Crowdsourcing Word-Color Associations M. Lafourcade 1 , N. Le Brun 2 , V. Zampa 3 1 LIRMM, Montpellier, France 2 Imagin@t, Lunel, France 3 LIDILEM, Grenoble, France Abstract. In Natural Language Processing and semantic analysis in particular, color information may be important in order to properly process textual information (word sense disambiguation, and indexing). More specifically, knowing what colors are generally associated with what terms is crucial information. In this paper, we explore how crowdsourcing through a game with a purpose (GWAP) can be an adequate strategy to collect such lexico-semantic data. Kaywords: Word Color Associations, Lexical Network, Crowdsourcing 1. Introduction Color information is important in our daily life and may be of interest in the context of Natural Language Processing. Although, this is beyond the scope of this paper, they are strong connections between colors and emotions. However, provide information about word-colors associations to a system dedicated to the semantic analysis of texts, in addition to other traditional knowledge (hypernym, parts of, semantic role, etc.) could greatly improve system performance. Association between word and color could be made for abstract nouns related to emotions (like fear, anger, danger, hope, ...) but in a more straightforward way for concrete nouns (like sky, lion, snow, sea, ...) There have been many psychological studies about colors and their association to words or their impact on communication (either verbal or not). Mohammad (2011a) explains that color is a crucial aspect for the successful delivery of information, whether it is in marketing a commercial product (Sable and Akcay, 2010), in web design (Meier, 1988; Pribadi et al., 1990), or in information visualization (Christ, 1975; Card et al.,1999). Color is definitely a strong bearer of sense. But the relation between colors and senses, for many researchers in psychology/sociology depends on a number of factors. Luscher (1969),a psychotherapist known for inventing the eponymous color test, a tool to measure a person's psychophysical state based on his or her color preferences, argues that the choice of one color by a person is related to (or even translate) his emotional status. Child et al. (1968), and Ou et al. (2011) show that people of different ages and genders may have different color preferences. Moreover, some colors may have lexical variation according to languages and culture (for example Hungarian and Turkish have two words for red). There is a very lively debate as to whether the associations between color and direction were independent of age, gender, or nationality. This might be the case, for example, for the relation between red and danger, since the red may be related to blood/fire regardless of others factors. Berlin and Kay (1969) argued that differences could be organized into a coherent hierarchy, and that there is a limited number of universal basic color terms used in various cultures. Berlin and Kay based their analysis on a comparison of color words in 20 languages from around the world, but their findings have been discussed a lot. In the same way, several expressions using colors have the same meaning in different languages especially when they are culturally close, which is hardly a surprise. For example, dark thoughts in English and idées noires are roughly equivalent, as well as to see red and voir rouge. Many studies, mostly in English, have been undertaken to determine relations between colors and words or colors and emotions, etc. Most of those studies of psycholinguistic are undertaken in a classical way, and their raw results are general modest in size and not freely available. Furthermore, as previously mentioned, it is extremely delicate (and probably unwise) to translate directly the result of