Estudo exploratório de categorias gramaticais com potencial de indicadores para a análise de sentimentos

Abstract

This paper reports on work in progress on grammatical categories that may be explored as indicators of emotionin written text. Unlike state-of-the-art research on sentiment analysis focusedon lexical items, this study draws on systemic-functional grammar [Halliday & Matthiessen 2014]in order to map patterns ofchoice ingrammatical systems that can be linked to emotionconstrualin language. The methodology is based on manual annotation of news report text samples carried out on a spreadsheet with categories pertaining to the main grammatical systems at clauselevel inthe rank scale. Frequency of annotated individual categories and category clusters isexamined with a view toidentifying the most productive categories to probe sentiment in text