The Effect of 2001 on UN General Assembly Speechs

©2020 Blaire Palmer

The UN General Assembly is a yearly meeting of all 193 members of the United Nations to discuss world problems. According to the United Nations original charter and various resolutions, this body is the main policymaking body of the United Nations, excluding issues of security being debated by the UNSC. The topics vary based on the pressing issue of the year. This dataset is a transcript of the UN General Assembly meetings from 1970 to 2016. The data is broken up into assembly sessions, years, nations, and text. This data was collected from Kaggle . For the purpose of this analysis, I have limited the data from 1991 onward, as historians commonly agree that this was a turning point in world history because of the disestablishment of the Soviet Union. The majority of this data will be focused solely on the transcript of the 2001 General Assembly debates.

Hypothesis:

2001 was a dramatic change in sentiment for the United States and the World in the UN General Assembly.

First, I downloaded all of the packages and activated them into the R library, and uploaded the kaggle dataset of the UNGA transcripts

install.packages("tidyverse") install.packages("tidytext") install.packages("wordcloud2") install.packages("ggplot2") install.packages("ggthemes") library(ggthemes) library(ggplot2) library(wordcloud2) library(tidyverse) library(tidytext)

2001 was a traumatic year for the United States. The events of 9/11 left a mark on the nation, and the world, as one of the first internationally broadcasted global terrorist acts. This is clearly shown by the graph below, of the average sentiment from 1991 to 2016 of the speeches made by the USA. The sentiment was collected through the “afinn” sentiment lexicon, which graphs the sentiments of words on a scale from -5 to 5. I chose the Afinn analysis because it was the most feasible to find the averages from, as “bing” and “nrc” categorize words with verbal definitions, or character vectors, instead of a numerical scale. This graph shows that overall, the sentiment of the United States’ speeches are positive. In fact the sentiment only dips below 0, going into a negative range, 6 times in the entire dataset. These instances are in 1998, 2001, 2002, 2003, 2013, and 2014.

un %>% filter(country == "USA", year >= 1991) %>% arrange(desc(year)) -> usa_un

usa_un %>% unnest_tokens(word, text) %>% anti_join(stop_words) %>% group_by(year, country)-> tokens

tokens %>% inner_join(get_sentiments("afinn")) -> sentiment

sentiment %>% group_by(year, country, session) %>% summarise(average = mean(value))-> sentiment_count

sentiment_count %>% ggplot() + geom_line(mapping = aes(x = year, y=average)) + ylim(-1.0, 1.0) + theme_few()

Average Sentiment Over Time (USA)

It can be seen that 2001 was the most negative sentiment said by the United States. This is because of the events of September 11th, 2001, when a terror attack hit the two towers in Washington, DC, and almost 3,000 citizens were killed. The word cloud below shows that the center of discussion was focused around these terrorist attacks. The table below shows the ten top spoken words. It should also be noted that although these words did not make the top ten, “Taliban”, “Afghanistan”, and “Murder” are the 11th, 12th, and 13th most spoken words, respectively.

un %>% filter(country == "USA" & year == 2001) -> usa_un

usa_un %>% select(text) %>% unnest_tokens(word, text) %>% anti_join(stop_words) %>% count(word, sort = TRUE) %>% head(200) -> usa_cloud

usa_cloud %>% wordcloud2(size = .6, minRotation = -pi/6, maxRotation = -pi/6, rotateRatio = 1, color="random-dark")

USA World Cloud (2001)

Word Frequency
nations 20
terror 16
united 16
terrorists 15
world 14
nation 11
country 10
people 10
terrorist 8
september 7

To expand this analysis of 2001, I also included an “afinn” sentiment analysis of countries in the Middle East. I am using the working definition of the “Middle East” as all of the nations in the Arab League, as well as Israel, Iran, and Turkey. More information on why this definition is used can be found in the introduction of Understanding the Contemporary Middle East. It is clear that most of these nations have a negative sentiment as the main object of their speeches. To see the list of countries and their appreviations, click here

un %>% filter(country == "DZA" | country == "BHR" | country == "COM" | country == "DJI"| country == "EGY" | country == "IRQ" | country == "IRN" | country == "JOR" | country == "KWT" | country == "LBN" | country == "LBY" | country == "MTQ" | country == "MAR" | country == "OMN" | country == "PSE" | country == "QAT" | country == "SAU" | country == "SOM" | country == "SDN" | country == "SYR" | country == "TUN" | country == "ARE" | country == "YEM" | country == "TUR" | country == "ISR" | country == "AFG") %>% filter(year == 2001) -> middle_east

middle_east %>% unnest_tokens(word, text) %>% anti_join(stop_words) %>% group_by(year, country)-> me_clean

me_clean %>% inner_join(get_sentiments("afinn")) -> me_sent

me_sent %>% group_by(year, country, session) %>% summarise(average = mean(value))-> me_sent_count

me_sent_count %>% ggplot() + geom_col(mapping = aes(x =country, y=-average, fill = average)) + coord_flip() + ylim(-1.0, 1.0) + theme_few()

Average Sentiment of Middle Eastern Nations (2001)

Next, I created a word cloud of these nations' most said words to see what the topics discussed were. It is clear from the word cloud that in 2001 the countries in the Middle East are focused on the nation of Israel as the topic of debate. The other word cloud is of the words said by the permanent 5 nations in the security council (excluding the United States), and this cloud shows that 9/11 was a more talked about topic by these nations. To learn more about the p5 nations and why they are important, click here. It should be noted that since there are 25 countries being analyzed together in the 'Middle East' word cloud, there are likely more variation in words said, and a larger dataset to analyze. With a more indepth read through of the text, I noticed that multiple Middle Eastern countries felt the need to publicly separate Islam from radical extremism.

