CROWDSOURCING A WORD–EMOTION ASSOCIATION LEXICON

Saif M. MohammadPeter D. Turney

article2013International Conference on Climate Informatics2,721 citations

Presents a practical crowdsourcing methodology for constructing large-scale word-emotion lexicons, proving that sense-verification questions and association-based framing substantially increase annotation quality and inter-annotator agreement.

Listen

The article addresses the scarcity of large, high-quality emotion lexicons compared to polarity resources, which limits progress in sentiment analysis for applications such as customer relations management, search engines, tutoring systems, and literary analysis. Emotions manifest in language through words, yet manual expert annotation has proven too costly and slow to build comprehensive resources.

The work set out to create a large English termemotion association lexicon quickly and inexpensively by harnessing crowdsourcing while solving quality-control challenges inherent to the method.

Researchers used Amazon’s Mechanical Turk to obtain annotations for more than 10,000 wordsense pairs drawn from frequent unigrams, bigrams, the General Inquirer, and the WordNet Affect Lexicon. They presented each term with a word-choice question to convey sense and filter out unfamiliar or malicious responses, then collected five independent ratings per term for eight basic emotions and for positive or negative polarity. Pilot experiments compared question phrasing, and post-processing retained only high-quality assignments.

The resulting lexicon, EmoLex, shows that roughly 36 percent of terms evoke at least one emotion and 30 percent carry strong positive or negative polarity. Adjectives and adverbs are most often emotive; trust and joy appear most frequently among the eight emotions. Annotator agreement reached fair to substantial levels, with at least four of five workers concurring on the majority of terms. Annotations aligned closely with existing gold-standard resources, and about 9 percent of terms were judged to name emotions directly. Certain emotions such as anger and sadness co-occur more often than others.

These results demonstrate that carefully designed crowdsourcing can produce reliable emotion data at low cost, enabling practical systems that detect customer anger, identify trusted products, or track emotional arcs in text. The findings also clarify which parts of speech and which emotion pairs warrant priority in downstream models.

The authors recommend expanding coverage to 40,000 terms, building parallel lexicons in other languages, testing the resource in live applications, and adopting maximum-difference scaling to further improve annotation quality. They note that word-sense disambiguation and context handling remain necessary before the lexicon reaches full utility.

The main limitations are that annotations reflect prior associations rather than specific contexts and that agreement is lower for emotions such as anticipation. Overall confidence in the core findings is high because of the large sample, multiple validation checks, and agreement with established lexicons.

arXiv: 1308.6297
Cover for CROWDSOURCING A WORD–EMOTION ASSOCIATION LEXICON

Abstract

Even though considerable attention has been given to the polarity of words (positive and negative) and the creation of large polarity lexicons, research in emotion analysis has had to rely on limited and small emotion lexicons. In this paper we show how the combined strength and wisdom of the crowds can be used to generate a large, high-quality, word-emotion and word-polarity association lexicon quickly and inexpensively. We enumerate the challenges in emotion annotation in a crowdsourcing scenario and propose solutions to address them. Most notably, in addition to questions about emotions associated with terms, we show how the inclusion of a word choice question can discourage malicious data entry, help identify instances where the annotator may not be familiar with the target term (allowing us to reject such annotations), and help obtain annotations at sense level (rather than at word level). We conducted experiments on how to formulate the emotion-annotation questions, and show that asking if a term is associated with an emotion leads to markedly higher inter-annotator agreement than that obtained by asking if a term evokes an emotion.

Table of Contents

  • 1 Introduction
  • 2 Applications
  • 3 Emotions
  • 4 Related work
  • 5 Target terms
  • 6 Mechanical Turk
  • 7 Issues with crowdsourcing and emotion annotation
  • 7.1 Key issues in crowdsourcing
  • 7.2 Finer points of emotion annotation
  • 8 Our approach
  • 9 Annotation Statistics and Post-Processing
  • 10 Analysis of Emotion Annotations
  • 10.1 Discussion
  • 10.2 Agreement
  • 10.3 Evokes versus Associated
  • 11 Analysis of Polarity Annotations
  • 11.1 Discussion
  • 11.2 Agreement
  • 12 Conclusions
  • 13 Future Directions
  • References

Knowls

  1. Knowl 1 — Quality-Controlled Crowdsourced Word-Sense Emotion Annotation Method

    model/method

    To collect emotion and polarity associations for target terms at the sense level while mitigating spam and careless entries from crowdsourcing, each Human Intelligence Task (HIT) on Amazon Mechanical Turk is structured around two sequential components:

    1. Word Choice Sense-Guiding Question: Before answering emotion questions, annotators are presented with a 4-choice vocabulary problem asking which word is closest in meaning to the target term. The correct answer is the head word of the Macquarie Thesaurus category corresponding to the target sense, while the remaining three options are randomly chosen head words from other categories acting as distractors. This question serves three functions:
      • It specifies the intended word sense without requiring annotators to read long dictionary definitions.
      • It filters out annotators unfamiliar with the target word (random guessers have a 75% error rate).
      • It detects automated bots or random clickers.
    2. Polarity and Emotion Association Questions: Annotators rate the target term on a 4-point ordinal scale (not, weakly, moderately, strongly):
      • Polarity ratings: positive and negative semantic orientation.
      • Emotion ratings across Plutchik's eight basic emotions: joy, sadness, anger, fear, trust, disgust, surprise, and anticipation.
      • Direct emotion identification: A binary question asking whether the word is itself an emotion (e.g., love is an emotion, whereas shark is associated with fear but is not an emotion).

