Abstract
Political speech acts are critical for politicians launching a regime because they can provide information that can be used to control people's thoughts and opinions. The purpose of this study was to conduct a qualitative content analysis of the inaugural and ascension addresses of Nigerian heads of state and presidents. The textual data used in this analysis were the ascension and inaugural addresses of Nigerian Heads of State and Presidents from 1960 to 2019. They were extracted and analysed using text-mining techniques. Textual data were clustered about their topical content using Latent Dirichlet Allocation (LDA), and speech cohesion between these addresses was examined using a similarity matrix and heatmap. Furthermore, term frequency and association analyses were performed to examine the high-frequency terms (tokens) and the terms (tokens) that are strongly correlated within each of the ascension/inaugural addresses (corpus). The summarization of characters and words in the ascension and inaugural addresses reveals that the Civilian Presidents used more characters and words than the Military Heads of State. There was an increase in the number of characters and words in the ascension and inaugural addresses among those who had served the nation multiple times. The total sentiment score in the ascension/inaugural addresses from 1960 to 2019 by Civilian Presidents and Military Heads of State revealed that the Civilian Presidents expressed more trust, surprise, sadness, joy, fear, disgust and anticipation in their addresses than the Military Heads of State. The most occurring term (token) in the ascension/inaugural addresses was the word government which appeared 221 times. The most token in the corpus government was found to be moderately correlated with the following tokens: loss, existing and majority. Similarly, economic was found to be moderately correlated with these tokens: inflation, building, education, exchange, loan, workers and technical. In this study, all the ascension/inaugural addresses share similar topic distribution: as seen in Abacha’s and Muritala’s addresses; and Shonekan’s inaugural address was very similar to Balewa, Azikwe and Babangida's addresses; Babangida's ascension, Abdulsalam’s 1998 ascension, Jonathan’s 2010 inaugural and Buhari’s 2015 inaugural addresses discussed similar topics to Obasanjo’s 1976 ascension address. The highest average sentiment score was observed in Obasanjo’s 2003 inaugural address and the lowest score was in Buhari’s 1983 ascension address. The sentiment score for the ascension/inaugural addresses showed that Civilian Presidents inaugural addresses expressed more positive, joy, trust and anticipation than Military Heads of State. These emotions showed that the Civilian President’s inaugural addresses are better when compared to Military Heads of State in terms of the sentiment scores.
Keywords: Data mining, Sentiment analysis, Head of State, Nigeria, Content analysis
Introduction
Language has played a pivotal role in gaining maximum control of many people’s thoughts, actions, opinions and values. People that crave power, most especially, political office holders have seen proper usage of the language as a medium of expressing their manifestos to the citizens, starting from the campaign to gain their vote of confidence, till the inauguration. The inaugural address is an avenue for giving the citizens' policy outlines penned down in the master plan for the new administration.
Nigeria has had fifteen different heads of state and presidents since gaining her independence in 1960. The country is currently in its fourth republic: the first was from 1963 to 1966, the second from 1979 to 1983, the third from 1992 to 1993, and the fourth, and longest, from 1999 to the present. Every republic marks the beginning of the democratic dispensation era. It is, however, customary for every government (civilian or military) as well as the President or Head of State to deliver a political speech. The political speech ceremony (inauguration) into newly elected executive positions like president and governor, usually, in Nigeria, takes place on the 29th May of the election year since the fourth republic (return to democratic government after military disruptions), except for a few states like Kogi, Bayelsa, Ondo, Anambra and Edo due to the pronouncement by the court.
Political speech is paramount for politicians initiating a regime as it provides information to regulate people's thoughts and opinions through political language. Permana and Mauriyat (2021a, b) and Charteris-Black (2018) classified political speech into policy making such as political decisions and putting in place shared values like consensus building. In the work of Osisanwo (2017) found in Ellah (2022), an inaugural speech is a speech delivered by every individual occupying a new political office during their ceremonial induction. Ellah and Nta (2020) stated in their study that political occupants not only express their qualities and traits during the inaugural speeches, but also declare their intentions and make promises to citizens.
