Two analysis methods were applied to the essays written by Toru Takemitsu and Pierre Boulez, two prominent contemporary composers of classical music. The purpose of the network analysis was to identify keywords and their surrounding words within the text. After applying a morphological analysis to the text, a home-brewed network creation software was employed to create a network for the extracted keywords. There are clearly several node clusters and a number of measures of centrality for the network indicate that the multi-centrality of the keyword space. The text corpus for the essays was then parsed in order to carry out a content analysis at the semantic level. The aim of the content analysis was to extract the structures within the concepts employed by the composers in talking about music. The interesting results include the findings that a) their aesthetic vocabulary is strongly associated to an abstract thinking vocabulary, and b) more ordinary emotion words tend to be associated with lower level music entities. These findings seem to substantiate the layered model of affective processes proposed in our previous report. With methodological perspectives we will argue for the following points; a) Network analysis of the words within a text can provide a better basis for text analysis. b) The modeling of affective process and text analysis may be mutually beneficial.