    Each HIT is completed independently by 5 annotators who must be native or fluent English speakers.

  2. Knowl 2 — Effect of Question Framing: Emotion Association vs. Emotion Evocation

    empirical result

    The phrasing of emotion annotation prompts significantly influences crowdsourced inter-annotator agreement. In a pilot study comparing 2,100 terms annotated under two phrasing conditions—asking whether a word is associated with an emotion versus whether it evokes an emotion—asking about association resulted in consistently higher inter-annotator agreement:

    Emotion Evokes (%) Associated (%)
    Anger 61.6 68.2
    Anticipation 34.8 49.6
    Disgust 65.4 66.4
    Fear 62.0 59.4
    Joy 54.6 62.3
    Sadness 66.7 65.3
    Surprise 54.0 67.3
    Trust 47.3 49.8
    Micro-average 55.8 61.0

    The table reports the percentage of terms in the pilot set for which all 5 annotators achieved complete unanimous agreement on binary classification (emotive vs. non-emotive). Asking for associated emotions yields an overall micro-average increase of 5.2 percentage points in unanimous consensus (61.0% vs. 55.8%), with substantial improvements for anticipation (+14.8%), surprise (+13.3%), joy (+7.7%), and anger (+6.6%). This discrepancy arises because asking what an emotion evokes prompts annotators to draw upon idiosyncratic personal experiences, whereas asking what a word is associated with guides them toward shared linguistic and cultural intuitions.

  3. Knowl 3 — Post-Processing and Quality-Control Filtering Pipeline for Crowdsourced Lexicons

    algorithm

    To filter out random, malicious, and noisy responses from crowdsourced annotators on Mechanical Turk, raw submissions are processed through a multi-stage validation algorithm:

    Input: Set of N×5N \times 5 raw assignments for NN target term-sense pairs
    Output: Master set of validated term-sense emotion annotations
    for each assignment in raw assignments:
        if assignment has any unanswered question:
            discard and reject assignment (do not pay annotator)
    for each target term:
        if 3 or more annotators choose an answer different from the thesaurus head word on Q1:
            mark target term as having an ambiguous or flawed Q1 question
            discard all 5 assignments for this target term (annotators paid in full)
    for each annotator:
        calculate overall score on Q1 across all attempted HITs
        if overall Q1 score < 66.67%:
            discard and reject all assignments submitted by this annotator
    for each assignment remaining:
        if Q1 answer is incorrect:
            discard assignment
    for each annotator:
        compute maximum likelihood probability pp of agreeing with the majority on emotion questions
    compute mean μ\mu and standard deviation σ\sigma of pp across all annotators
    for each annotator:
        if p<μ2σp < \mu - 2\sigma:
            discard all assignments from this outlier annotator
    retain all terms having 3\ge 3 valid assignments in the final master set

    Applying this pipeline to 50,850 initial assignments (10,170 terms ×\times 5 annotators) resulted in:

    • Discarding 2,666 assignments containing unanswered questions.
    • Discarding 1,045 terms (5,225 assignments) due to flawed distractors in Q1.
    • Discarding all assignments from Turkers scoring below the 66.67% Q1 accuracy threshold.
    • Discarding assignments from 111 outlier Turkers scoring more than 2σ2\sigma below the mean majority agreement probability.
    • A finalized master set of 8,883 terms containing 38,726 assignments from 2,216 annotators (averaging 4.45 valid assignments per term).
  4. Knowl 4 — Aggregation and Binarization of Crowdsourced Emotion and Polarity Ratings

    model/method

    Crowdsourced emotion and polarity annotations collected on a 4-point ordinal scale (no, weak, moderate, strong) are aggregated and converted into binary categories for natural language processing applications:

    1. Ordinal Consensus (Majority Intensity): For each term–emotion and term–polarity pair, the consolidated intensity is the majority intensity level chosen across all valid assignments. Ties between intensity levels are resolved by selecting the stronger intensity level.
    2. Binary Conversion (Emotive vs. Non-Emotive / Evaluative vs. Non-Evaluative):
      • The no and weak intensity ratings are grouped into the non-emotive (or non-evaluative) bin.
      • The moderate and strong intensity ratings are grouped into the emotive (or evaluative) bin.
      • The final binary classification is assigned according to whichever bin contains the majority of annotator assignments.
  5. Knowl 5 — Target Term Composition of the EmoLex Word-Emotion Lexicon

    data/table

    The NRC Word-Emotion Association Lexicon (EmoLex) dataset covers 10,170 initially selected term–sense pairs, yielding 8,883 validated terms in the master set after automated quality control:

    Subset Initial Terms Master Terms Annotations / Term
    EmoLex-Uni (Thesaurus Unigrams)
    Adjectives 200 190 4.4
    Adverbs 200 187 4.5
    Nouns 200 178 4.5
    Verbs 200 195 4.4
    EmoLex-Bi (Thesaurus Bigrams)
    Adjectives 200 162 4.4
    Adverbs 187 171 4.3
    Nouns 200 185 4.5
    Verbs 200 178 4.4
    EmoLex-GI (General Inquirer)
    Negative terms 2,119 1,837 4.4
    Neutral terms 4,226 3,653 4.4
    Positive terms 1,787 1,541 4.4
    EmoLex-WAL (WordNet Affect)
    Anger terms 165 160 4.5
    Disgust terms 37 34 4.4
    Fear terms 100 89 4.4
    Joy terms 165 149 4.5
    Sadness terms 120 112 4.5
    Surprise terms 53 51 4.4
    Union Total 10,170 8,883 4.45

    The target terms were selected from three primary sources:

    • The 200 most frequent monosemous unigrams and bigrams per part of speech from the Macquarie Thesaurus based on Google n-gram frequencies (187 adverb bigrams met these criteria).
    • All terms from the General Inquirer with at most three senses in the thesaurus (8,132 terms).
    • All terms from the Ekman subset of the WordNet Affect Lexicon with at most two senses (640 terms).
  6. Knowl 6 — Distribution of Emotion Associations Across Lexical Subsets and Parts of Speech

    data/table

    Consolidated binary emotion associations across the 8,883 master terms in EmoLex reveal distinct patterns across grammatical categories and emotion types:

    Subset Anger Anticip. Disgust Fear Joy Sadness Surprise Trust Any Emotion
    EmoLex (All) 13% 12% 10% 14% 16% 12% 6% 16% 54%
    Unigrams
    Adjectives 14% 14% 10% 13% 29% 14% 10% 15% 68%
    Adverbs 13% 20% 8% 10% 23% 11% 7% 23% 67%
    Nouns 7% 18% 3% 7% 16% 6% 3% 24% 46%
    Verbs 11% 21% 5% 16% 14% 11% 7% 15% 52%
    Bigrams
    Adjectives 12% 25% 8% 14% 30% 15% 8% 16% 66%
    Adverbs 6% 23% 1% 7% 19% 3% 9% 29% 54%
    Nouns 9% 23% 6% 14% 20% 9% 7% 29% 58%
    Verbs 8% 25% 5% 7% 21% 6% 3% 27% 60%

    Key empirical findings from this distribution include:

    • Overall, 54% of terms in EmoLex are associated with at least one emotion.
    • Trust (16%) and joy (16%) are the most common emotion associations, whereas surprise (6%) is the least common.
    • Adjectives (68%) and adverbs (67%) have the highest proportion of emotive terms, reflecting their syntactic role as qualifiers.
    • Nouns associate most frequently with trust (24% for unigrams, 29% for bigrams), whereas adjectives associate most frequently with joy (29% for unigrams, 30% for bigrams).
    • In total, 9.3% of target terms (826 terms) directly refer to emotions rather than merely being associated with them.
  7. Knowl 7 — Inter-Annotator Agreement on Word-Emotion Associations

    data/table

    Inter-annotator agreement for Plutchik's eight basic emotions across 8,883 master terms was evaluated at both the 4-level ordinal intensity scale and the binarized (emotive vs. non-emotive) level:

    4-Level Agreement (%) Binary Agreement (%) Fleiss' κ\kappa
    Emotion 3\ge 3 Agree 4\ge 4 Agree = 5 Agree 4\ge 4 Agree κ\kappa Interpretation
    Anger 86.1 64.4 67.2 86.6 0.39 Fair agreement
    Anticipation 80.7 49.0 48.4 81.0 0.14 Slight agreement
    Disgust 86.0 65.3 68.1 86.5 0.31 Fair agreement
    Fear 83.2 55.5 59.7 84.5 0.32 Fair agreement
    Joy 83.7 59.4 61.0 83.6 0.36 Fair agreement
    Sadness 85.4 61.6 66.9 87.1 0.39 Fair agreement
    Surprise 88.1 62.8 66.2 89.0 0.18 Slight agreement
    Trust 81.0 53.6 50.7 79.5 0.24 Fair agreement
    Micro-average 84.3 59.0 61.0 84.7 0.29 Fair agreement

    At the binary level, all five annotators reach unanimous consensus for 61.0% of terms, and at least four out of five annotators agree for 84.7% of terms. Fleiss' κ\kappa averages 0.29 across the eight emotions (fair agreement), with highest agreement on anger (κ=0.39\kappa = 0.39) and sadness (κ=0.39\kappa = 0.39), and lowest agreement on anticipation (κ=0.14\kappa = 0.14) and surprise (κ=0.18\kappa = 0.18). The κ\kappa values are conservative estimates because terms are presented out of context and the non-emotive class is heavily predominant.

