröffentlichung: 16.12.2019
13 16
14 Lizenz: Sofern nicht anders angegeben 17Version 2.0: 23.07.2021
17 Medienlizenzen: Medienrechte liegen bei den Autoren
19 Letzte Überprüfung aller Verweise: 27.11.2019
21 GND-Verschlagwortung: Gefühl | Hermeneutik | Literaturwissenschaft | Netzwerkanalyse (Soziologie) | Textanalyse |
23 Empfohlene Zitierweise: Evgeny Kim, Roman Klinger: A Survey on Sentiment and Emotion Analysis for Computational Literary Studies. In: Zeitschrift für digitale Geisteswissenschaften. Wolfenbüttel 2019. text/html Format. DOI: 10.17175/2019_008
29 Abstract
30 Emotions are a crucial part of compelling narratives: literature tells us about
31 people with goals, desires, passions, and intentions. In the past, the
32 affective dimension of literature was mainly studied in the context of literary
33 hermeneutics. However, with the emergence of the research field known as
34 Digital Humanities (DH), some studies of emotions in a literary context have
35 taken a computational turn. Given the fact that DH is still being formed as a
36 field, this direction of research can be rendered relatively new. In this
37 survey, we offer an overview of the existing body of research on sentiment and
38 emotion analysis as applied to literature. The research under review deals with
39 a variety of topics including tracking dramatic changes of a plot development,
40 network analysis of a literary text, and understanding the emotionality of
41 texts, among other topics.
46Emotionen sind ein wichtiger Bestandteil überzeugender Erzählungen,
47 Literatur beschreibt schließlich Menschen und ihre Ziele, Wünsche,
48 Leidenschaften und Absichten. In der Vergangenheit wurde diese affektive
49 Dimension hauptsächlich im Rahmen der literarischen Hermeneutik
50 untersucht. Mit dem Aufkommen des Forschungsfeldes Digital Humanities
51 (DH) wurde jedoch in einigen Studien bezüglich des Aspekts der Emotionen
52 im literarischen Kontext eine Wende hin zu komputationellen Methoden
53 vorgenommen. Diese Forschungsrichtung ist aktuell durch die Prozesse in
54 den DH in einer Neugestaltung. In diesem Artikel berichten wir über den aktuellen
55 Forschungsstand zur
56 Sentiment- und Emotionsanalyse zur Analyse von Literatur. Wir behandeln
57 eine Vielzahl von Themen, wie zum Beispiel die Veränderungen der
58 emotionalen Konnotation im Verlauf eines Texts, der Netzwerkanalyse
59 eines literarischen Textes und dem Verständnis der Emotionalität von Texten.
63Zu diesem Artikel ist eine überarbeitete Version erschienen: Version 2
66 1 Introduction and Motivation
67 1.1 Emotions and Arts
68 2 Affect and Emotion
69 2.1 Ekman’s Theory of Basic Emotions
70 2.2 Plutchik’s Wheel of Emotions
71 2.3 Russel’s Circumplex Model
72 3 Emotion Analysis in Non-computational Literary Studies
73 4 Emotion and Sentiment Analysis in Computational Literary Studies
74 4.1 Emotion Classification
75 4.1.1 Classification based on emotions
76 4.1.2 Classification of happy ending vs. non-happy endings
77 4.2 Genre and Story-type Classification
78 4.2.1 Story-type clustering
79 4.2.2 Genre classification
80 4.3 Temporal Change of Sentiment
81 4.3.1 Topography of emotions
82 4.3.2 Tracking sentiment
83 4.3.3 Sentiment recognition in historical texts
84 4.4 Character Network Analysis and Relationship Extraction
85 4.4.1 Sentiment dynamics between characters
86 4.4.2 Character analysis and character relationships
87 4.5 Other Types of Emotion Analysis
88 4.5.1 Emotion flow analysis and visualization
89 4.5.2 Miscellaneous
90 5 Discussion and Conclusion
91 Acknowledgements
92 Bibliographic References
93 List of Figures with Captions
96 1 Introduction and Motivation
98 This article deals with emotion and sentiment analysis in computational literary studies.
99 Following Liu[1], we define sentiment as a
100 positive or negative feeling
101 underlying the opinion. The term opinion in this sense is
102 close to attitude in psychology and both sentiment analysis
103 and opinion mining are often used interchangeably. Sentiment analysis is an area of
104 computational linguistics that analyzes people’s sentiments and opinions regarding
105 different objects or topics. Though sentiment analysis is primarily text-oriented,
106 there are multimodal approaches as well.[2]
107 Defining the concept of emotion is a challenging task. As
108 Scherer puts it, defining emotion is a notorious problem.[3] Indeed, different methodological and conceptual
109 approaches to dealing with emotions lead to different definitions. However, the
110 majority of emotion theorists agree that emotions involve a set of expressive,
111 behavioral, physiological, and phenomenological features.[4] In this view, an emotion can be defined as an
112 integrated feeling state involving physiological changes, motor-preparedness,
113 cognitions about action, and inner experiences that emerges from an appraisal of the
114 self or situation.[5]
115 Similar to sentiment, emotions can be analyzed computationally. However, the goal
116 of
117 emotion analysis is to recognize the emotion, rather than sentiment, which makes it
118 a
119 more difficult task as differences between emotions are subtler than those between
120 positive and negative.
122 Although sentiment and emotion analysis are different tasks, our review of the
123 literature shows that the use of either term is not always consistent. There are
124 cases where researchers analyze only positive and negative aspects of a text but
125 refer to their analysis as emotion analysis. Likewise, there are cases where
126 researchers look into a set of subjective feelings including emotions but call it
127 sentiment analysis. Hence, to avoid confusion, in this survey, we use the terms emotion analysis and sentiment analysis
128 interchangeably. In most cases, we follow the terminology used by the authors of the
129 papers we discuss (i.e., if they call emotions sentiments, we do the same).
131 Finally, we talk about sentiment and emotion analysis in the context of computational
132 literary studies. Da defines computational literary studies as the statistical
133 representation of patterns discovered in text mining fitted to currently existing
134 knowledge about literature, literary history, and textual production.[6] Computational literary studies are
135 synonymous to distant reading[7] and digital
136 literary studies,[8]
137 each of which refers to the practice of running a textual analysis on a computer to
138 yield quantitative results. In this survey, we use all of these terms interchangeably
139 and when we refer to digital humanities as a field, we refer to those groups of
140 researchers whose primary objects of study are texts.
143 1.1 Emotions and Arts
145 Much of our daily experiences influence and are influenced by the emotions we
146 experience.[9] This experience is
147 not limited to real events. People can feel emotions because they are reading a novel
148 or watching a play or a movie.[10] There is a growing
149 body of literature that pinpoints the importance of emotions for literary comprehension,
150 [11] as well as research
151 that recognizes the deliberate choices people make with regard to their emotional
152 states when seeking narrative enjoyment such as a book or a film[12]
153 The link between emotions and arts in general is a matter of debate that dates back
154 to the Ancient period, particularly to Plato, who viewed passions and desires as the
155 lowest kind of knowledge and treated poets as undesirable members in his ideal
156 society.[13] In contrast, Aristotle’s
157 view on emotive components of poetry expressed in his Poetics[14] differed from Plato’s in that
158 emotions do have great importance, particularly in the moral life of a person.[15] In the late nineteenth
159 century the emotion theory of arts stepped into the spotlight of philosophers. One
160 of
161 the first accounts on the topic is given by Leo Tolstoy in 1898 in his essay What is Art?.[16] Tolstoy argues that art
162 can express emotions experienced in fictitious context and the degree to which the
163 audience is convinced of them defines the success of the artistic work.[17]
164 New methods of quantitative research emerged in humanities scholarship bringing forth
165 the so-called digital revolution[18] and the transformation of the
166 field into what we know as digital humanities.[19] The adoption of computational
167 methods of text analysis and data mining from the fields of then fast-growing areas
168 of computational linguistics and artificial intelligence provided humanities scholars
169 with new tools of text analytics and data-driven approaches to theory
170 formulation.[20]
171 To the best of our knowledge, the first work[21] on a computer-assisted modeling of emotions in
172 literature appeared in 1982. Challenged by the question of why some texts are more
173 interesting than others, Anderson and McMaster concluded that the emotional tone of
174 a
175 story can be responsible for the reader’s interest. The results of their study
176 suggest that a large-scale analysis of the emotional tone of a collection of texts
177 is
178 possible with the help of a computer program. There are two implications of this
179 finding. First, they suggested that by identifying emotional tones of text passages
180 one can model affective patterns of a given text or a collection of texts, which in
181 turn can be used to challenge or test existing literary theories. Second, their
182 approach to affect modeling demonstrates that the stylistic properties of texts can
183 be defined on the basis of their emotional interest and not only their linguistic
184 characteristics. With regard to these implications, this work is an important early
185 piece as it laid out a roadmap for some of the basic applications of sentiment and
186 emotion analysis of texts, namely sentiment and emotion pattern recognition from text
187 and computational text characterization based on sentiment and emotion.
189 With the development of research methods used by digital humanities researchers, the
190 number of approaches and goals of emotion and sentiment analysis in literature has
191 grown. The goal of this survey is to provide an overview of these recent methods of
192 emotion and sentiment analysis as applied to a text. The survey is directed at
193 researchers looking for an introduction to the existing research in the field of
194 sentiment and emotion analysis of a (primarily, literary) text. The survey does not
195 cover applications of emotion and sentiment analysis in the areas of digital
196 humanities that are not focused on text. Neither does it provide an in-depth overview
197 of all possible applications of emotion analysis in the computational context outside
198 of the DH line of research.
203 2 Affect and Emotion
205 The history of emotion research has a long and rich tradition that followed Darwin’s
206 1872 publication of The Expression of the Emotions in Man and Animals[22]. The subject of emotion theories is vast
207 and diverse. We refer the reader to Maria Gendron’s paper[23] for a brief history of ideas about emotion
208 in psychology. Here, we will focus on three views on emotion that are popular in
209 computational analysis of emotions: Ekman’s theory of basic
210 emotions, Plutchik’s wheel of emotion, and Russel’s
211 circumplex model.
214 2.1 Ekman’s Theory of Basic Emotions
216 The basic emotion theory was first articulated by Silvan Tomkins[24] in the early 1960s. Tomkins postulated that each instance
217 of a certain emotion is biologically similar to other instances of the same emotion
218 or shares a common trigger. One of Tomkins’ mentees, Paul Ekman, put in question the
219 existing emotion theories that proclaimed that facial expressions of emotion are
220 socially learned and therefore vary from culture to culture. Ekman, Sorenson and
221 Friesen challenged this view[25]
222 in a field study with the outcome that facial displays of fundamental emotions are
223 not learned but innate. However, there are culture-specific prescriptions about how
224 and in which situations emotions are displayed.
226 Based on the observation of facial behavior in early development or social
227 interaction, Ekman’s theory also postulates that emotions should be considered discrete categories[26]
228 rather than continuous. Though this
229 view allows for conceiving of emotions as having different intensities, it does not
230 allow emotions to blend and leaves no room for more complex affective states in which
231 individuals report the co-occurrence of like-valenced discrete
232 emotions.[27] This and other theory
233 postulates were widely criticized and disputed in literature.[28]
236 2.2 Plutchik’s Wheel of Emotions
238 Another influential model of emotions was proposed by Robert Plutchik in the early
239 1980s.[29] The important difference
240 between Plutchik’s theory and Ekman’s theory is that apart from a small set of basic
241 emotions, all other emotions are mixed and derived from the various combinations of
242 basic ones. He further categorized these other emotions into the primary dyads (very likely to co-occur), secondary
243 dyads (less likely to co-occur) and tertiary dyads
244 (seldom co-occur).
246 In order to represent the organization and properties of emotions as defined by his
247 theory, Plutchik proposed a structural model of emotions known nowadays as Plutchik’s wheel of emotions. The wheel Figure 1 is constructed in the fashion of a color wheel, with
248 similar emotions placed closer together and opposite emotions 180 degrees apart. The
249 intensity of an emotion in the wheel depends on how far from the center a part of
250 a
251 petal is, i.e., emotions become less distinguishable the further they are from the
252 center of the wheel. Essentially, the wheel is constructed from eight basic bipolar
253 emotions: joy versus sorrow, anger versus fear, trust versus disgust, and surprise versus anticipation. The blank spaces
254 between the leaves are so-called primary dyads – emotions that
255 are mixtures of two of the primary emotions.
