This paper presents a crowdsourcing approach to generate a large, high-quality word-emotion association lexicon. The authors, Saif M. Mohammad and Peter D. Turney, from the National Research Council Canada, describe the challenges and solutions for emotion annotation in a crowdsourcing scenario. They propose a method that includes a word choice question to guide annotators to the intended sense of a target word, reducing the risk of incorrect annotations. The lexicon, named EmoLex, contains over 10,000 terms and focuses on eight basic emotions: joy, sadness, anger, fear, trust, disgust, surprise, and anticipation. The authors conducted experiments to determine the effectiveness of different question formulations and found that asking if a term is *associated* with an emotion leads to higher inter-annotator agreement compared to asking if it *evokes* an emotion. The paper also discusses the applications of emotion-aware systems in customer relationship management, sentiment analysis, and other areas. The results show that a significant percentage of terms are emotive, with trust and joy being the most common emotions associated with nouns, verbs, adjectives, and adverbs. The authors conclude that the crowdsourcing approach is effective in generating a comprehensive and high-quality emotion lexicon, despite the challenges of quality control and finding enough annotators.This paper presents a crowdsourcing approach to generate a large, high-quality word-emotion association lexicon. The authors, Saif M. Mohammad and Peter D. Turney, from the National Research Council Canada, describe the challenges and solutions for emotion annotation in a crowdsourcing scenario. They propose a method that includes a word choice question to guide annotators to the intended sense of a target word, reducing the risk of incorrect annotations. The lexicon, named EmoLex, contains over 10,000 terms and focuses on eight basic emotions: joy, sadness, anger, fear, trust, disgust, surprise, and anticipation. The authors conducted experiments to determine the effectiveness of different question formulations and found that asking if a term is *associated* with an emotion leads to higher inter-annotator agreement compared to asking if it *evokes* an emotion. The paper also discusses the applications of emotion-aware systems in customer relationship management, sentiment analysis, and other areas. The results show that a significant percentage of terms are emotive, with trust and joy being the most common emotions associated with nouns, verbs, adjectives, and adverbs. The authors conclude that the crowdsourcing approach is effective in generating a comprehensive and high-quality emotion lexicon, despite the challenges of quality control and finding enough annotators.