  8. Knowl 8 — Polarity Annotation Distribution and Agreement in EmoLex

    data/table

    Polarity (positive and negative semantic orientation) was annotated alongside emotions for the 8,883 terms in EmoLex. Agreement and distribution metrics across 4-level and binary intensity categories are as follows:

    Polarity Evaluative (%) 4-Level 4\ge 4 (%) Binary = 5 (%) Binary 4\ge 4 (%) Fleiss' κ\kappa Interpretation
    Negative 30.0 59.7 66.1 88.4 0.62 Substantial agreement
    Positive 35.0 47.8 49.3 75.6 0.45 Moderate agreement
    Combined / Avg. 65.0 (either) 53.8 57.7 82.0 0.54 Moderate agreement

    Key empirical findings include:

    • Overall, 65% of terms carry polarity (35% positive, 30% negative).
    • Unigram nouns are heavily skewed toward positive polarity (39% positive vs. 8% negative), suggesting that default noun concepts are positive or neutral, requiring modifying adjectives to become negative.
    • Negative polarity exhibits substantially higher inter-annotator agreement (κ=0.62\kappa = 0.62, substantial agreement) than positive polarity (κ=0.45\kappa = 0.45, moderate agreement). This occurs because the semantic boundary between positive and neutral words is more subjective and fuzzy than the boundary between negative and neutral words.
  9. Knowl 9 — Validation of Crowdsourced Emotion and Polarity Annotations Against Gold-Standard Lexicons

    empirical result

    To evaluate the quality of non-expert crowdsourced annotations, subsets of EmoLex originating from expert-curated resources were compared against their original gold-standard labels:

    1. WordNet Affect Lexicon (WAL) Comparison:
      • WAL anger words: 83% annotated as anger by Turkers (and 53% disgust, 18% fear, 0% joy).
      • WAL disgust words: 94% annotated as disgust (and 44% anger, 14% fear, 0% joy).
      • WAL fear words: 74% annotated as fear (and 20% sadness, 19% disgust, 17% anger, 1% joy).
      • WAL joy words: 78% annotated as joy (and 28% trust, 14% anticipation, 2% anger).
      • WAL sadness words: 94% annotated as sadness (and 13% disgust, 13% fear, 9% anger, 0% joy).
      • WAL surprise words: 66% annotated as surprise (and 42% joy, 26% positive, 8% fear).
      • Terms annotated as anger in WAL but rejected as anger by Turkers (baffled, exacerbate, gravel, pesky, pestering) were confirmed not to be genuine anger words.
    2. General Inquirer (GI) Comparison:
      • GI negative words: 83% marked negative and 1% positive by Turkers, associating strongly with anger (36%), fear (34%), sadness (33%), and disgust (29%), with 0% joy.
      • GI positive words: 82% marked positive and 2% negative by Turkers, associating strongly with joy (40%), trust (33%), and anticipation (13%), with 1%\le 1\% negative emotions.
      • GI neutral words: 12% marked negative and 30% positive by Turkers, reflecting the fuzzier boundary of neutral/positive terms.

    These results confirm that crowdsourcing with automated quality-control mechanisms produces annotations highly concordant with expert lexicons while capturing natural emotion co-occurrences (e.g., anger with disgust, joy with trust, and surprise with joy).

Coverage note — Deliberately omitted general introductory reviews on sentiment analysis applications (e.g., customer relationship management details), broad historical/psychological emotion taxonomy surveys, and prospective future work descriptions (e.g., Maximum Difference Scaling proposals).