Several works have been studied in inaugural speeches in Nigeria. Adeyanju (2006) scrutinized the pragmatic features in the political speeches of six prominent Nigerian leaders: civilian leaders (Alhaji Abubakar Tafawa Balewa, Chief Obafemi Awolowo and Dr. Nnamdi Azikiwe); and military leaders (General Yakubu Gowon, General Olusegun Obasanjo and General Ibrahim Babangida). Akinwotu (2013) examined the political party nomination acceptance speeches of two Nigerian presidential candidates: Chief Moshood Kashimawo Olawale (MKO) Abiola of the Social Democratic Party (SDP) in the 1993 general election and Chief Obafemi Awolowo of the Unity Party of Nigeria (UPN) in the1979 election. Ellah (2022) observed that President Mohammadu Buhari’s 2015 inaugural speech from a discourse-pragmatic perspective, with a special interest in the incorporation and inculcation of other texts in the speech. Adegbija (1995) studied the discourse tactics in military coup speeches in Nigeria while seeking public approval. A comparative pragmatic investigation of the second inaugural speeches of Nigeria’s President Olusegun Obasanjo and United States’ President George Bush was also investigated, the work reveals that those having the same use of language and persuasive words in their speeches (Adetunji 2009).
Sentiment analysis has been used in wide areas of study like e-commerce, communication, use of opinion mining to predict election outcomes and most recently coronavirus pandemic. Pak and Paroubek (2010) employed TreeTagger for the collected corpus of Twitter data to train a sentiment classifier based on the multinomial Naïve Bayes classifier. A sentiment classifier can identify positive, negative and neutral sentiments in documents. Yang et al. (2020) proposed a novel sentiment analysis model named Sentiment Lexicon on Chinese Based and Deep Learning (SLCABG) based on the sentiment lexicon and deep learning techniques to improve the sentiment analysis on product reviews. It was concluded that the new model can be used to assist merchants on e-commerce platforms to get feedback from users. Sentiment analysis has lately been used during the coronavirus pandemic. Sentiment analysis was applied to see how Indians felt towards the imposition of lockdown. It was revealed that most Indians' responses seem positive and follow the imposition of lockdown by the government to flatten the curve (Barkur et al. 2020).
Onyenwe et al. (2020) used Natural Language Processing (NLP) technique referred to as sentiment analysis and tweets text exploratory to empirically examine the impact of political party control on winning an election over its candidates and vice versa. This was like the study of Oyebode and Orji (2019) that used lexicon-based and supervised machine learning approaches to election-related posts on social media to identify negative or positive sentiment polarity. According to Onyenwe et al. (2020), it was concluded that factors that play significant roles in winning an election were the influences of the political party and the candidate's disposition.
The act of dissecting texts of speeches (especially political figures such as an executive head of state) has recently become a must-have for every household. There are two main techniques used, these are the assignment of predefined categories to texts (that is, text classification model); and the act of identifying specific information from the text (i.e. text extractor). Studies that used sentiment analysis in inaugural speeches can be found in Aremu (2017); Balogun and Amodu (2018); Enyi (2016b); Nnamdi-Eruchalu (2017); Ogungbe (2021); Osisanwo 2017). Ogungbe (2021) examined the lexico-syntactic expressiveness in President Muhammadu Buhari's 1983 and 2015 inaugural speeches and found that President Muhammadu Buhari employed stylistic devices like reference, harmony of words (collocation), enumeration, pronouns etc. to win the attention, support, trust and loyalty of Nigerians to the ideas expressed. The pragmatic and goals-oriented acts were identified in the use of words in President Muhammadu Buhari's 2015 inaugural speech in the investigation of Osisanwo 2017) using Mey (2001) pragmatic acts theory for descriptive analysis.
