top of page

Social trends can be reflected by text.  

 

Written text serves as a window into people’s attitudes in cultures. 

     Through natural language processing and text mining, we can analyze texts of different eras and understand social trends within these eras. For instance:
 
  1. Ying-jeou Ma (the 12th and 13th President of Taiwan, ROC) and Ing-Wen Tsai (the 14th and 15th President), what are their political emphases? 
  2. Medium in Taiwan are keen to report "Ing-Wen Tsai dare not mention ROC (中華民國)." Is this true?  
      To answer these two questions, I collected these two presidents' speeches on the Double Tenth days (the National Day of Taiwan, ROC) in these 10 years (2011 - 2020) from the website of president (http://www.president.gov.tw). Every year, the president presents their administrative report and political vision toward people around the world on this day. With natural language processing and text mining, we can provide relatively quantitative and objective opinions when discussing social trends.  
中華民國第12、13任總統馬英九先生官方肖像照.jpg
蔡英文官方元首肖像照.png
121348295_3701854613166580_9151525744180
Figure 1. Word clouds from each presidents' speeches on the Double Tenth Day
Source of the presidents' photos: https://en.wikipedia.org/wiki/Ma_Ying-jeou
                                                           https://en.wikipedia.org/wiki/Tsai_Ing-wen
     If we viewed the words from the speeches on the Double Tenth day, unlike the claims from the medium in Taiwan, we can ensure that President Tsai dares mention 中華民國 "ROC," although the frequency is less than President Ma. Interestingly, these two presidents' preference for writing Taiwan are different (臺灣 vs. 台灣). President Ma frequently mentioned 臺灣 "Taiwan," 中華民國 "ROC," 經濟 "Economics," 國家 "country," 民主 "democracy," 大陸 "mainland (China)," and 兩岸 "Cross-strait (relationship)." President Tsai emphasized on 台灣 "Taiwan," 國家 "country," 發展 "development," 國際 "democracy," 經濟 "economy," 區域 "region," and 社會 "society."
Figure 2. Discriminative words from each presidents' speeches on the Double Tenth Day
     However, if we simply view the word cloud, the two presidents' specific policies are still unclear. I extracted the discriminative words from each presidents with ratio differences. In figure 2, it seems that President Ma has different policies from President Tsai. President Ma's discriminative words (the left side) included 經濟 "economy," 兩岸 "cross-straight (relationship)", 合作 "cooperation," 日本 "Japan," 美國 "America, " and 大陸 "Mainland China." President Tsai's discriminative words (the right side) included 國家 "country," 發展 "development", 產業 "industry," 區域 "region," 國防 "national defense  " and 團結 "union." President Ma focused on "diplomacy and economy," and President Tsai focused on "development of country, industry, and regional defense." 
Figure 3. Pair correlation between the speeches on the Double Tenth Day
     Furthermore, I performed the pair correlation between the speeches on the Double Tenth Day. In figure 3, two presidents' speeches can be roughly classified: the speeches were correlated with each other within each president. In other words, these two presidents have different political visions. Interestingly, President Tsai's speech in her first year (2016) is much more correlated with President Ma. This year may act as a connection between these two presidents' political visions. 

     As we can see from the above analyses, we can easily understand social trend with the relatively objective and quantitative methods. This method can be applied to discuss (1) the dominant syntactic construction with time (Hsieh, 2019a), (2) the preferred topics by newspapers with time (Hsieh, 2019b), and (3) product review (in preparation). 
Reference: 

1. Hsieh, M. (2019). Newspapers speak “two” languages: Evidence from the use of fānzhuǎn “flip” in
          Taiwan. IEICE Technical Report, 119(151), 55-59.
2. Hsieh, M. (2019). Intertwining complexities between social and dynamic changes of language use:
          An example of fānzhuǎn “flip” in Chinese. In Yoshimoto (eds), Proceeding of the 21th Annual
       Conference of the Japanese Society of Language Sciences (pp. 95-98). Sendai, Japan: JSLS Main
          Office.
bottom of page