257 The wheel model of emotions proposed by Plutchik had a great impact on the field of affective computing
258 being primarily used as a basis for emotion categorization in emotion recognition
259 from text.[30] However, some postulates of the theory are criticized,
260 for example, there is no empirical support for the wheel structure.[31] Another criticism is that
261 Plutchik’s model of emotion does not explain the mechanisms by which love, hate, relief, pride, and other everyday emotions emerge
262 from the basic emotions, nor does it provide reliable
263 measurements of these emotions.[32]
266 Fig. 1: Plutchik’s wheel of emotions. [Plutchik 2011.
267 PD]
272 2.3 Russel’s Circumplex Model
274 Attempts to overcome the shortcomings of basic emotions theory and its unfitness for
275 clinical studies led researchers to suggest various dimensional models, the most
276 prominent of which is the circumplex model of affect proposed by James Russel.[33] The word circumplex
277 in the name of the model refers to the fact that emotional episodes do not cluster
278 at
279 the axes but rather at the periphery of a circle Figure 2. At the core of the
280 circumplex model is the notion of two dimensions plotted on a circle along horizontal
281 and vertical axes. These dimensions are valence (how pleasant
282 or unpleasant one feels) and arousal (the degree of calmness
283 or excitement). The number of dimensions is not strictly fixed and there are
284 adaptations of the model that incorporate more dimensions. One example of this is
285 the
286 Valence-Arousal-Dominance model that adds an additional
287 dimension of dominance, the degree of control one feels over the situation that
288 causes an emotion.[34]
289 By moving from discrete categories to a dimensional representation, the researchers
290 are able to account for subjective experiences that do not fit nicely into the
291 isolated non-overlapping categories. Accordingly, each affective experience can be
292 depicted as a point in a circumplex that is described by only
293 two parameters – valence and arousal –
294 without need for labeling or reference to emotion concepts for which a name might
295 only exist in particular subcommunities or which are difficult to describe.[35] However, the strengths of the model turned
296 out to be its weaknesses: for example, it is not clear whether there are basic
297 dimensions in the model[36] nor is it
298 clear what should be done with qualitatively different events of fear, anger, embarrassment and
299 disgust that fall in identical places in the circumplex
300 structure.[37] Despite these
301 shortcomings, the circumplex model of affect is widely used in psychologic and
302 psycholinguistic studies. In computational linguistics, the circumplex model is
303 applied when the interest is in continuous measurements of valence and arousal rather than in the specific
304 discrete emotional categories.
308 Fig. 2: Circumplex model of affect: Horizontal axis represents the valence dimension,
309 the vertical axis represents the arousal dimension. Drawn after Posner et al. 2005. [Kim / Klinger 2019]
314 3 Emotion Analysis in Non-computational Literary Studies
316 Until the end of the twentieth century, literary and art theories often disregarded
317 the importance of the aesthetic and affective dimension of literature, which in part
318 stemmed from the rejection of old-fashioned literary history that had explained the
319 meaning of art works by the biography of the author.[38] However, the affective turn taken by a wide range of
320 disciplines in the past two decades – from political and sociological sciences to
321 neurosciences or media studies – has refueled the interest of literary critics in
322 human affects and sentiments.
324 We said in Section 1 that there seems to be a consensus among literary critics that
325 literary art and emotions go hand in hand. However, one might be challenged to define
326 the specific way in which emotions come into play in the text. The exploration of
327 this problem is presented by van Meel.[39]
328 Underpinning the centrality of human destiny, hopes, and feelings in the themes of
329 many artworks – from painting to literature – van Meel explores how emotions are
330 involved in the production of arts. Pointing out big differences between the two
331 media in their attempts to depict human emotions (painting conveys nonverbal behavior
332 directly, but lacks temporal dimensions that novels have and use to describe
333 emotions), van Meel provides an analysis of the nonverbal descriptions used by the
334 writers to convey their characters’ emotional behavior. Description of visual
335 characteristics, van Meel speculates, responds to a fundamental need of a reader to
336 build an image of a person and their behavior. Moreover, nonverbal descriptions add
337 important information that can in some cases play a crucial hermeneutical role, such
338 as in Kafka’s Der Prozess, where the fatal decisions for K. are made clear by gestures rather than
339 words. His verdict is not announced, but is implied by the judge who refuses a
340 handshake. The same applies to his death sentence that is conveyed to him by his
341 executioners playing with a butcher’s knife above his head.
343 A hermeneutic approach through the lense of emotions is presented by Kuivalainen[40] and provides a detailed analysis of
344 linguistic features that contribute to the characters’ emotional involvement in
345 Mansfield’s prose. The study shows how, through the extensive use of adjectives,
346 adverbs, deictic markers, and orthography, Mansfield steers the reader towards the
347 protagonist’s climax. Subtly shifting between psycho-narration and free indirect
348 discourse, Mansfield is making use of evaluative and emotive descriptors in
349 psycho-narrative sections, often marking the internal discourse with dashes,
350 exclamation marks, intensifiers, and repetition that thus trigger an emotional
351 climax. Various deictic features introduced in the text are used to pinpoint the
352 source of emotions, which helps in creating a picture of characters’ emotional world.
353 Verbs (especially in the present tense), adjectives, and adverbs serve the same goal
354 in Mansfield’s prose of describing the characters’ emotional world. Going back and
355 forth from psycho-narration to free indirect discourse provides Mansfield with a tool
356 to point out the significant moments in the protagonists’ lives and establish a
357 separation between characters and narration.
359 Both van Meel’s and Kuivalainen’s works, separated from each other by more than a
360 decade, underpin the importance of emotions in the interpretation of characters’
361 traits, hopes, and tragedy. Other authors find these connections as well. For
362 example, Barton[41] proposes instructional
363 approaches to teach school-level readers to interpret character’s emotions and use
364 this information for story interpretation. Van Horn[42] shows that understanding characters emotionally or trying to help
365 them with their problems made reading and writing more meaningful for middle school
366 students.
368 Emotions in text are often conveyed with emotion-bearing words.[43] At the same time their role in the creation
369 and depiction of emotion should not be overestimated. That is, saying that someone
370 looked angry or fearful or sad, as well as directly expressing characters’ emotions,
371 are not the only ways authors build believable fictional spaces filled with
372 characters, action, and emotions. In fact, many novelists strive to express emotions
373 indirectly by way of figures of speech or catachresis,[44] first of all because emotional language can be
374 ambiguous and vague, and, second, to avoid any allusions to Victorian emotionalism
375 and pathos.
377 How can an author convey emotions indirectly? A book chapter by Hillis Miller in Exploring Text and Emotions[45] seeks the answer to exactly this
378 question. Using Conrad’s Nostromo opening scenes as material, Hillis Miller shows how Conrad’s descriptions of
379 an imaginary space generate emotions in readers without direct communication of
380 emotions. Conrad’s Nostromo opening chapter is an objective description of Sulaco, an imaginary land. The
381 description is mainly topographical and includes occasional architectural metaphors,
382 but it combines wide expanse with hermetically sealed enclosure, which generates
383 depthless emotional detachment[46]. Through the use of present tense, Conrad makes the readers suggest
384 that the whole scene is timeless and does not change. The topographical descriptions
385 are given in a pure materialist way: there is nothing behind clouds, mountains,
386 rocks, and sea that would matter to humankind, not a single feature of the landscape
387 is personified, and not a single topographical shape is symbolic. Knowingly or
388 unknowingly, Miller argues, by telling the readers what they should see – with no
389 deviations from truth – Conrad employs a trope that perfectly matches Kant’s concept of the sublime. Kant’s view of poetry was that true poets tell the truth without
390 interpretation; they do not deviate from what their eyes see. Conrad, or to be more
391 specific, his narrator in Nostromo, is an example of sublime seeing with a latent presence of strong emotions.
392 On the one hand, Conrad’s descriptions are cool and detached. This coolness is caused
393 by the indifference of the elements in the scene. On the other hand, by dehumanizing
394 sea and sky, Conrad generates awe, fear, and a dark foreboding about the kinds of
395 life stories that are likely to be enacted against such a backdrop[47].
397 Hillis Miller’s analysis resonates with some premises from emotion theory that we
398 have discussed previously, namely, Plutchik’s belief that emotions should be studied
399 not by a certain way of expression but by the overall behavior of a person.
400 Considering that such a formula cannot be applied to all literary theory studies
401 about emotions (as not all authors choose to convey emotions indirectly, as well as
402 not all authors tend to comment on characters’ nonverbal emotional behavior), it
403 seems that one should search for a balance between low-level linguistic feature
404 analysis of emotional language and a rigorous high-level hermeneutic inquiry
405 dissecting the form of the novel and its under-covered philosophical layers.
409 4 Emotion and Sentiment Analysis in Computational Literary Studies
411 With this section, we proceed to an overview of the existing body of research on
412 computational analysis of emotion and sentiment in computational literary studies.
413 An
414 overview of the papers including their properties is shown in
415 Table 1. The table, as
416 well as this section, is divided into several subsections, each of which corresponds
417 to a specific application of emotion and sentiment analysis to literature.
418 Section 4.1 reviews the papers that deal with the classification of literary texts in terms
419 of emotions they convey; Section 4.2 examines the papers that address text
420 classification by genre or other story-types based on sentiment and emotion features;
421 Section 4.3 is dedicated to research in modeling sentiments and emotions in texts
422 from previous centuries, as well as research dealing with applications of sentiment
423 analysis to texts written in the past; Section 4.4 provides an overview of sentiment
424 analysis applications to character analysis and character network construction, and
425 Section 4.5 is dedicated to more general applications of sentiment and emotion
426 analysis to literature.
429 4.1 Emotion Classification
431 A straightforward approach to sentiment and emotion analysis is phrasing them as a
432 text classification[48]. A fundamental
433 question of such a classification is how to find the best features and algorithms
434 to
435 classify the data (sentences, paragraphs, entire documents) into predefined classes.
436 When applied to literature, such a classification may be of use for grouping
437 different literary texts in digital collections based on the emotional properties
438 of
439 the stories. For example, books or poems can be grouped based on the emotions they
440 convey or based on whether or not they have happy endings or not.
443 4.1.1 Classification based on emotions
445 Barros et al.[49] aim at answering two
446 research questions: 1) is the classification of Quevedo’s works proposed by the
447 literary scholars consistent with the sentiment reflected by the corresponding
448 poems?; and 2) which learning algorithms are the best for the classification? To that
449 end, they perform a set of experiments on the classification of 185 Francisco de
450 Quevedo’s poems that are divided by literary scholars into four categories and that
451 Barros et al. map to emotions of joy, anger, fear, and sadness.
452 Using the terms joy, anger, fear, and sadness as points of
453 reference, Barros et al. construct a list of emotion words by looking up the synonyms
454 of English emotion words and adjectives associated with these four emotions and
455 translating them into Spanish. Each poem is converted into a vector where each item
456 is a normalized count of words relating to a certain emotion. The experiments with
457 different algorithms show the superiority of decision trees achieving accuracy of
458 almost 60%. However, this result is biased by an unbalanced distribution of classes.
459 To avoid the bias, Barros et al. apply a resampling strategy that leads to a more
460 balanced distribution and repeat the classification experiments. After resampling,
461 the accuracy of decision trees in a 10-fold cross validation achieves 75,13%, thus
462 demonstrating an improvement over the previous classification performance. Based on
463 these results the authors conclude that a meaningful classification of the literary
464 pieces based only on the emotion information is possible.
466 Reed[50] offers a proof-of-concept for performing sentiment analysis on a corpus of
467 twentieth-century American poetry. Specifically, Reed analyzes the expression of
468 emotions in the poetry of the Black Arts Movement of the 1960s
469 and 1970s. The paper describes the project Measured Unrest in the Poetry of the Black
470 Arts Movement whose goal is to understand 1) how the feelings associated with
471 injustice are coded in terms of race and gender, and 2) what sentiment analysis can
472 show us about the relations between affect and gender in poetry. Reed notes that
473 surface affective value of the words does not always align with their more nuanced
474 affective meaning shaped by poetic, social, and political contexts.
476 Yu[51] explores what linguistic patterns
477 characterize the genre of sentimentalism in early American novels. To that end, they
478 construct a collection of five novels from the mid-nineteenth century and annotate
479 the emotionality of each of the chapters as high or low. The respective chapters are then classified using
480 support-vector machines and naïve Bayes classifiers as highly emotional or the
481 opposite. The results of the evaluation suggest that arbitrary feature reduction
482 steps such as stemming and stopword removal should be taken very carefully, as they
483 may affect the prediction. For example, Yu shows that no stemming leads to better
484 classification results. A possible explanation is that stemming conflates and
485 neutralizes a large number of discriminative features. The author provides an example
486 of such a conflation with the words wilderness and wild. While the latter can appear anywhere in the text, the
487 former one is primarily encountered in the chapters filled with emotions.
491 4.1.2 Classification of happy ending vs. non-happy endings
493 Zehe et al.[52] argue that automatically
494 recognizing a happy ending as a major plot element could help to better understand
495 a
496 plot structure as a whole. To show that this is possible, they classify 212 German
497 novels written between 1750 and 1920 as having happy or non-happy endings. A novel
498 is
499 considered to have a happy ending if the situation of the main characters in the
500 novel improves towards the end or is constantly favorable. The novels were manually
501 annotated with this information by domain experts. For feature extraction, the
502 authors first split each novel into n segments of the same
503 length. They then calculate sentiment values for each of the segments by counting
504 the
505 occurrences of words that appear in the respective segment and that are found in the
506 German version of the NRC Word-Emotion Association Lexicon[53] and divide this number by the
507 length of the dictionary. Finally, they calculate the sentiment score for the
508 sections by taking the average of all sentiment scores in the segments that are part
509 of the section. These steps are then followed by classification with a support-vector
510 machine and the F1 score of 0.73, which the authors consider a good starting point
511 for future work.
516 4.2 Genre and Story-type Classification
518 The papers we have discussed so far focus on understanding the emotion associated
519 with units of texts. This extracted information can further be used for downstream
520 tasks and also for downstream evaluations. We discuss the following downstream
521 classification cases here. The papers in this category use sentiment and emotion
522 features for a higher-level classification, namely story-type clustering and literary
523 genre classification. The assumption behind these works is that different types of
524 literary text may show different composition and distribution of emotion vocabulary
525 and thus can be classified based on this information. The hypothesis that different
526 literary genres convey different emotions stems from common knowledge: we know that
527 horror stories instill fear and that mysteries evoke anticipation and anger while romances
528 are filled with joy and love. However
529 as we will see in this section, the task of automatic classification of these genres
530 is not always that straightforward and reliable.