Balogun and Amodu (2018) analysed the patterned repetitions that draw the attention of the audience in the inaugural speeches of President Goodluck Jonathan in 2011 and President Barack Obama in 2013. Being that President Obama is from an English-speaking nation and his speech was prepared by a skillful orator made his speech has an edge over President Jonathan in terms of being captivating and compelling. A word embedding (a subset of supervised machine learning technique) that can detect similar words and levels of negativity in Austrian parliamentary speeches from 1996 to 2013 was used. It was found that there is a greater usage of negative words in the opposition parties than in ruling parties (Rudkowsky et al. 2018).
Text mining and sentiment analysis have been used in various facets of life but are limitedly explored in the inaugural speeches, especially in Nigeria. In the work of Han and Lim (2021), inaugural speeches delivered by 59 US Presidents from the two major political parties were classified using named entities and key phrases by using the Support Vector Machine (SVM). Han and Lim (2021) and Lim and Han (2020) used various N-grams lexical features rather than using topics or semantics to build a feature set.
In this work, the ascension and inaugural addresses of Nigerian Heads of State and Presidents from 1960 to 2019 were analysed using text-mining techniques because only a few (if at all) works have been carried out using text mining and sentiment analysis on Nigeria’s presidents’ inaugural speeches. Governance is a very big challenge in the world. Nigeria is currently faced with a large distrust between the citizens and the leaders (Botha and Abdile 2019). The importance of this research work is to know the sentiment scores of Nigerian leaders and compare Civilian Presidents and Heads-of-State inaugural speeches using sentiment analysis. The methods in this study can also be applied to any country in the world with a military transition to democracy or vice versa.
Methods
The textual data used in this analysis are the ascension and inaugural addresses of Nigerian Heads of State and Presidents from 1960 to 2019, which were accessible online at www.dawodu.com. It is the first website about Nigeria's socio-economic problems, politics and history. Dr. Segun Dawodu started it as a blog for the first time in 1996. The data were saved into a CSV file format given that preprocessing of data is an integral part of any analysis task.
The inherent structure of the ascension and inaugural addresses was analysed using text-mining techniques, which provide an efficient and reliable way of quantifying textual data. Textual data were clustered about their topical content using Latent Dirichlet Allocation (LDA), and speech cohesion between these addresses was examined using a similarity matrix and heatmap. R statistical software version 4.1.0 (2021-05-18)—"Camp Pontanezen" was used for the analysis.
Results
Distribution of characters & words in the ascension/inaugural addresses
We performed some simple calculations and tidied the textual data by counting the number of characters (Speech Length) and words (Speech Words) in each of the ascension and inaugural addresses, with the longest address having 28,267 characters and 4,725 words.
The summarization of characters and words in the ascension and inaugural addresses reveals that the Civilian Presidents used more characters and words than the Military Heads of State (Table 1).
Table 1.
Distribution of characters & words in the speech of the ascension/inaugural addresses
| Head | N | Sum of characters | Sum of words | Mean characters used | Mean words used |
|---|---|---|---|---|---|
| Civilian | 10 | 146,649 | 24,137 | 14,665 | 2,414 |
| Military | 8 | 92,104 | 14,562 | 11,513 | 1,820 |
Specifically, the highest and the lowest characters for civilian presidents were found in Azikwe’s 1963 address which consists of 28,267 characters and in Shonekan’s 1993 address, which consists of 3909, respectively (Table 2). Also, there is an increase in the number of characters and words in the ascension and inaugural addresses among those who have served the nation multiple times. Obasanjo has the highest mean number of characters and words (Table 3).
Table 2.