References

  1. 1.Adamic, L. A., Zhang, J., Bakshy, E., and Ackerman, M. S. (2008). Knowledge sharing and yahoo answers: everyone knows something. In Proceeding of the 17th international conference on World Wide Web, WWW ’08, pages 665–674, New York, NY, USA. ACM.
  2. 2.Alm, C. O., Roth, D., and Sproat, R. (2005). Emotions from text: Machine learning for text-based emotion prediction. In Proceedings of the Joint Conference on Human Language Technology / Empirical Methods in Natural Language Processing (HLT/EMNLP-2005), pages 579–586, Vancouver, Canada.
  3. 3.Aman, S. and Szpakowicz, S. (2007). Identifying expressions of emotion in text. In V. Matoušek and P. Mautner, editors, Text, Speech and Dialogue, volume 4629 of Lecture Notes in Computer Science, pages 196–205. Springer Berlin / Heidelberg.
  4. 4.Aristotle (1913). Physiognomonica. In W. D. Ross, editor, The Works of Aristotle, pages 805–813. Oxford, England: Clarendon. Translated by T. Loveday and E. S. Forster.
  5. 5.Bellegarda, J. (2010). Emotion analysis using latent affective folding and embedding. In Proceedings of the NAACL-HLT 2010 Workshop on Computational Approaches to Analysis and Generation of Emotion in Text, Los Angeles, California.
  6. 6.Bernard, J., editor (1986). The Macquarie Thesaurus. Macquarie Library, Sydney, Australia.
  7. 7.Boucouvalas, A. C. (2002). Real time text-to-emotion engine for expressive internet communication. Emerging Communication: Studies on New Technologies and Practices in Communication, 5, 305–318.
  8. 8.Bougie, J. R. G., Pieters, R., and Zeelenberg, M. (2003). Angry customers don’t come back, they get back: The experience and behavioral implications of anger and dissatisfaction in services. Open access publications from tilburg university, Tilburg University.
  9. 9.Bradley, M. and Lang, P. (1999). Affective norms for english words (anew): Instruction manual and affective ratings. In Technical Report, C-1 , The Center for Research in Psychophysiology, University of Florida.
  10. 10.Brants, T. and Franz, A. (2006). Web 1t 5-gram version 1. Linguistic Data Consortium.
  11. 11.Breazeal, C. and Brooks, R. (2004). Robot emotions: A functional perspective. In Who Needs Emotions. Oxford University Press.
  12. 12.Brennan, R. L. and Prediger, D. J. (1981). Coefficient Kappa: Some Uses, Misuses, and Alternatives. Educational and Psychological Measurement, 41(3), 687–699.
  13. 13.Callison-Burch, C. (2009). Fast, cheap and creative: Evaluating translation quality using amazon’s mechanical turk. In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP-2009), pages 286–295, Singapore.
  14. 14.Cohen, J. (1960). A Coefficient of Agreement for Nominal Scales. Educational and Psychological Measurement, 20(1), 37–46.
  15. 15.Cohen, S. H. and Associates, S. . (2003). Maximum difference scaling: Improved measures of importance and preference for segmentation. In Sawtooth Software Conference Proceedings, Sawtooth Software, Inc. 530 W. Fir St, pages 61–74.
  16. 16.Dabholkar, P. A., Shepherd, C. D., and Thorpe, D. I. (2000). A comprehensive framework for service quality: an investigation of critical conceptual and measurement issues through a longitudinal study. Journal of Retailing, 76(2), 139–173.
  17. 17.Darwin, C. (1872). The Expressions of the Emotions in Man and Animals. John Murray.
  18. 18.Davenport, T. H., Harris, J. G., and Kohli, A. K. (2001). How do they know their customers so well? 42(2), 63–73.
  19. 19.Díaz, A. B. C. and Ruz, F. J. M. (2002). The consumers reaction to delays in service. International Journal of Service Industry Management, 13(2), 118–140.
  20. 20.Dubé, L. and Maute, M. (1996). The antecedents of brand switching, brand loyalty and verbal responses to service failure. Advances in Services Marketing and Management, 5, 127–151.
  21. 21.Ekman, P. (1992). An argument for basic emotions. Cognition and Emotion, 6(3), 169–200.
  22. 22.Ekman, P. (2005). Emotion in the Human Face. Oxford University Press.
  23. 23.Ekman, P. and Friesen, W. V. (2003). Unmasking the Face: A Guide to Recognizing Emotions From Facial Expressions. Malor Books.
  24. 24.Elfenbein, H. A. and Ambady, N. (1994). Is there universal recognition of emotion from facial expression? a review of the cross-cultural studies. Psychological Bulletin, 115, 102–141.
  25. 25.Elliott, C. (1992). The affective reasoner: A process model of emotions in a multi-agent system. Ph.D. thesis, Institute for the Learning Sciences, Northwestern University.
  26. 26.Fleiss, J. L. (1971). Measuring nominal scale agreement among many raters. Psychological Bulletin, 76(5), 378–382.
  27. 27.Folkes, V. S., Koletsky, S., and Graham, J. L. (1987). A field study of causal inferences and consumer reaction: The view from the airport. Journal of Consumer Research: An Interdisciplinary Quarterly, 13(4), 534–39.
  28. 28.Francisco, V. and Gervás, P. (2006). Automated mark up of affective information in english texts. In P. Sojka, I. Kopecek, and K. Pala, editors, Text, Speech and Dialogue, volume 4188 of Lecture Notes in Computer Science, pages 375–382. Springer Berlin / Heidelberg.
  29. 29.Genereux, M. and Evans, R. P. (2006). Distinguishing affective states in weblogs. In AAAI-2006 Spring Symposium on Computational Approaches to Analysing Weblogs, pages 27–29, Stanford, California.
  30. 30.Goyal, A., Riloff, E., Daume III, H., and Gilbert, N. (2010). Toward plot units: Automatic affect state analysis. In Proceedings of the NAACL-HLT 2010 Workshop on Computational Approaches to Analysis and Generation of Emotion in Text, Los Angeles, California.
  31. 31.Gross, J. J. (1998). The emerging field of emotion regulation: An integrative review. Review of General Psychology, 2(3), 271–299.
  32. 32.Guo, K., Hall, C., Hall, S., K., M., and Mills, D. (2007). Left gaze bias in human infants, rhesus monkeys, and domestic dogs. Animal Cognition, 12, 409–418.
  33. 33.Harris, L. C. and Goode, M. M. H. (2004). The four levels of loyalty and the pivotal role of trust: a study of online service dynamics. Journal of Retailing, 80(2), 139–158.
  34. 34.Hollinger, G., Georgiev, Y., Manfredi, A., Maxwell, B. A., Pezzementi, Z. A., and Mitchell, B. (2006). Design of a social mobile robot using emotion-based decision mechanisms. In Intelligent Robots and Systems, 2006 IEEE/RSJ International Conference on, pages 3093–3098.