533 4.2.1 Story-type clustering
535 Similarly to Zehe et al., Reagan et al.[54] are interested in automatically understanding a plot structure as a
536 whole, not limited to a book ending. The inspiration for their work comes from Kurt
537 Vonnegut’s lecture on emotional arcs of stories.[55]
538 Reagan et al. test the idea that the plot
539 of each story can be plotted as an emotional arc, i.e. a time
540 series graph, where the x-axis represents a time point in a
541 story, and the y-axis represents the events happening to the
542 main characters that can be favorable (peaks on a graph) or unfavorable (troughs on
543 a
544 graph). As Vonnegut puts it, the stories can be grouped by these arcs and the number of such groupings is limited. To test this idea, Reagan
545 et al. collect the 1,327 most popular books from the Project Gutenberg.[56] Each book is then split into segments for which
546 sentiment scores (happy vs. sad) are
547 calculated and compared. The results of the analysis show support for six emotional
548 patterns that are shared between subgroupings of the corpus:
551 Rise: the arc starts at a low point and steadily increases towards the end;
552 Fall: the arc starts at a high point and steadily decreases towards the end;
553 Fall-rise: the arc drops in the middle of the story but increases towards the
554 end;
556 Rise-fall: the arc hits the high point in the middle of the story and decreases
557 towards the end;
559 Rise-fall-rise: the arc fluctuates between high and low points but ends with an
560 increase;
562 Fall-rise-fall: the arc fluctuates between high and low points but ends with a
563 decrease.
566 Additionally, Reagan et al. find that Icarus, Oedipus, and Man in the hole arcs are
567 the three most popular emotional arcs among readers, based on download counts.
571 4.2.2 Genre classification
573 There are other studies[57] that are similar in spirit to the work done by
574 Reagan. Samothrakis and Fasli examine the hypothesis that different genres clearly
575 have different emotion patterns to reliably classify them with machine learning. To
576 that end, they collect works of the genres mystery, humor, fantasy, horror, science fiction and western from the Project Gutenberg.
578 Using WordNet-Affect[58] to
579 detect emotion words as categorized by Ekman’s fundamental emotion classes, they
580 calculate an emotion score for each sentence in the text. Each work is then
581 transformed into six vectors, one for each basic emotion. A random forest classifier
582 achieves a classification accuracy of 0.52. This is significantly higher than a
583 random baseline, which allows the authors to conclude that such a classification is
584 feasible.
586 A study by Kim et al.[59] originates from
587 the same premise as the work by Samothrakis and Fasli but puts emphasis on finding
588 genre-specific correlations of emotion developments. Extending the set of tracked
589 emotions to Plutchik’s classification, Kim et al. collect 2,000 books from the
590 Project Gutenberg that belong to five genres found in the Brown corpus[60], namely adventure, science fiction, mystery, humor and romance.
591 The authors extend the set of classification algorithms beyond random forests using
592 a
593 multi-layer perceptron and convolutional
594 neural networks, which achieves the best performance (0.59 F1-score). To
595 understand how uniform the emotion patterns in different genres are, the authors
596 introduce the notion of prototypicality, which is computed as
597 average of all emotion scores. Using this as a point of reference for each genre Kim
598 et al. use Spearman correlation to calculate the uniformity of emotions per genre.
599 The results of this analysis suggest that fear and anger are the most salient plot devices in fiction, while joy is only of mediocre stability, which is in line with
600 findings of Samothrakis and Fasli.
602 The study by Henny-Krahmer[61] pursues
603 two goals: 1), to test whether different subgenres of Spanish American literature
604 differ in degree and kind of emotionality, and 2), whether emotions in the novels
605 are
606 expressed in direct speech of characters or in narrated text. To that end, they
607 conduct a subgenre classification experiment on a corpus of Spanish American novels
608 using sentiment values as features. To answer the first question, each novel is split
609 into five segments and for each sentence in the segment the emotion score (polarity
610 values + Plutchik’s basic emotions) is calculated using SentiWordNet[62] and NRC[63] dictionaries. The classifier achieves an average F1
611 of 0.52, which is higher than the most-frequent class baseline and, hence, provides
612 a
613 support for emotion-based features in subgenre classification. The analysis of
614 feature importance shows that the most salient features come from the sentiment
615 scores calculated from the characters’ direct speech and that novels with higher
616 values of positive speech are more likely to be sentimental novels.
618 There are some limitations to the studies presented in this section. On the one hand,
619 it is questionable how reliable coarse emotion scoring is that
620 takes into account only presence or absence of words found in specialized
621 dictionaries and overlooks negations and modifiers that can either negate an emotion
622 word or increase/decrease its intensity. On the other hand, a limited view of the
623 emotional content as a sum of emotion bearing words reserves no room for qualitative
624 interpretation of the texts – it is not clear how one can distinguish between emotion
625 words used by the author to express their sentiment, between words used to describe
626 characters’ feelings, and emotion words that characters use to address or describe
627 other characters in a story.
632 4.3 Temporal Change of Sentiment
634 The papers that we have reviewed so far approach the problem of sentiment and emotion
635 analysis as a classification task. However, applications of sentiment analysis are
636 not only limited to classification. In other fields, for example computational social
637 sciences, sentiment analysis can be used for analyzing political preferences of the
638 electorate or for mining opinions about different products or topics. Similarly,
639 several digital humanities studies incorporate sentiment analysis methods in a task
640 of mining sentiments and emotions of people who lived in the past. The goal of these
641 studies is not only to recognize sentiments, but also to understand how they were
642 formed.
645 4.3.1 Topography of emotions
647 Heuser et al.[64] start with a premise
648 that emotions occur at a specific moment in time and space, thus making it possible
649 to link emotions to specific geographical locations. Consequently, having such
650 information at hand, one can understand which emotions are hidden behind certain
651 landmarks. As a proof-of-concept, Heuser et al. build an interactive map,
652 Mapping emotions in Victorian London[65], where each location is tagged with emotion
653 labels. To construct a corpus for their analysis, Heuser et al. collect a large
654 corpus of English books from the eighteenth and nineteenth century and extract 383
655 geographical locations of London that have at least ten mentions each. The resulting
656 corpus includes 15,000 passages, each of which has a toponym in the middle and 100
657 words directly preceding and following the location mention. The data is then given
658 to annotators who are asked to define whether each of the passages expressed happiness or fear, or neutrality. The same data is also analyzed by a custom sentiment analysis
659 program that would assign each passage one of these emotion categories.
661 Some striking observations are made with regard to the data analysis. First, there
662 is
663 a clear discrepancy between fiction and reality – while toponyms from the West End
664 with Westminster and the City are over-represented in the books, the same does not
665 hold true for the East End with Tower Hamlets, Southwark, and Hackney. Hence, there
666 is less information about emotions pertaining to these particular London locations.
667 Another striking detail is that the resulting map is dominated by the neutral
668 emotion. Heuser et al. argue that this has nothing to do with the absence of emotions
669 but rather stems from the fact that emotions tend to be silenced in public domain,
670 which influenced the annotators decision.
672 The space and time context are also used by Bruggman and Fabrikant[66] who model sentiments of Swiss
673 historians towards places in Switzerland in different historical periods. As the
674 authors note, it is unlikely that a historian will directly express attitudes towards
675 certain toponyms, but it is very likely that words they use to describe those can
676 bear some negative connotation (e.g. cholera, death). Correspondingly, such places
677 should be identified as bearing negative sentiment by a sentiment analysis tool.
678 Additionally, they study the changes of sentiment towards a particular place over
679 time. Using the General Inquirer (GI) lexicon[67] to identify
680 positive and negative terms in the document, they assign each document a sentiment
681 score by summing up the weights of negative and positive words and normalizing them
682 by the document length. The authors conclude that the results of their analysis look
683 promising, especially regarding negatively scored articles. However, the authors find
684 difficulties in interpreting positively ranked documents, which may be due to the
685 fact that negative information is more salient.
689 4.3.2 Tracking sentiment
691 Other papers in this category link sentiment and emotion to certain groups, rather
692 than geographical locations. The goal of these studies is to understand how sentiment
693 within and towards these groups was formed.
695 Taboada et al.[68]
696 aim at tracking the literary reputation of six authors writing in the first half of
697 the twentieth century. The research questions raised in the project are how the
698 reputation is made or lost, and how to find correlation between what is written about
699 the author and their work to the author’s reputation and subsequent canonicity. To
700 that end, the project’s goal is to examine critical reviews of six authors’ writing
701 and to map information contained in texts critical to the author’s reputation. The
702 material they work with includes not only reviews, but also press notes, press
703 articles, and letters to editors (including from the authors themselves). For the
704 pilot project with Galsworthy and Lawrence they collected and scanned 330 documents
705 (480,000 words). The documents are tagged for the parts of speech and relevant words
706 (positive and negative) are extracted using custom-made sentiment dictionaries. The
707 sentiment orientation of rhetorically important parts of the texts is then measured.
710 Chen et al.[69] aim to understand personal
711 narratives of Korean comfort women who had been forced into
712 sexual slavery by Japanese military during World War II. Adapting the WordNet-Affect lexicon,[70] Chen et
713 al. build their own emotion dictionary to spot emotional keywords in women’s stories
714 and map the sentences to emotion categories. By adding variables of time and space,
715 Chen et al. provide a unified framework of collective remembering of this historical
716 event as witnessed by the victims.
718 Finally, an interesting project to follow is the Oceanic Exchanges[71] project that started in late 2017. One goal of the project is
719 to trace information exchange in nineteenth-century newspapers and journals using
720 sentiment as one of the variables under analysis.
724 4.3.3 Sentiment recognition in historical texts
726 Other papers put emphasis not so much on the sentiments expressed by writers but
727 instead focus on the particularities of historical language.
729 Marchetti et al.[72] and Sprugnoli et
730 al.
731 [73] present the integration of
732 sentiment analysis in the ALCIDE (Analysis of Language and Content In a Digital Environment)
733 project[74]. The sentiment analysis module is
734 based on WordNet-Affect, SentiWordNet[75] and MultiWordNet.[76] Each
735 document is assigned a polarity score by summing up the words with prior polarity
736 and
737 dividing by the number of words in the document. A positive global score leads to
738 a
739 positive document polarity and a negative global score leads to a negative document
740 polarity. The overall conclusion of their work is that the assignment of a polarity
741 in the historical domain is a challenging task largely due to lack of agreement on
742 polarity of historical sources between human annotators.
744 Challenged by the problem of applicability of existing emotion lexicons to historical
745 texts, Buechel et al.[77] propose a new
746 method of constructing affective lexicons that would adapt well to German texts
747 written up to three centuries ago. In their study, Buechel et al. use the
748 representation of affect based on the Valence-Arousal-Dominance
749 model (an adaptation of Russel’s circumplex model, see Section 2.3).
750 Presumably, such a representation provides a finer-grained insight into the literary
751 text,[78] which is more expressive
752 than discrete categories, as it quantifies the emotion along three different
753 dimensions. As a basis for the analysis, they collect German texts from the Deutsches Textarchiv[79] written
754 between 1690 and 1899. The corpus is split into seven slices, each spanning 30 years.
755 For each slice they compute word similarities and obtain seven distinct emotion
756 lexicons, each corresponding to specific time period. This allows for, the authors
757 argue, the tracing of the shift in emotion association of words over time.
759 Finally, Leemans et al.[80] aim to
760 trace historical changes in emotion expressions and to develop methods to trace these
761 changes in a corpus of 29 Dutch language theatre plays written between 1600 and 1800.
762 Expanding the Dutch version of Linguistic Inquiry and Word Count (LIWC) dictionary[81] with
763 historical terms, the authors are able to increase the recall of emotion recognition
764 with a dictionary. In addition, they develop a fine-grained vocabulary mapping body
765 terms to emotions, and show that a combination of LIWC and their lexicon lead to
766 improvement in the emotion recognition.
771 4.4 Character Network Analysis and Relationship Extraction
773 The papers reviewed above address sentiment analysis of literary texts mainly on a
774 document level. This abstraction is warranted if the goal is to get an insight into
775 the distribution of emotions in a corpus of books. However, emotions depicted in
776 books do not exist in isolation but are associated with characters who are at the
777 core of any literary narrative.[82] This leads us to ask what sentiment and emotion analysis can tell us
778 about the characters. How emotional are they? And what role do emotions play in their
779 interaction?
781 Character relationships have been analyzed in computational linguistics from a graph
782 theoretic perspective, particularly using social network analysis.[83] Fewer works,
783 however, address the problem of modeling character relationships in terms of
784 sentiment. Below we provide an overview of several papers that propose the
785 methodology for extracting this information.
788 4.4.1 Sentiment dynamics between characters
790 Several studies present automatic methods for analyzing sentiment dynamics between
791 plays’ characters. The goal of the study by Nalisnick and Baird[84] is to track the emotional trajectories of
792 interpersonal relationships. The structured format of a dialog allows them to
793 identify who is speaking to whom, which makes it possible to mine
794 character-to-character sentiment by summing the valence values of words that appear
795 in the continuous direct speech and are found in the lexicon[85]
796 of affective norms. The extension[86] of the previous research from the same authors
797 introduces the concept of a sentiment network, a dynamic social network of
798 characters. Changing polarities between characters are modeled as edge weights in
799 the
800 network. Motivated by the desire to explain such networks in terms of a general
801 sociological model, Nalisnick and Baird test whether Shakespeare’s plays obey the
802 Structural Balance Theory by Marvel et al.[87] that postulates that a friend of a
803 friend is also your friend. Using the procedure proposed by Marvel et al. on their
804 Shakespearean sentiment networks, Nalisnick and Baird test whether they can predict
805 how a play’s characters will split into factions using only information about the
806 state of the sentiment network after Act II. The results of their analysis are varied
807 and do not provide adequate support for the Structural Balance Theory as a benchmark
808 for network analysis in Shakespeare’s plays. One reason for that, as the authors
809 state, is inadequacy of their shallow sentiment analysis methods that cannot detect
810 such elements of speech as irony and deceit that play a pivotal role in many literary
811 works.
