The Emotions of Abstract Words: A Distributional Semantic Analysis

Abstract

Recent psycholinguistic and neuroscientific research has emphasized the crucial role of emotions for abstract words, which would be grounded by affective experience, instead of a sensorimotor one. The hypothesis of affective embodiment has been proposed as an alternative to the idea that abstract words are linguistically coded and that linguistic processing plays a key role in their acquisition and processing. In this paper, we use distributional semantic models to explore the complex interplay between linguistic and affective information in the representation of abstract words. Distributional analyses on Italian norming data show that abstract words have more affective content and tend to co-occur with contexts with higher emotive values, according to affective statistical indices estimated in terms of distributional similarity with a restricted number of seed words strongly associated with a set of basic emotions. Therefore, the strong affective content of abstract words might just be an indirect byproduct of co-occurrence statistics. This is consistent with a version of representational pluralism in which concepts that are fully embodied either at the sensorimotor or at the affective level live side-by-side with concepts only indirectly embodied via their linguistic associations with other embodied words.

Exploring What Is Encoded in Distributional Word Vectors: A Neurobiologically Motivated Analysis

“…Kousta et al (2011) demonstrated using behavioral experiments that abstract words were more emotionally valenced than concrete words when other variables such as imageability and context availability were controlled. Lenci et al (2018) lent support to their finding using language statistics. These findings imply that emotional information is easier to extract from abstract words than from concrete words; that is, our result lends further support to these findings.…”

“…However, emotional attributes for abstract words were predicted more accurately than all other attributes, as shown in Section 4.3. These split results for emotional attributes can be explained by the strong association between abstract words and emotional values (Kousta, Vigliocco, Vinson, Andrews, & Del Campo, 2011; Lenci et al, 2018). Kousta et al (2011) demonstrated using behavioral experiments that abstract words were more emotionally valenced than concrete words when other variables such as imageability and context availability were controlled.…”

“…Emotional information has been addressed most extensively (Bestgen & Vincze, 2012; Hollis et al, 2017; Lenci, Lebani, & Passaro, 2018; Passaro et al, 2017; Recchia & Louwerse, 2015; Tillmand & Louwerse, 2018; Westbury, Keith, Briesemeister, Hofmann, & Jacobs, 2015). Bestgen and Vincze (2012) proposed a method for predicting three affective variables, that is, valence, arousal, and dominance, of ANEW norms (Bradley & Lang, 1999).…”

“…Language statistics also explained reaction times required for truthfulness judgment of emotion sentences, but perceptual simulation factors did not. Lenci et al (2018) examined the role of affective information for abstract concepts and found using language statistics that abstract words had more affective content and were more likely to co‐occur with contexts with higher emotive values.…”

Hybrid CNN Classification for Sentiment Analysis under Deep Learning

“…Text data set generally contains texts of various grammatical nature namely, adjective, verb, adverb, abstract nouns etc. Among the texts of various grammatical nature, it is generally felt that only the abstract nouns express the strongest and intense emotions [3]. Moreover, there is only a limited amount of research being done with reference to abstract nouns.…”