Distribution of sum of characters & words in the ascension/inaugural addresses by name (alphabetically) and year
| Name | Year | Time spent in the office | Sum of characters | Sum of words |
|---|---|---|---|---|
| Abacha | 1993 | 4yrs, 203 days | 16860 | 2685 |
| Abdulsalam | 1998 | 355 days | 24396 | 3768 |
| Azikiwe | 1963 | 2yrs, 107 days | 28267 | 4725 |
| Babangida | 1985 | 7yrs, 364 days | 11823 | 1883 |
| Balewa | 1960 | 8yrs, 138 days | 6863 | 1147 |
| Buhari | 1983 | 1 yr, 239 days | 8501 | 1330 |
| Buhari | 2015 | 4yrs | 11617 | 1915 |
| Buhari | 2019 | 4yrs* | 19014 | 3021 |
| Gowan | 1966 | 8yrs, 362 days | 13353 | 2110 |
| Ironsi | 1966 | 194 days | 4915 | 772 |
| Jonathan | 2010 | 5yrs, 24 days | 13202 | 2213 |
| Muritala | 1975 | 199 days | 7062 | 1135 |
| Obasanjo | 1976 | 3yrs, 230 days | 5194 | 879 |
| Obasanjo | 1999 | 4yrs | 17466 | 2837 |
| Obasanjo | 2003 | 4yrs | 25098 | 4160 |
| Shagari | 1979 | 4yrs, 91 days | 12888 | 2056 |
| Shonekan | 1993 | 83 days | 3909 | 669 |
| Yaradua | 2007 | 2yrs, 341 days | 8325 | 1394 |
The * indicates the incumbent president
Table 3.
Distribution of mean characters & words in the ascension/inaugural addresses
| Name | Number of time served | Mean characters used | Mean words used |
|---|---|---|---|
| Buhari | 3 | 13,044 | 2089 |
| Obasanjo | 3 | 15,919 | 2625 |
Corpus preparation and document-term matrix creation
To prepare the corpus used for this analysis, we converted all text in the corpus to lower case, removed punctuation, numbers, stop words and stemmed words. The result is that each ascension and inaugural address is a string of tokens, where a token is a sequence of characters that are grouped as a useful semantic unit that is sometimes not always immediately recognizable as words. A document-term matrix was then created where each column represents a token and each row represents ascension and inaugural addresses.
We created a document-term matrix, keeping only tokens longer than three characters since shorter tokens are very hard to interpret. The document’s term matrix has 89% sparsity (zero cells) with 4,921 terms (tokens) (Figs. 1, 2).
Fig. 1.
Distribution of the number of characters in the ascension/inaugural addresses
Fig. 2.
Distribution of the number of Words in the Ascension/Inaugural Addresses by Head
Similarly, terms that appear in the collection of ascension and inaugural addresses less often than the specified lower bound were ignored, and terms that appear only once in the whole corpus were also ignored when building the second document-term matrix. This resulted in 1,889 terms (tokens) with a sparsity of 73%.
The number of times each unique term (tokens) appears within each cell in the first document-term matrix was used to create a word cloud visualization to see the words occurring most frequently (Fig. 3), and the second-word cloud was created with the second document-term matrix (Fig. 4). Figure 4 becomes clearer than Fig. 3 after removing terms appearing in the collection of ascension and inaugural addresses less often than the specified lower bound.
Fig. 3.
Word cloud from the first document-term matrix
Fig. 4.
Word cloud from the second document-term matrix
Sentiment analysis of the ascension/inaugural addresses
The computational study of the sentiments and emotions expressed in the ascension and inaugural addresses was examined. There is a package in the R programming language called “syuzhet”. It is designed to classify the sentiments into positive, negative, fear, joy etc., by extracting the sentiments from the text. These sentiment words are available in the sentiment lexicon. The sentiment extraction method was developed by Standford for Natural Processing Language (Jockers 2020). The process searches each address for the appearance of certain words that are scored individually, which produces values that are marked as either exhibiting a positive or negative sentiment.