815 4.4.2 Character analysis and character relationships
817 Elsner[88] aims at answering the
818 question of how to represent a plot structure for summarization and generation tools.
819 To that end, Elsner presents a kernel for comparing novelistic
820 plots at the level of character interactions and their relationships. Using sentiment
821 as one of the characteristics of a character, Elsner demonstrates that the kernel
822 approach leads to meaningful plot representation that can be used for a higher-level
823 processing.
825 Kim and Klinger[89] aim at understanding
826 the causes of emotions experienced by literary characters. To that end, they
827 contribute the REMAN
828 corpus[90] of literary texts with annotations of emotions,
829 experiencers, causes and targets of the emotions. The goal of the project is to
830 enable the automatic extraction of emotions and causes of emotions experienced by
831 the
832 characters. The authors suggest that the results of coarse-grained emotion
833 classification in literary text are not readily interpretable as they do not tell
834 much about who the experiencer of the emotion is. Indeed, if a text mentions two
835 characters, one of whom is angry and another one who is scared because of that, text classification models will only
836 tell us that the text is about anger and fear. Hence, a finer-grained approach towards character relationship
837 extraction is warranted. Kim and Klinger conduct experiments on the annotated dataset
838 showing that the fine-grained approach to emotion prediction with long short-term
839 memory networks outperforms bag-of-words models (an increase
840 in F1 by 12 pp). At the same time, the results of their experiments suggest that
841 joint prediction of emotions and experiencers can be more beneficial than studying
842 these categories separately.
844 Barth et al.[91] develop the character
845 relation analysis tool rCAT with the goal of visualization and
846 analysis of character networks in a literary text. The tool implements a distance
847 parameter (based on token space) for finding pairs of interacting characters. In
848 addition to the general context words that characterize each pair of characters, the
849 tool provides an emotion filter to restrict character relationship analysis to
850 emotions only.
852 A tool presented by Jhavar and Mirza[92] provides a similar functionality: given an input of two character
853 names from the Harry Potter series, the EMoFiel[93] tool identifies the emotion flow between a
854 given directed pair of story characters. These emotions are identified using
855 categorical[94] and continuous[95] emotion models.
857 Egloff et al.[96] present an ongoing
858 work on the Ontology of Literary Characters (OLC) that allows
859 us to capture and infer characters’ psychological traits from their linguistic
860 descriptions. The OLC incorporates the Ontology of Emotion[97] that is based on both Plutchik’s and
861 Hourglass’s[98] models of emotions.
862 The ontology encodes 32 emotion concepts. Based on their natural language
863 description, characters are attributed to a psychological profile along the classes
864 of Openness to experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism. The ontology links
865 each of these profiles to one or more archetypal categories of hero, anti-hero, and villain.
866 Egloff et al. argue that, by using the semantic connections of the OLC, it is
867 possible to infer the characters’ psychological profiles and the role they play in
868 the plot.
870 Kim and Klinger[99] propose a new task
871 of emotion relationship classification between fictional characters. They argue that
872 joining character network analysis with sentiment and emotion analysis may contribute
873 to a computational understanding of narrative structures, as characters are at the
874 center of any plot development. Building a corpus of 19 fan fiction short stories
875 and
876 annotating it with emotions, Kim and Klinger propose several models to classify
877 emotion relations of characters. They show that a deep learning architecture with
878 character position indicators is the best for the task of predicting both directed
879 and undirected emotion relations in the associated social network graph. As an
880 extension to this study, Kim and Klinger[100] explore how emotions are expressed between characters in the same
881 corpus via various non-verbal communication channels.[101] They find
882 that facial expressions are predominantly associated with joy
883 while gestures and body postures are more likely to occur with trust.
885 Finally, a small body of work focuses on mathematical modeling of character
886 relationships. Rinaldi et al.[102]
887 contribute a model that describes the love story between the Beauty and the Beast
888 through ordinary differential equations. Zhuravlev et al.[103] introduce a distance function to model the
889 relationship between the protagonist and other characters in two masochistic short
890 novels by Ivan Turgenev and Sacher-Masoch. Borrowing some instruments from the
891 literary criticism and using ordinary differential equations, Zhuravlev et al. are
892 able to reproduce the temporal and spatial dynamics of the love plot in the two
893 novellas more precisely than it had been done in previous research. Jafari et
894 al.[104] present a dynamic model
895 describing the development of character relationships based on differential
896 equations. The proposed model is enriched with complex variables that can represent
897 complex emotions such as coexisting love and hate.
902 4.5 Other Types of Emotion Analysis
904 We have seen that sentiment analysis as applied to literature can be used for a
905 number of downstream tasks, such as classification of texts based on the emotions
906 they convey, genre classification based on emotions, and sentiment analysis in the
907 historical domain. However, the application of sentiment analysis is not limited to
908 these tasks. In this concluding part of the survey, we review some papers that do
909 not
910 formulate their approach to sentiment analysis as a downstream task. Often, the goal
911 of these works is to understand how sentiments and emotions are represented in
912 literary texts in general, and how sentiment or emotion content varies across
913 specific documents or a collection of them with time, where time can be either
914 relative to the text in question (from beginning to end) or to the historical changes
915 in language (from past to present). Such information is valuable for gaining a deeper
916 insight into how sentiments and emotions change over time, allowing us to bring
917 forward new theories or shed more light onto existing literary or sociological
918 theories.
921 4.5.1 Emotion flow analysis and visualization
923 A set of authors aimed to visualize the change of emotion content through texts or
924 across time. One of the earliest works in this direction is a paper by Anderson and
925 McMaster[105] that starts from
926 the premise that reading enjoyment stems from the affective tones of a text. These
927 affective tones create a conflict that can rise to a climax through a series of
928 crises, which is necessary for a work of fiction to be attractive to the reader.
929 Using a list of 1,000 of the most common English words annotated with valence,
930 arousal, and dominance ratings,[106] they
931 calculate the conflict score by taking the mean of the ratings for each word in a
932 text passage. The more negative the score is, the higher the conflict is, and vice
933 versa. Additionally, they plot conflict scores for each consecutive 100 words of a
934 test story and provide qualitative analysis of the peaks. They argue that a reader
935 who has access to the text would be able to find correlation between events in the
936 story and peaks on the graph. However, the authors still stress that such
937 interpretation remains dependent upon the judgement of the reader. Further, other
938 contributions by the authors are based on the same premises.[107]
939 Alm and Sproat[108] present the results of
940 the emotion annotation task of 22 tales by the Grimm brothers and evaluate patterns
941 of emotional story development. They split emotions into positive and negative categories and divide each
942 story into five parts from which aggregate frequency counts of combined emotion
943 categories are computed. The resulting numbers are plotted on a graph that shows a
944 wave-shaped pattern. From this graph, Alm and Sproat argue, one can see that the
945 first part of the fairy tales is the least emotional, which is probably due to scene
946 setting, while the last part shows an increase in positive emotions, which may
947 signify the happy ending.
949 Two other studies by Mohammad[109] focus on differences in emotion word density as well as emotional
950 trajectories between books of different genres. Emotion word density is defined as
951 a
952 number of times a reader will encounter an emotion word on reading every X words. In addition, each text is assigned several emotion
953 scores for each emotion that are calculated as a ratio of words associated with one
954 emotion to the total number of emotion words occurring in a text. Both metrics use
955 the NRC Affective Lexicon to find occurrences of emotion
956 words. They find that fairy tales have significantly higher anticipation, disgust, joy and
957 surprise word densities, but lower trust word densities when compared to novels.
959 A work by Klinger et al.[110] is a case
960 study in an automatic emotion analysis of Kafka’s Amerika and Das Schloss. The goal of the work is to analyze the development of emotions in both texts
961 as well as to provide a character-oriented emotion analysis that would reveal
962 specific character traits in both texts. To that end, Klinger et al. develop German
963 dictionaries of words associated with Ekman’s fundamental emotions plus contempt and
964 apply them to both texts in question to automatically detect emotion words. The
965 results of their analysis for Das Schloss show a striking increase of surprise towards the end
966 and a peak of fear shortly after start of chapter 3. In the
967 case of Amerika, the analysis shows that there is a decrease in enjoyment after a peak in chapter 4.
969 Yet another work that tracks the flow of emotions in a collection of texts is
970 presented by Kim et al.[111] The authors
971 hypothesize that literary genres can be linked to the development of emotions over
972 the course of text. To test this, they collect more than 2,000 books from five genres
973 (adventure, science fiction, mystery, humor and romance) from Project Gutenberg and identify prototypical emotion shapes for
974 each genre. Each novel in the corpus is split into five consecutive equally-sized
975 segments (following the five-act theory of dramatic acts).[112] All five genres show close correspondence with regard to sadness, anger, fear and disgust, i.e., a consistent increase of
976 these emotions from Act 1 to Act 5, which may correspond to an entertaining
977 narrative. Mystery and science fiction
978 books show increase in anger towards the end, and joy shows an inverse decreasing pattern from Act 1 to Act 2,
979 with the exception of humor.
981 The work by Kakkonen and Galic Kakkonen[113] aims at supporting the literary analysis of Gothic texts at the sentiment level. The authors introduce a
982 system called SentiProfiler that generates visual
983 representations of affective content in such texts and outlines similarities and
984 differences between them, however, without considering the temporal dimension. The
985 SentiProfiler uses WordNet-Affect to
986 derive a list of emotion-bearing words that will be used for analysis. The resulting
987 sentiment profiles for the books are used to visualize the presence of sentiment in
988 a
989 particular document and to compare two different texts.
993 4.5.2 Miscellaneous
995 In this section, we review studies that are different in goals and research questions
996 from the papers presented in previous sections and do not constitute a category on
997 their own.
999 Koolen[114] claims that there is a bias among
1000 readers that put works by female authors on par with »women’s books«, which, as
1001 stated by the author, tend to be perceived as of lower literary quality. She
1002 investigates how much »women’s books« (here, romantic novels
1003 written by women) differ from novels perceived as literary (female and male-authored
1004 literary fiction). The corpus used in the study is a collection of European and
1005 North-American novels translated into Dutch. Koolen uses a Dutch version of the Linguistic Inquiry and Word Count,[115] a dictionary that contains content and sentiment-related categories
1006 of words to count the number of words from different categories in each type of
1007 fiction. Her analysis shows that romantic novels contain more positive emotions and
1008 words pertaining to friendship than in literary fiction. However, female-authored
1009 literary novels and male-authored ones do not significantly differ on any category.
1012 Kraicer and Piper[116] explore the
1013 women’s place within contemporary fiction starting from the premise that there is
1014 a
1015 near ubiquitous underrepresentation and decentralization of women. As a part of their
1016 analysis, Kraicer and Piper use sentiment scores to look at social balance and
1017 »antagonism«, i.e., how different gender pairings influence positive and negative
1018 language surrounding the co-occurrence of characters (using the sentiment dictionary
1019 presented by Liu[117] to calculate a
1020 sentiment score for a character pair). Having analyzed a set of 26,450 characters
1021 from 1,333 novels published between 2001 and 2015, the authors find that sentiment
1022 scores give little indication that the character’s gender has an effect on the state
1023 of social balance.
1025 Morin and Acerbi[118] focus on
1026 larger-scale data spanning a hundred thousand of books. The goal of their study is
1027 to
1028 understand how emotionality of written texts changed throughout the centuries. Having
1029 collected 307,527 books written between 1900 and 2000 from the Google Books
1030 corpus[119] they collect, for each
1031 year, the total number of case-insensitive occurrences of emotion terms that are
1032 found under positive and negative taxonomies of LIWC
1033 dictionary.[120] The main findings
1034 of their research show that emotionality (both positive and
1035 negative emotions) declines with time, and this decline is
1036 driven by the decrease in usage of positive vocabulary. Morin and Acerbi remind us
1037 that the Romantic period was dominated by emotionality in
1038 writing, which could be the effect of a group of writers who wrote above the mean.
1039 If
1040 one assumes that each new writer tends to copy the emotional style of their
1041 predecessors, then writers at one point of time are disproportionally influenced by
1042 this group of above-the-mean writers. However, this trend does not last forever and,
1043 sooner or later, the trend reverts to the mean, as each writer reverts to a normal
1044 level of emotionality.
1046 An earlier work[121] written in
1047 collaboration with Acerbi provides a somewhat different
1048 approach and interpretation of the problem of the decline in positive vocabulary in
1049 English books of the twentieth century. Using the same dataset and lexical resources
1050 (plus WordNet-Affect) Bentley et al. find a strong correlation
1051 between expressed negative emotions and the U.S. economic misery
1052 index, which is especially strong for the books written during and after
1053 the World War I (1918), the Great Depression (1935), and the energy crisis (1975).
1054 However, in the present study,[122] the
1055 authors argue that the extent to which positive emotionality correlates with
1056 subjective well-being is a debatable issue. Morin and Acerbi provide more possible
1057 reasons for this effect as well as detailed statistical analysis of the data, so we
1058 refer the reader to the original paper for more information.