  35. 35.Holsti, O. R. (1969). Content analysis for the social sciences and humanities. Addison-Wesley, Reading, MA.
  36. 36.Holzman, L. E. and Pottenger, W. M. (2003). Classification of emotions in internet chat: An application of machine learning using speech phonemes. Technical report, Leigh University.
  37. 37.Howe, J. and Robinson, M. (2006). Crowdsourcing: A definition. In Crowdsourcing: Tracking the Rise of the Amateur . Weblog.
  38. 38.James, W. (1884). What is an emotion? Mind, 9, 188–205.
  39. 39.John, D., Boucouvalas, A. C., and Xu, Z. (2006). Representing emotional momentum within expressive internet communication. In Proceedings of the 24th IASTED international conference on Internet and multimedia systems and applications, pages 183–188, Anaheim, CA. ACTA Press.
  40. 40.Kalra, A. and Karahalios, K. (2005). Texttone: Expressing emotion through text. In M. F. Costabile and F. Patern, editors, Human-Computer Interaction - INTERACT 2005 , volume 3585 of Lecture Notes in Computer Science, pages 966–969. Springer Berlin / Heidelberg.
  41. 41.Knautz, K., Siebenlist, T., and Stock, W. G. (2010). Memose: search engine for emotions in multimedia documents. In Proceeding of the 33rd international ACM SIGIR conference on Research and development in information retrieval, SIGIR ’10, pages 791–792, New York, NY. ACM.
  42. 42.Kövecses, Z. (2003). Metaphor and Emotion: Language, Culture, and Body in Human Feeling (Studies in Emotion and Social Interaction). Cambridge University Press.
  43. 43.Landis, J. R. and Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33(1), 159–174.
  44. 44.Lazarus, R. S. (1984). On the primacy of cognition. American Psychologist, 39(2), 124–129.
  45. 45.Lazarus, R. S. (2000). The cognition-emotion debate: A bit of history. In M. Lewis and J. Haviland-Jones, editors, Handbook of Cognition and Emotion, pages 1–20. New York: Guilford Press.
  46. 46.Lehrer, A. (1974). Semantic fields and lexical structure. North-Holland, American Elsevier, Amsterdam, NY.
  47. 47.Litman, D. J. and Forbes-Riley, K. (2004). Predicting student emotions in computer-human tutoring dialogues. In Proceedings of the 42nd Annual Meeting on Association for Computational Linguistics, ACL ’04, Morristown, NJ, USA. Association for Computational Linguistics.
  48. 48.Liu, H., Lieberman, H., and Selker, T. (2003). A model of textual affect sensing using real-world knowledge. In Proceedings of the 8th international conference on Intelligent user interfaces, IUI ’03, pages 125–132, New York, NY. ACM.
  49. 49.Louviere, J. J. (1991). Best-worst scaling: A model for the largest difference judgments. Technical report, University of Alberta.
  50. 50.Louviere, J. J. and Finn, A. (1992). Determining the appropriate response to evidence of public concern: The case of food safety. Journal of Public Policy and Marketing, 11(2), 12–25.
  51. 51.Ma, C., Prendinger, H., and Ishizuka, M. (2005). Emotion estimation and reasoning based on affective textual interaction. In J. Tao and R. W. Picard, editors, First International Conference on Affective Computing and Intelligent Interaction (ACII-2005), pages 622–628, Beijing, China.
  52. 52.Masson, J. M. (1996). When Elephants Weep: The Emotional Lives of Animals. Delta.
  53. 53.Matykiewicz, P., Duch, W., and Pestian, J. P. (2009). Clustering semantic spaces of suicide notes and newsgroups articles. In Proceedings of the Workshop on Current Trends in Biomedical Natural Language Processing, BioNLP ’09, pages 179–184, Stroudsburg, PA, USA. Association for Computational Linguistics.
  54. 54.Maute, M. F. and Forrester, W. J. (1993). The structure and determinants of consumer complaint intentions and behavior. Journal of Economic Psychology, 14(2), 219–247.
  55. 55.Mihalcea, R. and Liu, H. (2006). A corpus-based approach to finding happiness. In AAAI-2006 Spring Symposium on Computational Approaches to Analysing Weblogs, pages 139–144. AAAI Press.
  56. 56.Mohammad, S. M. (2011a). Even the abstract have colour: Consensus in wordcolour associations. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Portland, OR, USA.
  57. 57.Mohammad, S. M. (2011b). From once upon a time to happily ever after: Tracking emotions in novels and fairy tales. In Proceedings of the ACL 2011 Workshop on Language Technology for Cultural Heritage, Social Sciences, and Humanities (LaTeCH), Portland, OR, USA.
  58. 58.Mohammad, S. M. and Turney, P. D. (2010). Emotions evoked by common words and phrases: Using mechanical turk to create an emotion lexicon. In Proceedings of the NAACL-HLT 2010 Workshop on Computational Approaches to Analysis and Generation of Emotion in Text, LA, California.
  59. 59.Mohammad, S. M. and Yang, T. W. (2011). Tracking sentiment in mail: how genders differ on emotional axes. In Proceedings of the ACL 2011 Workshop on Computational Approaches to Subjectivity and Sentiment Analysis (WASSA), Portland, OR, USA.
  60. 60.Mohammad, S. M., Dunne, C., and Dorr, B. (2009). Generating high-coverage semantic orientation lexicons from overtly marked words and a thesaurus. In Proceedings of Empirical Methods in Natural Language Processing (EMNLP-2009), pages 599–608, Singapore.
  61. 61.Neviarouskaya, A., Prendinger, H., and Ishizuka, M. (2009). Compositionality principle in recognition of fine-grained emotions from text. In Proceedings of the Proceedings of the Third International Conference on Weblogs and Social Media (ICWSM-09), pages 278–281, San Jose, California.
  62. 62.Neviarouskaya, A., Prendinger, H., and Ishizuka, M. (2010). Affect analysis model: novel rule-based approach to affect sensing from text. Natural Language Engineering, 17(Part 1), 95–135.
  63. 63.Nowlis, V. and Nowlis, H. H. (2001). The description and analysis of mood. Annals of the New York Academy of Sciences, 65(4), 345–355.
  64. 64.Oliver, R. L. (1997). Satisfaction a behavioral perspective on the consumer . New York: McGraw-Hill.
  65. 65.Ortony, A. and Turner, T. J. (1990). What’s basic about basic emotions? Psychological Review, 97, 315–331.
  66. 66.Ortony, A., Clore, G. L., and Collins, A. (1988). The Cognitive Structure of Emotions. Cambridge University Press.
  67. 67.Osgood, C. E. and Walker, E. G. (1959). Motivation and language behavior: A content analysis of suicide notes. Journal of Abnormal and Social Psychology, 59(1), 58–67.