The total Sentiment Score in the ascension/inaugural addresses from 1960 to 2019 by Civilian & Military Heads of State/Presidents shown in Fig. 5 reveals that the Civilian Presidents express more Trust, Surprise, Sadness, Joy, Fear, Disgust and Anticipation in their addresses than the Military Heads of State (Fig. 5). Obasanjo’s 2003 inaugural address has the highest positive and trust emotions, followed by Azikwe’s inaugural address in 1963. Surprisingly, Abacha’s ascension address in 1993 has no disgust emotion, while Obasanjo’s address in 1999 had the highest disgust emotion when compared with other Heads-of-State/Presidents' addresses (Fig. 6).
Fig. 5.
Total sentiment score in the ascension/inaugural addresses from 1960 to 2019 by civilian & military heads of state/presidents
Fig. 6.
Total Sentiment score in the ascension/inaugural addresses from 1960 to 2019 by name of heads of state/presidents and year
The highest average sentiment score was observed in Obasanjo’s 2003 inaugural address, and the lowest score was in Buhari’s 1983 ascension address (Fig. 7).
Fig. 7.
Average sentiment Score by year of address and name of heads of state/presidents
Exploring the document-term matrix
Term frequency and association analyses were performed to examine the high-frequency terms (tokens) and the terms (tokens) that are strongly correlated within each of the ascension/inaugural addresses (corpus). The most occurring term (token) in the ascension/inaugural addresses was the government, which occurred 221 times. Other terms are presented in Table 4. The strength of the association (correlation) of the tokens with the twelve (12) most frequently occurring tokens in the ascension/inaugural address is presented in Table 5. Most token in the corpus government was found to be moderately correlated with the following tokens: loss, existing and majority. Similarly, economic was found to be strongly correlated with these tokens: inflation, building, education, exchange, loan, workers and technical ( See Table 5).
Table 4.
High-frequency tokens in the ascension/inaugural addresses (corpus)
| Frequency | Terms (Tokens) |
|---|---|
| 221 | Government |
| 150 | Country, government, Nigeria |
| 100 | Country, government, Nigeria, Nigerians |
| 80 | Country, government, nation, national, Nigeria, Nigerians, people, political |
| 60 | Country, government, nation, national, Nigeria, Nigerians, people, administration, federal, military, political, economic |
| 50 | Country, government, nation, national, Nigeria, Nigerians, people, public, world, administration, continue, federal, fellow, military, political, economic |
Table 5.
Token associations in the ascension/inaugural addresses (corpus) occurring 60 times (limited to the 10 tokens with the strongest associations with correlations > = 0.65)
| Token | Associated Tokens (Strength of Correlation) |
|---|---|
| Country | peaceful (.85), stability (.75), ongoing (.74), survival (.72), fellow (.71), culminating (.69), economic (.69), permanent (.69), chairman (.69), future (.68) |
| Government | loss (.69), existing (.68), majority (.67) |
| Nation | election (.81), foundation (.80), institutions (.77), industry (.76), individuals (.74), democracy (.71), firmly (.71), opportunity (.70), partners (.68), structure (.68) |
| National | institutions (.86), individuals (.81), peaceful (.80), true (.80), require (.79), provisional (.79), reconciliation (.79), ruling (.79), concern (.77), industry (.77) |
| Nigeria | assumption (.84), character (.84), coexistence (.84), criteria (.84), electorate (.84), ethics (.84), exercising (.84), fabric (.84), fourth (.84), hurt (.84) |
| Nigerians | healthcare (.86), inputs (.85), quantities (.85), subregion (.85), water (.85), output (.84), investment (.82), oil (.80), manufacturing (.80), fellow (.79) |
| People | understanding (.85), resolve (.78), required (.76), charge (.74), land (.74), administrations (.70), apply (.70), exploration (.70), real (.69), ensure (.68) |