1062 Tab. 1: Summary of characteristics of methods used in the papers reviewed
1063 in this survey. Download as PDF. [Kim / Klinger 2019]
1069 5 Discussion and Conclusion
1071 We have shown throughout this survey that there is a growing interest in sentiment
1072 and emotion analysis within digital humanities. Given the fact that DH have emerged
1073 into a thriving science within the past decade, it may safely be said that this
1074 direction of research is relatively new. At the same time, the research in sentiment
1075 analysis started in computational linguistic more than two decades ago and is
1076 nowadays an established field that has dedicated workshops and tracks in the main
1077 computational linguistics conferences. Moreover, a recent meta-study by Mäntylä et
1078 al.[123] shows that the number of
1079 papers in sentiment analysis is rapidly increasing each year. Indeed, the topic has
1080 not yet outrun itself and we should not expect to see it vanishing within the next
1081 decade or two, provided that no significant paradigm shift in the computational
1082 sciences takes place. One may wonder whether the same applies to sentiment analysis
1083 in digital humanities scholarship. Will the interest in the topic grow continuously
1084 or will it rally to the peak and vanish in a few years?
1086 There is no decisive answer. The popularity of sentiment analysis may have reached
1087 a
1088 peak but is far from fading. Application-wise, not a lot has changed during the past
1089 years: researchers are still interested in predicting sentiment and emotion from text
1090 for different purposes. If anything has changed, it is methodology. Early research
1091 in
1092 sentiment analysis relied on word polarity and specific dictionaries. Modern
1093 state-of-the-art approaches rely on word embeddings and deep learning architectures.
1094 Having started with simple polarity detection, contemporary sentiment analysis has
1095 advanced to a more nuanced analysis of sentiments and emotions.
1097 The situation is somewhat different in digital humanities research. Most of the works
1098 rely on affective lexicons and word counts, a technique for detecting emotions in
1099 literary text first used by Anderson and McMaster in 1982.[124] Even the most recent works base the
1100 interpretation of the results on the use of dictionaries and counts of
1101 emotion-bearing words in a text, passage, or sentence. In fact, around 70% of the
1102 papers we discussed in Section 4 substantially rely on the use of various lexical
1103 resources for detecting emotions (see Table 1 for a summary of methods used in the
1104 reviewed papers). We have discussed some limitations of this approach in Section 4.2.
1105 Let us reiterate its weakness with the following small example. Consider the sentence
1106 ›Jack was afraid of John because John held a knife in his hand‹. Assuming a
1107 dictionary of emotion-bearing words is used, the sentence can be categorized as
1108 expressing fear, because of the two strong fear markers, afraid and knife. Indeed, the sentence
1109 does express fear. But does it do it equally for Jack and
1110 John? The answer is no: Jack is the one who is afraid and John holding a knife is
1111 the
1112 reason for Jack being afraid. Let us assume that a researcher is interested in the
1113 emotion analysis of a book that contains thousands of sentences expressing emotions
1114 in different ways: some sentences describe characters who feel emotions just as in
1115 the sentence above, some are narrator’s digressions filled with emotions, some
1116 contain emotion-bearing words (knife, baby) but do not in fact express the same emotion in any given context. No
1117 doubt, a dictionary and count-based approach will be helpful in understanding the
1118 distribution of the emotion lexicon throughout the story. But is it enough for the
1119 interpretation? Can hermeneutics, in its traditional form, make use of such
1120 knowledge? Barely. In fact, some of the works that we reviewed pinpoint that the
1121 surface affective value of the words does not always align with their more nuanced
1122 affective meaning and that sentiment analysis tools make mistakes when classifying
1123 a
1124 text as emotional or not.[125] If so, how reliable
1125 is the interpretation? In other words, what kind of interpretation should we expect
1126 from the sentiment and emotion analysis research in the DH community?
1128 We do not have a ready answer to that question. At the one extreme, there is
1129 traditional hermeneutics, the examples of which are presented in a Section 3. At the
1130 other extreme, there is interpretation in the form of ›Author A writes with more
1131 emotion than author B because the numbers say so‹. We do, however, suggest that a
1132 balance should be made somewhere between these two extremes. Even as simple as it
1133 is,
1134 the approach of detecting sentiment and emotion-related words can be used to deliver
1135 a high-quality interpretation such as in Heuser et al.[126] or Morin and Acerbi.[127] However, we note again that there are still limits posed by the
1136 simplicity of this approach.
1138 This leads us to an outline of the reality of sentiment analysis research in digital
1139 humanities: the methods of sentiment analysis used by some of the DH scholars
1140 nowadays have gone or are almost extinct among computational linguists. This in turn
1141 affects the quality of the interpretation.
1143 However, we admit that this criticism may be unfair. In fact, there is a possible
1144 reason why DH researchers have taken this approach to sentiment analysis. Digital
1145 humanities are still being formed as an independent discipline and it is easier to
1146 form something new in a step-by-step fashion. Resorting to a metaphor from the
1147 construction world, one should first learn how to stack single bricks to build a wall
1148 rather than starting from the design of a communications system. It is necessary to
1149 make sure that appropriate tools and methods are chosen instead of using what proved
1150 to be successful in other domains without reflection. It is true that much digital
1151 humanities research (especially dealing with text) uses the methods of text analysis
1152 that were in fashion in computational linguistic twenty years ago. One may argue that
1153 new research in digital humanities should start with the state-of-the-art methods. Indeed, some arguments that methodology is at
1154 the root of the interpretation have already been made.[128] So, if there is anything that digital humanities can learn from
1155 computational linguistics, it is that methodology cannot stall. What really matters
1156 for digital humanities is interpretation, and if methodology is not going forward,
1157 the interpretation is not either.
1161 Acknowledgements
1164 We thank Laura Ana Maria Bostan, Sebastian Padó, and Enrica Troiano
1165 for fruitful discussions and the ZfDG team for their help in preparation of this
1166 article. This research has been conducted within the CRETA project which is funded by the German Ministry for Education and
1167 Research (BMBF) and partially funded by the German Research Council (DFG), projects
1168 SEAT (Structured Multi-Domain Emotion Analysis from Text, KL 2869/1-1).
1174 Footnotes
1177 [1]
1179 Liu 2015, p.2.
1182 [2]
1184 Soleymani et al. 2017.
1187 [3]
1189 Scherer 2005, p. 695.
1192 [4]
1194 Scarantino 2016, p. 36.
1197 [5]
1199 Mayer et al. 2008, p. 510.
1202 [6]
1204 Da 2019, p. 602.
1207 [7]
1209 Moretti 2005.
1212 [8]
1214 Hoover et al. 2014.
1217 [9]
1219 Schwarz 2000, p. 433.
1222 [10]
1224 Johnson-Laird / Oatley 2016,
1225 passim; Djikic et al. 2009, passim.
1228 [11]
1230 Robinson 2005;
1231 Hogan 2010;
1232 Hogan 2011;
1233 Bal / Veltkamp 2013;
1234 Djikic et al. 2013;
1235 Johnson 2012;
1236 Samur et al. 2018.
1238 [12]
1240 Zillmann et al. 1980;
1241 Ross 1999;
1242 Bryant / Zillmann 1984;
1243 Oliver 2008;
1244 Mar et al.
1245 2011.
1247 [13]
1249 Plato 1969
1250 , passim.
1253 [14]
1255 Aristotle 1996, passim.
1258 [15]
1260 De Sousa / Scarantino 2018.
1263 [16]
1265 Tolstoy 1962, passim.
1268 [17]
1270 Anderson / McMaster 1986, p. 3;
1271 Hogan 2010, p. 187; Piper /
1272 Jean So 2015.
1275 [18]
1277 Lanham 1989.
1280 [19]
1282 Berry 2012; Schreibman et al. 2015.
1285 [20]
1287 Vanhoutte 2013, p. 142;
1288 Jockers / Underwood
1289 2016, pp. 292f.
1292 [21]
1294 Anderson /
1295 McMaster 1982.
1298 [22]
1300 Darwin 1872, passim.
1303 [23]
1305 Gendron / Feldman Barrett 2009.
1308 [24]
1310 Tomkins 1962, passim.
1313 [25]
1315 Ekman et al. 1969, pp. 86-88.
1318 [26]
1320 Ekman 1993, p. 386.
1323 [27]
1325 Feldman Barrett 1998, pp. 580f.
1328 [28]
1330 Russell 1994;
1331 Russell et al. 2003;
1332 Gendron et al. 2014;
1333 Feldman Barrett 2017.
1336 [29]
1338 Plutchik 1991, passim.
1341 [30]
1343 Cambria et al. 2012;
1344 Kim et al. 2012; Suttles / Ide 2013;
1345 Borth et al. 2013; Abdul-Mageed /
1346 Ungar 2017.
1349 [31]
1351 Smith / Schneider 2009, passim.
1354 [32]
1356 Richins 1997, p. 128.
1359 [33]
1361 Russell 1980.
1364 [34]
1366 Bradley / Lang 1994, p. 50.
1369 [35]
1371 Russell 2003, p. 154.
1374 [36]
1376 Larsen / Diener 1992, p. 25.
1379 [37]
1381 Russell / Feldman Barrett 1999, p. 807.
1384 [38]
1386 Sætre
1387 et al. 2014b, passim.
1390 [39]
1392 Van Meel 1995, passim.
1395 [40]
1397 Kuivalainen 2009, passim.
1400 [41]
1402 Barton 1996, passim.
1405 [42]
1407 Van Horn
1408 1997, passim.
1411 [43]
1413 Johnson-Laird / Oatley 1989, passim.
1416 [44]
1418 Miller 2014, p. 92.
1421 [45]
1423 Sætre et al. 2014a, p. 91ff.
1426 [46]
1428 Miller 2014, p.
1429 93.
1432 [47]
1434 Miller 2014, p. 115.
1437 [48]
1439 Liu 2015, p. 47.
1442 [49]
1444 Barros et al. 2013, passim.
1447 [50]
1449 Reed 2018, passim.
1452 [51]
1454 Yu 2008, passim.
1457 [52]
1459 Zehe et al. 2016, passim.
1462 [53]
1464 Mohammad / Turney 2013, passim.
1467 [54]
1469 Reagan et al.
1470 2016, passim.
1473 [55]
1475 Vonnegut 2010 (2005), passim.
1478 [56]
1480 Project Gutenberg 1971-2019.
1483 [57]
1485 Samothrakis / Fasli 2015;
1486 Kim et al.
1487 2017a; Kim et al. 2017b.
1490 [58]
1492 Strapparava / Valitutti 2004.
1495 [59]
1497 Kim et al. 2017a, passim.
1500 [60]
1502 Francis / Kucera 1979, passim.
1506 [61]
1508 Henny-Krahmer 2018, passim.
1511 [62]
1513 Baccianella et al. 2010.
1517 [63]
1519 Mohammad / Turney 2013.
1522 [64]
1524 Heuser et al. 2016, passim.
1527 [65]
1529 Historypin 2010-2017.
1532 [66]
1534 Bruggmann / Fabrikant 2014, passim.
1537 [67]
1539 Stone et al. 1968.
1542 [68]
1544 Taboada et al. 2006, passim; Taboada et al. 2008, passim.
1547 [69]
1549 Chen et al. 2012, passim.
1552 [70]
1555 Strapparava / Valitutti 2004.
1558 [71]
1560 Oceanic Exchanges 2017.
1563 [72]
1565 Marchetti et al. 2014, passim.
1568 [73]
1570 Sprugnoli et al. 2016, passim.
1573 [74]
1575 ALCIDE Demo 2014-2015.
1578 [75]
1580 Baccianella et al. 2010, passim.
1583 [76]
1585 Pianta et al. 2002, passim.
1588 [77]
1590 Buechel et al. 2017, passim.
1593 [78]
1595 Buechel et al. 2016, p. 54, p. 59.
1598 [79]
1600 Deutsches Textarchiv 2007-2019.
1603 [80]
1605 Leemans et al. 2017, passim.
1608 [81]
1610 Pennebaker et al. 2007.
1613 [82]
1615 Ingermanson / Economy 2009, p.
1616 107.
1619 [83]
1621 Agarwal et al. 2013;
1622 Elson et al. 2011.
1625 [84]
1627 Nalisnick / Baird 2013a, passim.
1630 [85]
1632 Nielsen 2011, passim.
1635 [86]
1637 Nalisnick / Baird 2013b, passim.
1640 [87]
1642 Marvel et al. 2011.
1645 [88]
1647 Elsner 2012, passim;
1648 Elsner 2015, passim.
1651 [89]
1653 Kim / Klinger 2018, passim.
1656 [90]
1658 REMAN - Relational Emotion Annotation for Fiction. Corpus 2018.
1661 [91]
1663 Barth et al. 2018, passim.
1666 [92]
1668 Jhavar / Mirza
1669 2018, passim.
1672 [93]
1674 EMoFiel: Emotion Mapping of Fictional Relationship 2018.
1677 [94]
1679 Plutchik 1991, passim.
1682 [95]
1684 Russell 1980, passim.
1687 [96]
1689 Egloff et al. 2018, passim.
1692 [97]
1694 Patti et al. 2015.
1697 [98]
1699 Cambria et al. 2012, passim.
1702 [99]
1704 Kim / Klinger 2019b, passim.
1707 [100]
1709 Kim / Klinger
1710 2019a, passim.