  68. 68.Pang, B. and Lee, L. (2008). Opinion mining and sentiment analysis. Foundations and Trends in Information Retrieval, 2(1–2), 1–135.
  69. 69.Parrot, W. (2001). Emotions in Social Psychology. Psychology Press.
  70. 70.Pearl, L. and Steyvers, M. (2010). Identifying emotions, intentions, and attitudes in text using a game with a purpose. In Proceedings of the NAACL-HLT 2010 Workshop on Computational Approaches to Analysis and Generation of Emotion in Text, Los Angeles, California.
  71. 71.Perreault, W. D. and Leigh, L. E. (1989). Reliability of nominal data based on qualitative judgments. Journal of Marketing Research, 26, 135–148.
  72. 72.Pestian, J. P., Matykiewicz, P., and Grupp-Phelan, J. (2008). Using natural language processing to classify suicide notes. In Proceedings of the Workshop on Current Trends in Biomedical Natural Language Processing, BioNLP ’08, pages 96–97, Stroudsburg, PA, USA. Association for Computational Linguistics.
  73. 73.Plutchik, R. (1962). The Emotions. New York: Random House.
  74. 74.Plutchik, R. (1980). A general psychoevolutionary theory of emotion. Emotion: Theory, research, and experience, 1(3), 3–33.
  75. 75.Plutchik, R. (1985). On emotion: The chicken-and-egg problem revisited. Motivation and Emotion, 9(2), 197–200.
  76. 76.Plutchik, R. (1994). The psychology and biology of emotion. New York: Harper Collins.
  77. 77.Rashid, R., Aitken, J., and Fels, D. (2006). Expressing emotions using animated text captions. In K. Miesenberger, J. Klaus, W. Zagler, and A. Karshmer, editors, Computers Helping People with Special Needs, volume 4061 of Lecture Notes in Computer Science, pages 24–31. Springer Berlin / Heidelberg.
  78. 78.Ravaja, N., Saari, T., Turpeinen, M., Laarni, J., Salminen, M., and Kivikangas, M. (2006). Spatial presence and emotions during video game playing: Does it matter with whom you play? Presence: Teleoperators and Virtual Environments, 15(4), 381–392.
  79. 79.Read, J. (2004). Recognising affect in text using pointwise-mutual information. Ph.D. thesis, Department of Informatics, University of Sussex.
  80. 80.Reichheld, F. F. and Schefter, P. (2000). E-loyalty: your secret weapon on the web. Harvard Business Review, pages 105–113.
  81. 81.Rentoumi, V., Giannakopoulos, G., Karkaletsis, V., and Vouros, G. A. (2009). Sentiment analysis of figurative language using a word sense disambiguation approach. In Proceedings of the International Conference RANLP-2009 , pages 370–375, Borovets, Bulgaria. Association for Computational Linguistics.
  82. 82.Richins, M. (1987). A multivariate analysis of responses to dissatisfaction. Journal of the Academy of Marketing Science, 15, 24–31.
  83. 83.Richins, M. L. (1984). Word of mouth communication as negative information. Advances in Consumer Research, 11, 697–702.
  84. 84.Russell, J. A. (1994). Is there universal recognition of emotion from facial expression? a review of the cross-cultural studies. Psychological Bulletin, 115, 102–141.
  85. 85.Sautera, D. A., Eisner, F., Ekman, P., and Scott, S. K. (2010). Cross-cultural recognition of basic emotions through nonverbal emotional vocalizations. Proceedings of the National Academy of Sciences, 107(6), 2408–2412.
  86. 86.Scherer, K. R. (1982). Emotion as a process: Function, origin and regulation. Social Science Information, 21(4–5), 555–570.
  87. 87.Scherer, K. R. (1984). Emotion as a multicomponent process: a model and some cross-cultural data. Review of Personality and Social Psychology, 5, 37–63.
  88. 88.Scott, W. A. (1955). Reliability of Content Analysis:. Public Opinion Quarterly, 19(3), 321–325.
  89. 89.Shankar, V., Urban, G. L., and Sultan, F. (2002). Online trust: a stakeholder perspective, concepts, implications, and future directions. The Journal of Strategic Information Systems, 11(3–4), 325–344.
  90. 90.Shaver, P., Schwartz, J., Kirson, D., and O’Connor, G. (1987). Emotion knowledge: Further exploration of a prototype approach. Journal of Personality and Social Psychology, 52, 1061–1086.
  91. 91.Singh, J. (1988). Consumer complaint intentions and behavior: Definitional and taxonomical issues. The Journal of Marketing, 52(1), 93–107.
  92. 92.Snow, R., O’Connor, B., Jurafsky, D., and Ng, A. (2008). Cheap and fast - but is it good? Evaluating nonexpert annotations for natural language tasks. In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP-2008), pages 254–263, Waikiki, Hawaii.
  93. 93.Steunebrink, B. R. (2010). The Logical Structure of Emotions. Ph.D. thesis, Dutch Research School for Information and Knowledge Systems.
  94. 94.Stewart, K. J. (2003). Trust transfer on the world wide web. Organization Science, 14, 5–17.
  95. 95.Stone, P., Dunphy, D. C., Smith, M. S., Ogilvie, D. M., and associates (1966). The General Inquirer: A Computer Approach to Content Analysis. The MIT Press.
  96. 96.Strapparava, C. and Valitutti, A. (2004). Wordnet-Affect: An affective extension of WordNet. In Proceedings of the 4th International Conference on Language Resources and Evaluation (LREC-2004), pages 1083–1086, Lisbon, Portugal.
  97. 97.Subasic, P. and Huettner, A. (2001). Affect analysis of text using fuzzy semantic typing. IEEE Transaction on Fuzzy Systems, 9(4), 483–496.
  98. 98.Turney, P. and Littman, M. (2003). Measuring praise and criticism: Inference of semantic orientation from association. ACM Transactions on Information Systems (TOIS), 21(4), 315–346.
  99. 99.Velásquez, J. D. (1997). Modeling emotions and other motivations in synthetic agents. In Proceedings of the fourteenth national conference on artificial intelligence and ninth conference on Innovative applications of artificial intelligence, AAAI’97/IAAI’97, pages 10–15. AAAI Press.
  100. 100.Weiner, B. (1985). An Attributional Theory of Achievement Motivation and Emotion. Psychological Review, 92(4), 548–73.
  101. 101.Wiebe, J. M. (1994). Tracking point of view in narrative. Computational Linguistics, 20(2), 233–287.
  102. 102.Zajonc, R. B. (1984). On the primacy of affect. American Psychologist, 39(2), 117–123.
  103. 103.Zhe, X. and Boucouvalas, A. (2002). Text-to-Emotion Engine for Real Time Internet CommunicationText-to-Emotion Engine for Real Time Internet Communication, pages 164–168.

