Sentiment analysis of Public Perceptions on Climate Change

This submission has open access

Abstract Summary

This study was conducted with the aim of analysing how the public in Malaysia, perceive reports about climate change in Malaysian newspapers. Data was collected from an online newspaper published in Malaysia (May 2017 until May 2021), with "climate change Malaysia" being used as the search term. A total of 509 news articles were collected, but only 112 news articles from the editorial news section were selected. The112 news were then filtered, resulting in 59 news articles as the data for this study. A corpus-driven sentiment analysis approach was carried out to classify the polarity of the Malaysian public perceptions, the sentiment lexicon, and the public sentiments. The Azure Machine Learning software was employed to analyse the data. 532 negative sentiment words, with 290 strong negative sentiment words were identified. Only 159 positive sentiment words were found, with 61 words having strong positive sentiment words. The results revealed that the public is reasonably insightful of climate change issues, although their sentiments appeared to be negative. Despite its limitation, the present study has contributed significantly to studies on climate change in Malaysia from the linguistics perspective, since studies focusing on linguistics analysis on climate change in Malaysia is significantly lacking.

Submission ID :

AILA1218

Submission Type

Oral Presentation

Select a Symposium

[SYMP59] OPEN CALL - Language & holistic ecology

Argument :

Climate change is a very concerning issue and is now affecting the globe even worse than before. This study was conducted with the aim of analysing how the public, specifically in Malaysia, perceive reports about climate change in Malaysian newspapers. Data was collected from an online newspaper published in Malaysia (May 2017 until May 2021), with "climate change Malaysia" being used as the search term to collect the required data. This specialised corpus that is named the Malaysian Diachronic Climate Change Corpus (MyDCCC) was developed from The Sun Daily. A total 509 news articles were collected, but only 112 news articles from the editorial news section were selected from the 509 news articles. The collected 112 news articles were then filtered, resulting in 59 news articles being used as the data for this study. The final corpus consisted of a total of 6,791-word types, comprising 48,821-word tokens. A corpus-driven sentiment analysis approach was carried out to classify the polarity of the Malaysian public perceptions, identify the sentiment lexicon, and analyse the public sentiments. The sentiment analysis software employed was Azure Machine Learning that generated the polarity result and the sentiment score of each news article. By performing the sentiment analysis, the researchers also managed to find the average score for each sentiment and the total number of polarities count for all news articles.The news articles were separated into two sub-corpora ; positive sentiment sub-corpora and negative sentiment sub-corpora. The wordlist for each sub-corpus were then extracted to match the wordlist with the Multi-Perspective Question Answering (MPQA) Subjectivity Lexicon that is used by Azure Machine Learning. The wordlist generated in AntConc was compared with the sentiment lexicon list in the MPQA Subjectivity Lexicon to obtain the frequency, sentiment, strength of sentiment and part of speech (POS) of each word. Discourse analysis on selected texts was also carried out. The results revealed the occurrence of 532 negative sentiment words, with 290 strong negative sentiment words, which supported the high occurrence of the 90% polarity of negative sentiments identified in this study. Only 159 positive sentiment words were found, with 61 words strong positive sentiment words. The discourse analysis revealed that the public is reasonably insightful of climate change issues, although their sentiments appeared to be negative. This study has its limitations in that the data source was only from one newspaper and data was collected for a period of five years only. Despite its limitation, the present study has contributed significantly to studies on climate change in Malaysia from the linguistics perspective, since studies focusing on linguistics analysis on climate change in Malaysia is significantly lacking.

Bibliography

Nor Fariza Mohd Nor (Associate Professor, PhD) is a senior lecturer at the Center for Research in Language and Linguistics, Faculty of Social Sciences and Humanities, UKM. Her area of expertise are critical discourse analysis and corpus linguistics.