1713 [101]
1715 Their
1716 analysis is based on Van Meel 1995 we mentioned in
1717 Section 3.
1720 [102]
1722 Rinaldi et al. 2013, passim.
1725 [103]
1727 Zhuravlev et al. 2014, passim.
1730 [104]
1732 Jafari et al. 2016, passim.
1735 [105]
1737 Anderson / McMaster 1986, passim.
1740 [106]
1742 Heise 1965, passim.
1745 [107]
1747 Anderson / McMaster 1982;
1748 Anderson / McMaster 1993.
1751 [108]
1753 Alm / Sproat 2005, passim.
1756 [109]
1758 Mohammad 2011, passim;
1759 Mohammad
1760 2012, passim.
1763 [110]
1765 Klinger et al. 2016, passim.
1768 [111]
1770 Kim et al. 2017b, passim.
1773 [112]
1775 Freytag 1863, passim.
1778 [113]
1780 Kakkonen /
1781 Galic Kakkonen 2011, passim.
1784 [114]
1786 Koolen 2018, passim.
1789 [115]
1791 Boot et al.
1792 2017.
1795 [116]
1797 Kraicer / Piper 2019, passim.
1800 [117]
1802 Liu et al. 2010, passim.
1805 [118]
1807 Morin / Acerbi 2017, passim.
1810 [119]
1812 Google Books Ngram Viewer 2012.
1815 [120]
1817 Pennebaker et al. 2007.
1820 [121]
1822 Bentley et al. 2014, passim.
1825 [122]
1827 Morin / Acerbi 2017, passim.
1830 [123]
1832 Mäntylä et al. 2018, passim.
1835 [124]
1837 Anderson / McMaster 1982, passim.
1840 [125]
1842 Reed 2018, passim.
1845 [126]
1847 Heuser
1848 et al. 2016, passim.
1851 [127]
1853 Morin and Acerbi
1854 2017, passim.
1857 [128]
1859 Da
1860 2019, passim.
1867 Bibliographic References
1870 Muhammad Abdul-Mageed / Lyle Ungar: EmoNet: Fine-grained emotion detection with
1871 gated recurrent neural networks. In: Proceedings of the 55th Annual Meeting of the
1872 Association for Computational Linguistics. (ACL: 55, Vancouver, 30.07.-04.08.2017)
1873 New York, NY 2017, i 1, pp. 718–728. DOI: 10.18653/v1/P17-1067Apoorv Agarwal / Anup Kotalwar / Owen Rambow: Automatic extraction of social
1874 networks from literary text: A case study on Alice in Wonderland. In: Proceedings
1875 of
1876 the Sixth International Joint Conference on Natural Language Processing. (IJCLP: 6,
1877 Nagoya 14.-18.10.2013) Nagoya 2013, pp. 1202–1208. [online]
1879 Cecilia Ovesdotter Alm / Richard Sproat: Emotional sequencing and development in
1880 fairy tales. In: Affective computing and intelligent interaction. First international
1881 conference. Proceedings. Ed. by Jianhua Tao et al. (ACII’05, Beijing,
1882 22.-24.10.2005) Berlin et al. 2005, pp. 668–674. [Nachweis im GVK]
1883 ALCIDE (Analysis of Language and Content In a Digital Environment). Demo. Ed. by Center
1884 for Information Technology Digital Humanities, Fondazione Bruno Kessler / Italian-German
1885 Historical Institute.
1886 In: Alcide Demo. Trento 2014-2015. [online]
1887 Clifford W. Anderson / George E. McMaster: Computer assisted modeling of affective
1888 tone in written documents. In: Computers and the Humanities 16 (1982), i. 1, pp. 1–9.
1889 [Nachweis im GVK]
1890 Clifford W. Anderson / George E. McMaster: Modeling emotional tone in stories
1891 using tension levels and categorical states. In: Computers and the Humanities 20
1892 (1986), i. 1, pp. 3–9. [Nachweis im GVK]
1893 Clifford W. Anderson / George E. McMaster: Emotional tone in Peter Rabbit before
1894 and after simplification. In: Empirical Studies of the Arts 11 (1993), i. 2, pp.
1895 177–185. [Nachweis im GVK]
1896 Aristotle: Poetics. Penguin 1996. (= Penguin Classics)Stefano Baccianella / Andrea Esuli / Fabrizio Sebastiani: Sentiwordnet 3.0: An
1897 enhanced lexical resource for sentiment analysis and opinion mining. In: Proceedings
1898 of the 7th International Conference on Language Resources and Evaluation. (LREC’10:
1899 7, Valetta, 17.05.-23.05.2010) Paris 2010, pp. 2200–2204. PDF. [online]
1900 P. Matthijs Bal / Martijn Veltkamp: How does fiction reading influence empathy? An
1901 experimental investigation on the role of emotional transportation. In: PLOS ONE 8
1902 (2013), i. 1, p. e55341. Article from 30.01.2013. DOI: 10.1371/journal.pone.0055341Lisa Feldman Barrett: Discrete emotions or dimensions? The role of valence focus
1903 and arousal focus. In: Cognition & Emotion 12 (1998), i. 4, pp. 579–599.
1904 [Nachweis im GVK]
1905 Lisa Feldman Barrett: How emotions are made: The secret life of the brain. Boston
1906 et al. 2017. [Nachweis im GVK]
1907 Linda Barros / Pilar Rodriguez / Alvaro Ortigosa: Automatic classification of
1908 literature pieces by emotion detection: a study on quevedo’s poetry. In: 2013 Humaine
1909 Association Conference on Affective Computing and Intelligent Interaction. (ACII
1910 2013: 5, Geneva, 02.-05.09.2013), Piscataway, NJ 2013, pp. 141–146. [Nachweis im GVK]
1911 Florian Barth / Evgeny Kim / Sandra Murr / Roman Klinger: A reporting tool for
1912 relational visualization and analysis of character mentions in literature. In: DHd
1913 2018: Kritik der digitalen Vernunft : Konferenzabstracts. Ed. by Georg Vogeler. (DHd
1914 2018: 5, Köln, 26.02.-02.03.2018), Cologne 2018, pp. 123–127. PDF. [online] [Nachweis im GVK]
1915 James Barton: Interpreting character emotions for literature comprehension. In:
1916 Journal of Adolescent & Adult Literacy 40 (1996), i. 1, pp. 22–28.
1917 [Nachweis im GVK]
1918 Alexander R. Bentley / Alberto Acerbi / Paul Ormerod / Vasileios Lampos: Books
1919 average previous decade of economic misery. In: PLOS ONE 9 (2014), i. 1, p. e83147.
1920 Article from 08.01.2014. DOI: 10.1371/journal.pone.0083147David M. Berry: Introduction: Understanding the digital humanities. In:
1921 Understanding digital humanities. Ed. by David M. Berry. Houndmills et al. 2012, pp.
1922 1–20. [Nachweis im GVK]
1923 Peter Boot / Hanna Zijlstra / Rinie Geenen: The Dutch translation of the
1924 linguistic inquiry and word count (LIWC) 2007 dictionary. In: Dutch Journal of
1925 Applied Linguistics 6 (2017), i. 1, pp. 65–76. [Nachweis im GVK]
1926 Damian Borth / Rongrong Ji / Tao Chen / Thomas Breuel / Shih-Fu Chang: Large-scale
1927 visual sentiment ontology and detectors using adjective noun pairs. In: Proceedings
1928 of the 21st ACM International Conference on Multimedia. (MM '13: 21, Barcelona,
1929 21.-25.10.2013) New York, NY 2013, pp. 223–232. [Nachweis im GVK]
1930 Margaret M. Bradley / Peter J. Lang: Measuring emotion: the self-assessment
1931 manikin and the semantic differential. In: Journal of behavior therapy and
1932 experimental psychiatry 25 (1994), i. 1, pp. 49–59. [Nachweis im GVK]
1933 André Bruggmann / Sara Irina Fabrikant: Spatializing a digital text archive about
1934 history. In: Workshop on Geographic Information Observatories 2014 : proceedings.
1935 Ed.
1936 by Krzysztof Janowicz / Benjamin Adams / Grant McKenzie / Tomi Kauppinen. (GIO 2014
1937 /
1938 GIScience: 8, Vienna, 23.09.2014) Aachen 2014, pp. 6–14. (CEUR Workshop Proceedings,
1939 1273) PDF. [online]
1940 Jennings Bryant / Dolf Zillmann: Using television to alleviate boredom and stress:
1941 Selective exposure as a function of induced excitational states. In: Journal of
1942 Broadcasting & Electronic Media 28 (1984), i. 1, pp. 1–20. [Nachweis im GVK]
1943 Sven Buechel / Johannes Hellrich / Udo Hahn: Feelings from the past – adapting
1944 affective lexicons for historical emotion analysis. In: Proceedings of the Workshop
1945 on Language Technology Resources and Tools for Digital Humanities. (LT4DH, Osaka,
1946 11.12.2016) Stroudsburg, PA 2016, pp. 54–61. PDF. [online]
1947 Sven Buechel / Johannes Hellrich / Udo Hahn: The course of emotion in three
1948 centuries of german text – a methodological framework. In: Digital Humanities 2017:
1949 Conference Abstracts. Ed. by Rhian Lewis et al. (DH 2017, Montreal, 08.-11.08.2017)
1950 Montreal 2017, pp. 176–179. [online]
1951 Erik Cambria / Andrew Livingstone / Amir Hussain: The hourglass of emotions. In:
1952 Cognitive behavioural systems. Ed. by Anna Esposito et al. (COST 2102, Dresden,
1953 21.-26.02.2011) Berlin 2012, pp. 144–157. [Nachweis im GVK]
1954 Annie T. Chen / Ayoung Yoon / Ryan Shaw: People, places and emotions: Visually
1955 representing historical context in oral testimonies. In: Proceedings of the Third
1956 Workshop on Computational Models of Narrative. (CMN’12: 3, Istanbul, 26.-27.05.2012),
1957 pp. 26–27. Cambridge, MA 2012. PDF. [online]
1958 Oceanic Exchanges: Tracing Global Information Networks in Historical Newspaper
1959 Repositories, 1840-1914. Ed. by Oceanic Exchanges Project Team. Boston, MA 2017.
1960 [online]
1961 Nan Z. Da: The computational case against computational literary studies. In:
1962 Critical Inquiry 45 (2019), i. 3, pp. 601–639. [Nachweis im GVK]
1963 Charles Darwin: The expression of emotion in animals and man. London 1872.
1964 [Nachweis im GVK]
1965 Deutsches Textarchiv. Grundlage für ein Referenzkorpus der neuhochdeutschen Sprache.
1966 Ed. by Berlin-Brandenburgischen Akademie der Wissenschaften.
1967 In: Berlin 2007-2019. [online]
1968 Maja Djikic / Keith Oatley / Sara Zoeterman / Jordan B. Peterson: On being moved
1969 by art: How reading fiction transforms the self. In: Creativity Research Journal 21
1970 (2009), i. 1, pp. 24–29. [Nachweis im GVK]
1971 Maja Djikic / Keith Oatley / Mihnea C. Moldoveanu: Reading other minds: Effects of
1972 literature on empathy. In: Scientific Study of Literature 3 (2013), i. 1, pp. 28–47.
1973 [Nachweis im GVK]
1974 Mattia Egloff / Antonio Lieto / Davide Picca: An ontological model for inferring
1975 psychological profiles and narrative roles of characters. In: Digital Humanities
1976 2018: Puentes-Bridges. Book of Abstracts. Hg. von Jonathan Girón Palau / Isabel
1977 Galina Russell. (DH 2018, Mexico City, 26.-29.06.2018) Mexico City 2018, pp. 649–650.
1978 PDF. [online]
1979 Paul Ekman: Facial expression and emotion. In: American psychologist 48 (1993), i.
1980 4, pp. 384–392. [Nachweis im GVK]
1981 Paul Ekman / Richard E. Sorenson / Wallace V. Friesen: Pan-cultural elements in
1982 facial displays of emotion. In: Science 164 (1969), i. 3875, pp. 86–88.
1983 [Nachweis im GVK]
1984 Micha Elsner: Character-based kernels for novelistic plot structure. In:
1985 Proceedings of the 13th Conference of the European Chapter of the Association for
1986 Computational Linguistics. (EACL’12: 13, Avignon, 23.-27.04.2012) Stroudsburg, PA
1987 2012, pp. 634–644. PDF. [online]
1988 Micha Elsner: Abstract representations of plot structure. In: Linguistic Issues in
1989 Language Technology 12 (2015), i. 5. PDF. [online]
1990 David K. Elson / Nicholas Dames / Kathleen R. McKeown: Extracting social networks
1991 from literary fiction. In: Proceedings of the 48th Annual Meeting of the Association
1992 for Computational Linguistics. (ACL: 48, Uppsala, 11.-18.07.2010) Red Hook, NY 2011,
1993 pp. 138–147. PDF. [online] [Nachweis im GVK]
1994 EMoFiel: Emotion Mapping of Fictional Relationship. Ed. by Harshita Jhavar / Paramita
1995 Mirza, Max Planck Institute for Informatics.
1996 In: EMoFiel. Saarbrücken 2018. [online]
1997 Winthrop Nelson Francis / Henry Kucera: Brown corpus manual. Preface to revised
1998 Edition. Providence, RI 1979. [online]
1999 Gustav Freytag: Die Technik des Dramas. Leipzig 1863. [Nachweis im GVK]
2000 Maria Gendron / Lisa Feldman Barrett: Reconstructing the past: A century of ideas
2001 about emotion in psychology. In: Emotion review 1 (2009), i. 4, pp. 316–339.
