The aim of this paper is to illustrate a methodology used to construct a radial network based on family resemblances or on semantic components that allows one to visualize and measure the relationship between a specific set of words through the use of GraphColl (Brezina et al. 2015). We develop two different models using the same methodology. The first model is formed by assembling the different categories of Taste terms in English elicited by native-speaker informants. Bagli devised a list of possible taste descriptors and asked a group of native speakers to organize them in mutually-exclusive categories on the basis of their semantic meaning. The second model is constructed by using a series of responses to online questionnaires that asked informants which of ten Manner components they felt were indicative of a specific Manner of Speaking (MoS) verb. In both cases, the authors uploaded the informant responses in a .txt file to the collocation software GraphColl, and then verified the correlation strength of the judged components by considering Mutual Information. The results illustrated through the software reveal a visually communicative graph that allows us to understand the correlations that are also fully verifiable through various statistical measures. The article is the result of the close collaboration between the two authors. However, for academic purposes, Jodi Sandford is responsible for 1, 2, and 4.2; Marco Bagli 3, 4.1, and 5.
GraphColl: A Methodology for Visualization and Quantification of Semantic Networks
Jodi L. Sandford
Membro del Collaboration Group
;BAGLI, MARCOMembro del Collaboration Group
2018
Abstract
The aim of this paper is to illustrate a methodology used to construct a radial network based on family resemblances or on semantic components that allows one to visualize and measure the relationship between a specific set of words through the use of GraphColl (Brezina et al. 2015). We develop two different models using the same methodology. The first model is formed by assembling the different categories of Taste terms in English elicited by native-speaker informants. Bagli devised a list of possible taste descriptors and asked a group of native speakers to organize them in mutually-exclusive categories on the basis of their semantic meaning. The second model is constructed by using a series of responses to online questionnaires that asked informants which of ten Manner components they felt were indicative of a specific Manner of Speaking (MoS) verb. In both cases, the authors uploaded the informant responses in a .txt file to the collocation software GraphColl, and then verified the correlation strength of the judged components by considering Mutual Information. The results illustrated through the software reveal a visually communicative graph that allows us to understand the correlations that are also fully verifiable through various statistical measures. The article is the result of the close collaboration between the two authors. However, for academic purposes, Jodi Sandford is responsible for 1, 2, and 4.2; Marco Bagli 3, 4.1, and 5.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