This is the case from the representative for Yemen. He made the statement that "We also affirm the Arab Islamic position that rejects attempts to link terrorism to Arabs and Muslims. Terrorism, as history teaches us, has no religion, no homeland and no identity." Similarly, The representative from the Comoros Islands felt the pressure to condemn terror acts, commenting that "The Islamic Federal Republic of the Comoros therefore strongly condemned those hateful and barbarous acts, just as it condemns terrorism in all its forms." The P5 nations, however, discussed terrorism in ways that did not focus on differing Islamic extremism from Islam (which could be because none of the P5 nations are majority Muslim), but instead with a focus on the acts that the nation has been carrying out to try to solve them. This can be seen below in the highlighted cells of the table, where the P5 nations said 'cooperation' and 'development' 21 times per word. For example, The representative for Russia commented on the talks that the USA and Russia had already begun commencing to brainstorm ways to eliminate global terrorism.

un %>% filter( country == "DZA" | country == "BHR" | country == "COM" | country == "DJI"| country == "EGY" | country == "IRQ" | country == "IRN" | country == "JOR" | country == "KWT" | country == "LBN" | country == "LBY" | country == "MTQ" | country == "MAR" | country == "OMN" | country == "PSE" | country == "QAT" | country == "SAU" | country == "SOM" | country == "SDN" | country == "SYR" | country == "TUN" | country == "ARE" | country == "YEM" | country == "TUR" | country == "ISR") %>% filter(year == 2001) -> MEA_un

MEA_un %>% select(text) %>% unnest_tokens(word, text) %>% anti_join(stop_words) %>% count(word, sort = TRUE) %>% head(500) -> MEA_cloud

MEA_cloud %>% wordcloud2(size = .7, minRotation = -pi/6, maxRotation = -pi/6, rotateRatio = 1, color="random-dark")


#P5 Word Cloud 2001

un %>% filter( country == "RUS" | country == "GBR" | country == "CHN" | country == "FRA") %>% filter(year == 2001) -> pfive_un

pfive_un %>% select(text) %>% unnest_tokens(word, text) %>% anti_join(stop_words) %>% count(word, sort = TRUE) %>% head(500) -> pfive_cloud

pfive_cloud %>% wordcloud2(size = .7, minRotation = -pi/6, maxRotation = -pi/6, rotateRatio = 1, color="random-dark")

Middle East Word Cloud (2001)

Most Used Words

Middle East Permanent 5
Word Frequency Word Frequency
international 414 international 60
terrorism 290 united 50
united 283 world 51
peace 278 nations 50
nations 230 terrorism 39
security 223 security 34
world 218 cooperation 21
people 202 development 21
countries 155 countries 20
palestine 127 nations 20


Permanent 5 Word Cloud (2001)

In the same vein, it is clear that many nations discussed terrorist acts in their speeches during this session of the UNGA. The United States' speech mentioned these words the most, as shown in the piechart below. This piechart shows the frequency of times that the words 'terror', 'terrorists', and 'terrorist' are said with all of the countries in the 2001. I included ten nations that said the words the most, and the USA takes the cake for the frequency of times the words were used. The USA mentioned a variation of these three words 39 times, while the following top nations (Israel, Turkey, Andora, and Croatia), mentioned the words 23, 23, 13, and 13 times respectively. It should be mentioned that global terrorism was the topic of many nations, as the attacks of September 11th were the first global terrorist attacks that the UNGA had to debate on.

un %>% filter(year == 2001) %>% arrange(desc(year)) %>% unnest_tokens(word, text) -> country_words

country_words %>% group_by(year, country) %>% filter(year == 2001 , word == "terror" | word == "terrorist" | word == "terrorists") %>% count(word, sort = TRUE) -> terror

terror %>% group_by(country) %>% summarize(total = sum(n)) %>% arrange(desc(total)) %>% head(10)-> terror_top

ggplot(terror_top, aes(fill=country, y=total, x="")) + geom_bar(stat="identity", width=1) + coord_polar("y", start=0) + theme_void() + theme(legend.position="left")

Nations Talking about Terror

Conclusion

In conclusion, it is clear that the sentiment of the UNGA in 2001 was negative because nations were reacting to the terrorist events of September 11th in the United States. In Middle Eastern Nations, there was more of a sense of going on the defense, with Islamic nations feeling a mandatory sense of having to condemn Islamic extremism. On the other hand, the permanent 5 nations in the UNSC spoke more of solutions and looking forward to eliminating global terrorism. The United States spoke about terror the most, which makes sense because the event was in the nation's borders.