2002 [Nachweis im GVK]
2003 Maria Gendron / Debi Roberso / Jacoba Marietta van der Vyver / Lisa Feldman
2004 Barrett: Perceptions of emotion from facial expressions are not culturally universal:
2005 Evidence from a remote culture. In: Emotion 14 (2014), i. 2, pp. 251–262.
2006 [Nachweis im GVK]
2007 Google Books Ngram Viewer. Ed. by Google. In: Version
2008 2. 2012. [online]
2009 David Reuben Jerome Heise: Semantic differential profiles for 1,000 most frequent
2010 English words. In: Psychological Monographs: General and Applied 79 (1965), i. 8,
2011 pp.
2012 1–31. [Nachweis im GVK]
2013 Ulrike Edith Gerda Henny-Krahmer: Exploration of sentiments and genre in Spanish
2014 American novels. In: Digital Humanities 2018: Puentes-Bridges. Book of Abstracts.
2015 Hg.
2016 von Jonathan Girón Palau / Isabel Galina Russell. (DH 2018, Mexico City,
2017 26.-29.06.2018) Mexico City 2018, pp. 399–403. PDF. [online]
2018 Ryan Heuser / Franco Moretti / Erik Steiner: The emotions of London. Stanford
2019 2016. (= Literary Lab Pamphlets, 13) PDF.[online]
2020 Mapping emotions in Victorian London. Ed. by. Historypin. In: New
2021 Orleans et al. 2010-2017. [online]
2022 Hillis J. Miller: Text; Action; Space; Emotion in Conrad’s Nostromo. In: Exploring
2023 Text and Emotions. Ed. by Lars Saetre / Lombardo / Julien Zanetta. Aarhus 2014, pp.
2024 91–117. [Nachweis im GVK]
2025 Patrick Colm Hogan: Fictions and feelings: On the place of literature in the study
2026 of emotion. In: Emotion Review 2 (2010), i. 2, pp. 184–195. [Nachweis im GVK]
2027 Patrick Colm Hogan: What Literature Teaches Us about Emotion. New York, NY 2011.
2028 [Nachweis im GVK]
2029 David Lowell Hoover / Jonathan Culpeper / Kieran O’Halloran: Digital literary
2030 studies: Corpus Approaches to Poetry, Prose, and Drama. New York, NY 2014.
2031 [Nachweis im GVK]
2032 Randy Ingermanson / Peter Economy. Writing fiction for dummies. Hoboken, NJ 2009.
2033 [Nachweis im GVK]
2034 Sajad Jafari / Julien Clinton Sprott / Seyed Mohammad Reza Hashemi Golpayegani:
2035 Layla and Majnun: A complex love story. In: Nonlinear Dynamics 83 (2016), i. 1, pp.
2036 615–622. [Nachweis im GVK]
2037 Harshita Jhavar / Paramita Mirza: EMOFIEL: Mapping emotions of relationships in a
2038 story. In: Companion Proceedings of the The Web Conference 2018. (WWW’18, Lyon,
2039 23.-27.04.2018) Geneva 2018, pp. 243–246. DOI: 10.1145/3184558.3186989Matthew Lee Jockers / Ted Underwood: Text-mining the humanities. In: A New
2040 Companion to Digital Humanities. Ed. by Susan Schreibman / Ray Siemens / John
2041 Unsworth. Pondicherry 2016, pp. 291–306. [Nachweis im GVK]
2042 Dan R. Johnson: Transportation into a story increases empathy, prosocial behavior,
2043 and perceptual bias toward fearful expressions. In: Personality and Individual
2044 Differences 52 (2012), i. 2, pp. 150–155. [Nachweis im GVK]
2045 Philip Nicholas Johnson-Laird / Keith Oatley: The language of emotions: An
2046 analysis of a semantic field. In: Cognition and emotion 3 (1989), i. 2, pp. 81–123.
2047 [Nachweis im GVK]
2048 Philip Nicholas Johnson-Laird / Keith Oatley: Emotions in Music, Literature, and
2049 Film. In: Handbook of emotions. Ed. by Lisa Feldman Barret / Michael Lewis /
2050 Jeannette M. Haviland-Jones. 4. edition. New York, NY et al. 2016. pp. 82–97.
2051 [Nachweis im GVK]
2052 Tuomo Kakkonen / Gordana Galic Kakkonen: Sentiprofiler: Creating comparable visual
2053 profiles of sentimental content in texts. In: Proceedings of the Workshop on Language
2054 Technologies for Digital Humanities and Cultural Heritage. Ed. by Cristina Vertan
2055 / Milena Slavcheva / Petya Osenova / Stelios Piperidis. (DigHum / RANLP: 8, Hissar,
2056 16.09.2011) Shoumen 2011, pp. 62–69. PDF. [online]
2057 [Nachweis im GVK]
2058 Evgeny Kim / Roman Klinger: Who feels what and why? Annotation of a literature
2059 corpus with semantic roles of emotions. In: Proceedings of the 27th International
2060 Conference on Computational Linguistics. (COLING: 27, Santa Fe, NM, 20.-26.08.2018)
2061 Stroudsburg, PA 2018, pp. 1345–1359. PDF. [online]
2062 Evgeny Kim / Roman Klinger (2019a): An analysis of emotion communication channels
2063 in fan-fiction: Towards emotional storytelling. In: Proceedings of the Second
2064 Workshop of Storytelling. Ed. by Francis Ferraro / Ting-Hao ›Kenneth‹ Huang / Stephanie
2065 M. Lukin / Margaret Mitchell. (Florence, 01.08.2019) Stroudsburg, PA 2019. DOI: 10.18653/v1/W19-3406Evgeny Kim / Roman Klinger (2019b): Frowning Frodo, wincing Leia, and a seriously
2066 great friendship: Learning to classify emotional relationships of fictional
2067 characters. In: Proceedings of the 2019 Conference of the North American Chapter of
2068 the Association for Computational Linguistics: Human Language Technologies. Volume
2069 1,
2070 Long and Short Papers. (NAACL-HLT, Minneapolis, MN, 02.-07.06.2019) Stroudsburg, PA
2071 2019, pp. 647–653. DOI: 10.18653/v1/N19-1067Evgeny Kim / Sebastian Padó / Roman Klinger (2017a): Investigating the
2072 relationship between literary genres and emotional plot development. In: Joint SIGHUM
2073 Workshop on Computational Linguistics for Cultural Heritage, Social Sciences,
2074 Humanities and Literature - proceedings of the workshop. (SIGHUM, Vancouver, 04.08.2017)
2075 Stroudsburg, PA 2017, pp. 17–26. DOI: 10.18653/v1/W17-2203Evgeny Kim / Sebastian Padó / Roman Klinger (2017b): Prototypical emotion
2076 developments in adventures, romances, and mystery stories. In: Digital Humanities
2077 2017: Conference Abstracts. Ed. by Rhian Lewis / Cecily Raynor / Dominic Forest /
2078 Michael Sinatra / Stéfan Sinclair. (DH 2017, Montreal, 08.-11.08.2017) Montreal 2017,
2079 pp. 288–291. PDF. [online]
2080 Suin Kim / JinYeong Bak / Alice Haeyun Oh: Do you feel what I feel? Social aspects
2081 of emotions in twitter conversations. In: Proceedings of the Sixth International AAAI
2082 Conference on Weblogs and Social Media. (ICWSM: 6, Dublin 04.-07.12.2012) Palo Alto,
2083 CA 2012, pp. 495–498. [Nachweis im GVK]
2084 Roman Klinger / Surayya Samat Suliya / Nils Reiter: Automatic Emotion Detection
2085 for Quantitative Literary Studies – A case study based on Franz Kafka’s “Das Schloss”
2086 and “Amerika”. In: Digital Humanities 2016: Conference Abstracts. Ed. by Maciej Eder
2087 / Jan Rybicki. (DH 2016, Kraków. 11.-16.07.2016) Kraków 2016, pp. 826–828. PDF. [online]
2088 Corina Koolen: Women’s books versus books by women. Digital Humanities 2018:
2089 Puentes-Bridges. Book of Abstracts. Hg. von Jonathan Girón Palau / Isabel Galina
2090 Russell. (DH 2018, Mexico City, 26.-29.06.2018) Mexico City 2018, pp. 219–222. PDF.
2091 [online]
2092 Eve Kraicer / Andrew Piper: Social characters: The hierarchy of gender in
2093 contemporary English-language fiction. In: Journal of Cultural Analytics (2019).
2094 Article from 30.01.2019. DOI: 10.22148/16.032Päivi Kuivalainen: Emotions in narrative: A linguistic study of Katherine
2095 Mansfield’s short fiction. In: The Electronic Journal of the Department of English
2096 at
2097 the University of Helsinki 5 (2009). [online]
2098 Richard A. Lanham: The electronic word: Literary study and the digital revolution.
2099 In: New Literary History 20 (1989), i. 2, pp. 265–290. [Nachweis im GVK]
2100 Randy J. Larsen / Edward Diener: Promises and problems with the circumplex model
2101 of emotion. In: Emotion. Ed. by Margaret S. Clark. (= Review of personality and
2102 social psychology, 13) Newbury Park et al. 1992, pp. 25–29. [Nachweis im GVK]
2103 Inger Leemans / Janneke M. van der Zwaan / Isa Maks / Erika Kuijpers / Kristine
2104 Steenbergh: Mining embodied emotions: a comparative analysis of sentiment and emotion
2105 in dutch texts, 1600–1800. In: Digital Humanities Quaterly 11 (2017), i. 4. [online]
2106 Bing Liu: Sentiment Analysis: mining opinions, sentiments, and emotions. New York,
2107 NY 2015. [Nachweis im GVK]
2108 Bing Liu: Sentiment analysis and subjectivity. In: Handbook of natural language
2109 processing. Ed. by Nitin Indurkhya / Fred Jacob Damerau. 2. edition. Boca Raton, FL
2110 2010, pp. 627–666. [Nachweis im GVK]
2111 Mika V. Mäntylä / Daniel Graziotin / Miikka Kuutila: The evolution of sentiment
2112 analysis – a review of research topics, venues, and top cited papers. In: Computer
2113 Science Review 27 (2018), pp. 16–32. [Nachweis im GVK]
2114 Raymond A. Mar / Keith Oatley / Maja Djikic / Justin Mullin: Emotion and narrative
2115 fiction: Interactive influences before, during, and after reading. In: Cognition
2116 & Emotion 25 (2011), i. 5, pp. 818–833. [Nachweis im GVK]
2117 Alessandro Marchetti / Rachele Sprugnoli / Sara Tonelli: Sentiment analysis for
2118 the humanities: the case of historical texts. In: Digital Humanities 2014: Conference
2119 Abstracts. (DH 2014, Lausanne 08.-12.07.2014), Lausanne 2014, pp. 254–257. PDF. [online] [Nachweis im GVK]
2120 Seth A. Marvel / Jon Kleinberg / Robert D. Kleinberg / Steven H. Strogatz:
2121 Continuous-time model of structural balance. In: Proceedings of the National Academy
2122 of Sciences 108 (2011), i. 5, pp. 1771–1776. DOI: 10.1073/pnas.1013213108 [Nachweis im GVK]
2123 John D. Mayer / Richard D. Roberts / Sigal G. Barsade: Human abilities: Emotional
2124 intelligence. In: Annual Review of Psychology 59 (2008), i. 1, pp. 507–536.
2125 [Nachweis im GVK]
2126 Jacques M. van Meel: Representing emotions in literature and paintings: a
2127 comparative analysis. In: Poetics 23 (1995), i. 1–2, pp. 159–176. [Nachweis im GVK]
2128 Saif M. Mohammad: From once upon a time to happily ever after: Tracking emotions
2129 in novels and fairy tales. In: Proceedings of the 5th ACL-HLT Workshop on Language
2130 Technology for Cultural Heritage, Social Sciences, and Humanities. Ed. by Kalliopi
2131 Zervanou / Piroska Lendvai. (ACL-HT: 5, Portland, OR, 23.-24.06.2011) Stroudsburg,
2132 PA
2133 2011, pp. 105–114. PDF. [online]
2134 Saif M. Mohammad: From once upon a time to happily ever after: Tracking emotions
2135 in mail and books. In: Decision Support Systems 53 (2012), i. 4, pp. 730–741.
2136 [Nachweis im GVK]
2137 Saif M. Mohammad / Peter D. Turney: Crowdsourcing a word–emotion association
2138 lexicon. In: Computational Intelligence 29 (2013), i. 3, pp. 436–465.
2139 [Nachweis im GVK]
2140 Franco Moretti: Graphs, maps, trees: abstract models for a literary history.
2141 London et al. 2005. [Nachweis im GVK]
2142 Olivier Morin / Alberto Acerbi: Birth of the cool: a two-centuries decline in
2143 emotional expression in anglophone fiction. In: Cognition and Emotion 31 (2017), i.
2144 8, pp. 1663–1675. [Nachweis im GVK]
2145 Eric T. Nalisnick / Henry S. Baird (2013a): Character-to-character sentiment
2146 analysis in shakespeare’s plays. In: Proceedings of the 51st Annual Meeting of the
2147 Association for Computational Linguistics. Ed. by Hinrich Schuetze / Pascale Fung
2148 / Massimo Poesio. 3 volumes. (ACL: 51, Sofia, 04.-09.08.2013) Red Hook, NY et al.