Citation

MLA
Mohammad, S. M., and P. D. Turney. “CROWDSOURCING A WORD–EMOTION ASSOCIATION LEXICON”. Computational Intelligence, vol. 29, no. 3, 2012, pp. 436–65, https://doi.org/10.1111/j.1467-8640.2012.00460.x.
APA
Mohammad, S. M., & Turney, P. D. (2012). CROWDSOURCING A WORD–EMOTION ASSOCIATION LEXICON. Computational Intelligence, 29(3), 436–465. https://doi.org/10.1111/j.1467-8640.2012.00460.x
Chicago
Mohammad, S. M., and P. D. Turney. 2012. “CROWDSOURCING A WORD–EMOTION ASSOCIATION LEXICON”. Computational Intelligence 29 (3): 436–65. https://doi.org/10.1111/j.1467-8640.2012.00460.x.
Harvard
Mohammad, S.M. and Turney, P.D. (2012) “CROWDSOURCING A WORD–EMOTION ASSOCIATION LEXICON”, Computational Intelligence, 29(3), pp. 436–465. Available at: https://doi.org/10.1111/j.1467-8640.2012.00460.x.
Vancouver
1. Mohammad SM, Turney PD (2012) CROWDSOURCING A WORD–EMOTION ASSOCIATION LEXICON. Computational Intelligence 29:436–465

BibTeX

@article{Mohammad_2012, title={CROWDSOURCING A WORD–EMOTION ASSOCIATION LEXICON}, volume={29}, ISSN={1467-8640}, url={http://dx.doi.org/10.1111/j.1467-8640.2012.00460.x}, DOI={10.1111/j.1467-8640.2012.00460.x}, number={3}, journal={Computational Intelligence}, publisher={Wiley}, author={Mohammad, Saif M. and Turney, Peter D.}, year={2012}, month=Sept, pages={436–465} }
Metadata:Crossref

Access the Paper

This paper is available from its original source. Click below to access the PDF.

Open PDF