國中生情緒主題寫作文本之情緒詞彙特徵與心理健康相關研究

No Thumbnail Available

Date

2018

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

本研究旨在瞭解國中生寫作文本所用的情緒詞彙與其心理健康情形之關聯,根據情緒理論以及過去情緒詞相關文獻資料,分成兩個研究進行探討。研究一蒐集北部266位國中學生共760篇情緒主題寫作文本,並使用「中文情緒網絡詞典」之情緒詞彙特徵指標:情緒共現性及情緒正負向性作為計算指標,將寫作文本以句法為單位後進行剖析,再對照詞典中各詞彙的情緒共現性與正負向性,計算文本的情緒共現性與正負向性,以進行寫作文本情緒類別及正負向性之判斷。結果顯示,以情緒特徵指標的計算結果區辨文本情緒的正負向性有超過八成的正確率,在預測文本情緒的類別時同樣也有超過八成的正確率。同時,以情緒特徵指標評判文本正負向性的結果與人工判斷的結果也有相當高的一致性。研究二透過研究一之266位學生進行心理健康相關量表之調查,並請學生另撰寫581篇隱含情緒寫作文本,依據研究一之計算程序計算文本的各項情緒特徵指標,並用迴歸分析檢驗學生的文本情緒特徵指標與其心理健康相關量表的關聯。結果顯示,以「自尊」而言,厭惡隱喻共現性的預測效果達負向顯著。以「身體症狀」而言,生氣共現性的預測效果達正向顯著。以「睡眠狀況」而言,悲傷共現性的預測效果達正向顯著,害怕共現性的預測效果達負向顯著。結果顯示,以情緒詞彙特徵指標能有效區辨文本情緒的正負向性,也能有效預測文本情緒的類別;而在以情緒詞彙特徵指標預測心理健康時也同樣能獲得顯著的預測效果。顯示透過寫作文本的內容分析能夠偵測學生的情緒,探討學生的心理健康程度,以利進行後續的心理輔導。
This thesis studies the relationship between mental health and the emotional words from the articles of junior high school students through two comprehensive research. The first research is a classification system labeling the emotions of article through co-occurrence analysis from “The Emotional Lexical Co-occurrence Corpus”. The dataset includes 760 articles from 266 northern Taiwan junior high school students. The input predictors of the classification include the weight vector of words, the emotional co-occurrence between words, and emotional characteristics from Chinese emotional online dictionary. This classification results demonstrate the effectiveness with 80% precision which is highly identical to supervised labeling. The second research studies the relationship between mental health and emotional characteristics from the articles through regression analysis. Based on the classification system developed from the first research, the second research analyzes another 581 articles from the same group. The results show that the number and frequency of emotional words are significant predictors of the level of negative emotion, physical symptoms, and anxiety levels. Furthermore, the co-occurrence of different aspects also shows significant prediction accuracy on self-respect, physical symptoms, and sleep quality. This thesis concludes written articles are a good resource to analyze emotion levels and mental symptoms, and emotional characteristics is a significant predictor on labeling the emotion of articles and predicting mental symptoms.

Description

Keywords

中文情緒網絡詞典, 心理健康, 情緒特徵指標, Chinese emotional online dictionary, Emotional characteristics, Mental health

Citation

Collections