2149 2013. Vol. 2: Short Papers, pp. 479–483. [online]
2150 [Nachweis im GVK]
2151 Eric T. Nalisnick / Henry S. Baird (2013b): Extracting sentiment networks from
2152 shakespeare’s plays. In: 12th International Conference on Document Analysis and
2153 Recognition. (ICDAR: 12, Washington, DC, 25.-28.08.2013) Piscataway, NJ 2013, pp.
2154 758–762. [Nachweis im GVK]
2155 Finn Årup Nielsen: AFINN Sentiment Lexicon. In: 2011. [online]
2156 Mary Beth Oliver: Tender affective states as predictors of entertainment
2157 preference. In: Journal of Communication 58 (2008), i. 1, pp. 40–61. [Nachweis im GVK]
2158 Viviana Patti / Federico Bertola / Antonio Lieto: Arsemotica for
2159 Emotion-driven exploration of online art collections. In: The Twenty-Eighth
2160 International Florida Artificial Intelligence Research Society Conference. Ed. by
2161 Ingrid Russell / William Eberle. (FLAIRS: 28, Hollywood, 18.-28.05.2015) Palo Alto,
2162 CA, pp. 288–293. [Nachweis im GVK]
2163 James W. Pennebaker / Cindy K. Chung / Molly Ireland / Amy Gonzales / Roger
2164 J. Booth: The development and psychometric properties of LIWC2007. In: LIWC2007
2165 Manual. 2007. PDF. [online]
2166 Emanuele Pianta / Luisa Bentivogli / Christian Girardi: MultiWordNet: Developing
2167 an aligned multilingual database. In: Proceedings of 1st International Global WordNet
2168 Conference. (GWC: 1, Mysore, 21.-25.02.2002) Mysore 2002, pp. 293–302. [online]
2169 [Nachweis im GVK]
2170 Andrew Piper / Richard Jean So: Quantifying the weepy bestseller. In: The New
2171 Rebublic. Article from 18.12.2015. [online]
2172 Plato: Plato in Twelve Volumes. Cambridge, MA 1969. Siehe auch [Nachweis im GVK]
2173 Jonathan Posner / James Russell / Bradley Peterson: The circumplex model of affect:
2174 An integrative approach to affective neuroscience, cognitive development, and psychopathology.
2175 In: Development and psychopathology 17 (2005), i. 3, pp. 715-734. [Nachweis im GVK]
2176 Robert Plutchik: The Emotions. Revided edition. Lanham et al. 1991.
2177 [Nachweis im GVK]
2178 Robert Plutchik: Wheel of Emotions, 12.02.2011. In: Wikipedia, the free Encyclopedia:
2179 Robert
2180 Plutchik. Article from 20.09.2019. [online]
2181 Project Gutenberg. Ed. by Project Gutenberg Literary Archive Foundation. In:
2182 Salt Lake City, UT 1971-. [online]
2183 Andrew J. Reagan / Lewis Mitchell / Dilan Kiley / Christopher M. Danforth /
2184 Peter Sheridan Dodds: The emotional arcs of stories are dominated by six basic
2185 shapes. In: EPJ Data Science 5 (2016), i. 1, pp. 31–43. DOI: 10.1140/epjds/s13688-016-0093-1Ethan Reed: Measured unrest in the poetry of the black arts movement. Digital
2186 Humanities 2018: Puentes-Bridges. Book of Abstracts. Hg. von Jonathan Girón Palau
2187 /
2188 Isabel Galina Russell. (DH 2018, Mexico City, 26.-29.06.2018) Mexico City 2018, pp.
2189 477–478. PDF. [online]
2190 REMAN - Relational Emotion Annotation for Fiction. Relational EMotion ANnotation –
2191 a corpus with 1720 fictional text exceprts from the Project Gutenberg.
2192 Ed. by Evgeny Kim / Roman Klinger, Universität Stuttgart, Institut für Maschinelle
2193 Sprachverarbeitung. In: Institut für Maschinelle Sprachverarbeitung.
2194 Forschung. Ressourcen
2195 Korpora. Stuttgart 2018. [online]
2196 Marsha L. Richins: Measuring emotions in the consumption experience. In: Journal
2197 of consumer research 24 (1997), i. 2, pp. 127–146. [Nachweis im GVK]
2198 Sergio Rinaldi / Pietro Landi / Fabio Della Rossa: Small discoveries can have
2199 great consequences in love affairs: the case of Beauty and the Beast. In:
2200 International Journal of Bifurcation and Chaos 23 (2013), i. 11. [Nachweis im GVK]
2201 Jenefer Robinson: Deeper than reason: Emotion and its role in literature, music,
2202 and art. New York, NY 2005. [Nachweis im GVK]
2203 Catherine Sheldrick Ross: Finding without seeking: the information encounter in
2204 the context of reading for pleasure. In: Information Processing & Management 35
2205 (1999), i. 6., pp. 783–799. [Nachweis im GVK]
2206 James A. Russell: A circumplex model of affect. In: Journal of Personality and
2207 Social Psychology 39 (1980), pp. 1161–1178. [Nachweis im GVK]
2208 James A. Russell: Is there universal recognition of emotion from facial
2209 expression? A review of the cross-cultural studies. In: Psychological bulletin 115
2210 (1994), i. 1, pp. 102–141. [Nachweis im GVK]
2211 James A. Russell: Core affect and the psychological construction of emotion. In:
2212 Psychological review 110 (2003), i. 1, pp. 145–172. [Nachweis im GVK]
2213 James A. Russell / Lisa Feldman Barrett: Core affect, prototypical emotional
2214 episodes, and other things called emotion: dissecting the elephant. In: Journal of
2215 Personality and Social Psychology 76 (1999), i. 5, pp. 805–819. [Nachweis im GVK]
2216 James A. Russell / Jo-Anne Bachorowski / José-Miguel Fernández-Dols: Facial and
2217 vocal expressions of emotion. In: Annual review of psychology 54 (2003), i. 1, pp.
2218 329–349. [Nachweis im GVK]
2219 Exploring Text and Emotions. Ed. by Lars Sætre / Patrizia Lombardo / Julien
2220 Zanetta (2014a). Aarhus 2014. [Nachweis im GVK]
2221 Lars Sætre / Patrizia Lombardo / Julien Zanetta (2014b): Text and Emotions. In:
2222 Exploring Text and Emotions. Ed. by Lars Sætre / Patrizia Lombardo / Julien Zanetta.
2223 Aarhus 2014, pp. 9–26. [Nachweis im GVK]
2224 Spyridon Samothrakis / Maria Fasli: Emotional sentence annotation helps predict
2225 fiction genre. In: PLOS ONE 10 (2015), i. 11, p. e0141922. Article from 02.11.2015.
2226 DOI: 10.1371/journal.pone.0141922Dalya Samur / Mattie Tops / Sander L. Koole: Does a single session of reading
2227 literary fiction prime enhanced mentalising performance? Four replication experiments
2228 of Kidd and Castano (2013). In: Cognition & Emotion 32 (2018), pp. 130–144.
2229 [Nachweis im GVK]
2230 Andrea Scarantino: The Phylosophy of Emotions and Its Impact on Affective
2231 Sciences. In: Handbook of emotions. Ed. by Lisa Feldman Barret / Michael Lewis /
2232 Jeannette M. Haviland-Jones. 4. edition. New York, NY et al. 2016. pp. 3–49.
2233 [Nachweis im GVK]
2234 Klaus R. Scherer: What are emotions? And how can they be measured? In: Social
2235 Science Information 44 (2005), i. 4, pp. 695–729. [Nachweis im GVK]
2236 Susan Schreibman / Ray Siemens / John Unsworth: A New Companion to Digital
2237 Humanities. Chichester et al. 2015/2016. [Nachweis im GVK]
2238 Norbert Schwarz: Emotion, cognition, and decision making. In: Cognition &
2239 Emotion 14 (2000), i. 4, pp. 433–440. [Nachweis im GVK]
2240 Herman Smith / Andreas Schneider: Critiquing models of emotions. In: Sociological
2241 Methods & Research 37 (2009), i. 4, pp. 560–589. [Nachweis im GVK]
2242 Mohammad Soleymani / David Garcia / Brendan Jou / Björn Schuller / Shih-Fu Chang /
2243 Maja Pantic: A survey of multimodal sentiment analysis. In: Image and Vision
2244 Computing 65 (2017), pp. 3–14. [Nachweis im GVK]
2245 Ronald de Sousa / Andrea Scarantino: Emotion. In: The Stanford Encyclopedia of
2246 Philosophy. Ed. by Edward N. Zalta. Stanford, CA 2018. Article from 25.09.2018. [online]
2247 Rachele Sprugnoli / Sara Tonelli / Alessandro Marchetti / Giovanni Moretti:
2248 Towards sentiment analysis for historical texts. In: Digital Scholarship in the
2249 Humanities 31 (2016), i. 4, pp. 762–772. DOI: 10.1093/llc/fqv027 [Nachweis im GVK]
2250 Philip J. Stone / Dexter C. Dunphy / Marshall S. Smith: The General Inquirer: A
2251 computer approach to content analysis. In: American Journal of Sociology 73 (1968),
2252 i. 5, pp. 634–635. [Nachweis im GVK]
2253 Carlo Strapparava / Alessandro Valitutti. WordNet-Affect: An affective extension
2254 of WordNet. In: Proceedings of the 4th International Conference on Language Resources
2255 and Evaluation. Ed. by Maria Teresa Lino / Maria Francisca Xavier / Fátima Ferreira
2256 /
2257 Rute Costa / Raquel Silva. 9 volumes. (LREC: 4, Lisbon, 24.-30.05.2004) Paris et al.
2258 2004. Vol. 4, pp. 1083–1086. PDF. [online]
2259 [Nachweis im GVK]
2260 Jared Suttles / Nancy Ide: Distant supervision for emotion classification with
2261 discrete binary values. In: Computational Linguistics and Intelligent Text
2262 Processing. Ed. by Alexander Gelbukh. 2 volumes. (CICLing: 14, Samos, 24.-30.03.2013)
2263 Berlin et al. 2013. Vol. 2, pp. 121–136. [Nachweis im GVK]
2264 Maite Taboada / Mary Ann Gillies / Paul McFetridge: Sentiment classification
2265 techniques for tracking literary reputation. In: LREC workshop: Towards computational
2266 models of literary analysis. (LREC: 5, Genoa, 22.-28.05.2006) , pp. 36–43. Paris
2267 2006. [online]
2268 Maite Taboada / Mary Ann Gillies / Paul McFetridge / Robert Outtrim: Tracking
2269 literary reputation with text analysis tools. In: Meeting of the Society for Digital
2270 Humanities. Vancouver 2008. PDF. [online]
2271 Leo Tolstoy: What is art? And essays on art. Harmondsworth 1962. (= Penguin
2272 classics) Siehe auch [Nachweis im GVK]
2273 Silvan Tomkins: Affect imagery consciousness. 4 vol. New York, NY et al. 1962.
2274 Vol. I: The positive affects. [Nachweis im GVK]
2275 Leigh Van Horn: The characters within us: Readers connect with characters to
2276 create meaning and understanding. In: Journal of Adolescent & Adult Literacy 40
2277 (1997), i. 5, pp. 342–347. [Nachweis im GVK]
2278 Edward Vanhoutte: The gates of hell: History and definition of
2279 digital|humanities|computing. In: Defining Digital Humanities. A Reader. Ed. by
2280 Meliss Terras / Julianne Hyhan / Edward Vanhoutte. Farnham 2013, pp. 119–156.
2281 [Nachweis im GVK]
2282 Kurt Vonnegut: Kurt Vonnegut at the Blackboard. Ed. by Seven Stories Press. New York,
2283 NY 2005.
2284 In: Lapham’s Quarterly (2010). Article from 26.03.2010. [online]
2285 Bei Yu: An evaluation of text classification methods for literary study. In:
2286 Literary and Linguistic Computing 23 (2008), i. 3, pp. 327–343. DOI: 10.1093/llc/fqn015Albin Zehe / Martin Becker / Lena Hettinger / Andreas Hotho / Isabella Reger /
2287 Fotis Jannidis: Prediction of happy endings in German novels based on sentiment
2288 information. In: Proceedings of the Workshop on Interactions between Data Mining and
2289 Natural Language Processing 2016. Ed. by Peggy Cellier / Thierry Charnois / Andreas
2290 Hotho / Stan Matwin / Marie-Francine Moens / Yannick Toussaint. (DMNLP: 3, Riva del
2291 Garda, 19.-23.09.2016) Aachen 2016, pp. 9–16. URN: urn:nbn:de:0074-1646-4Mikhail Zhuravlev / Irina Golovacheva / Polina de Mauny: Mathematical modelling of
2292 love affairs between the characters of the pre-masochistic novel. In: 2014 Second
2293 World Conference on Complex Systems (WCCS: 2, Adagir, 10.-12.11.2014) Piscataway,
2294 NJ 2014,
2295 pp. 396–401. [Nachweis im GVK]
2296 Dolf Zillmann / Richard T. Hezel / Norman J. Medoff: The effect of affective
2297 states on selective exposure to televised entertainment fare. In: Journal of Applied
2298 Social Psychology 10 (1980), i. 4, pp. 323–339. [Nachweis im GVK]
2303 List of Figures with Captions
2306 Fig. 1: Plutchik’s wheel of emotions. [Plutchik 2011.
2307 PD]
2310 Fig. 2: Circumplex model of affect: Horizontal axis represents the valence dimension,
2311 the vertical axis represents the arousal dimension. Drawn after Posner et al. 2005. [Kim / Klinger 2019]
2314 Tab. 3: Summary of characteristics of methods used in the papers reviewed
2315 in this survey. Download as PDF. [Kim / Klinger 2019]
2317 18