| Administration | microeconomic (.99), funds (.98), abacha (.96), absence (.96), acceptability (.96), accommodation (.96), accomplish (.96), acknowledges (.96), acutely (.96), adjudication (.96) |
| Federal | corporation (.78), promulgated (.78), subject (.78), country (.71), council (.70), illegal (.67) |
| Military | corporation (.72), promulgated (.72), subject (.72), genuine (.69), region (.69), carry (.68), illegal (.68), provisions (.68), seize (.68), normal (.67) |
| Political | love (.87), structure (.81), constitutional (.80), committee (.80), institutions (.78), true (.78), time (.77), concern (.77), association (.76), judiciary (.76) |
| Economic | inflation (.81), building (.76), education (.68), exchange (.67), loan (.66), workers (.66), technical (.65) |
Topic modelling of tokens in the ascension/inaugural addresses (corpus)
We perform topic modelling which is mixed-method modelling to uncover hidden thematic structures in the ascension/inaugural addresses (corpus). Before we dive into generating the topics and analysing the output, we ran a function to loop over different topic numbers to decide on the number of topics to use in the Latent Dirichlet Allocation (LDA) model. The model with the highest log-likelihood value between 2 and 50 topics indicates the optimum number of the topic that is the best fit for the corpus, in this loop 25 topics were chosen (Fig. 8). Table 6 shows the top-ranked 10 tokens associated with each of the 25 topics generated by the LDA. The topic association plot is a visualization showing the distribution of the 25 topics in each of the ascension/inaugural addresses (corpus) (Fig. 9).
Fig. 8.
LDA model selection results show the log likelihood of the corpus for different numbers of topics
Table 6.
Top-ranked 10 tokens associated with each of the 25 topics generated by the LDA model
| Topic 1 | Topic 2 | Topic 3 | Topic 4 | Topic 5 | Topic 6 | Topic 7 | Topic 8 | Topic 9 | Topic 10 | Topic 11 | Topic 12 | Topic 13 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Government | President | Nigeria | Nigerians | Government | Government | Free | Government | Administration | Law | Political | Government | Government |
| Joint | Country | Independence | Country | Federal | People | Determined | Military | Government | Government | National | Economic | Military |
| People | Citizens | Day | Nigeria | Military | Nigerians | Nation | Federation | Political | People | Head | Country | Ensure |
| Governance | Build | Nation | Government | Nigeria | Boko | Support | Nigeria | Nigerians | Plans | Chief | Nigerians | Democracy |
| Federal | Set | Queen | People | Resources | Haram | Food | Federal | National | Progress | Dissolved | International | People |
| Electoral | Nigerian | Representatives | Fellow | Nation | Country | Economic | Region | Parties | Fellow | Nigeria | Nation | Measures |
| Country | Nation | World | Challenges | Nigerians | Nigeria | Policy | Army | Public | Rebuilding | Country | Nigeria | Past |
| Chief | Justice | Constitutional | Rural | Political | Nigerian | Government | Nigerian | Elections | Poverty | Nation | Military | Public |
| Power | Fellow | Country | Economic | Financial | Forces | World | Regional | Country | Past | Forces | Nigerian | Corruption |
| Nigerians | Property | Friends | Education | Constitution | National | Housing | Continue | Nation | Set | Commanderinchief | History | National |
| Topic 14 | Topic 15 | Topic 16 | Topic 17 | Topic 18 | Topic 19 | Topic 20 | Topic 21 | Topic 22 | Topic 23 | Topic 24 | Topic 25 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Government | Country | Nigeria | Democracy | Nigeria | Nigeria | Government | Government | Chief | Government | Africa | Public | |
| Country | Government | Country | Social | Government | Nigerians | Military | Country | Col | Country | Commonwealth | Nigeria | |
| Parties | National | Continue | Sational | Africa | Country | Coup | Region | Staff | Education | African | Nigerians | |
| Federal | Interim | Nation | Nigeria | Power | Leadership | Public | Military | Government | National | Political | Corruption | |
| Courage | Economic | Nigerian | Process | Freedom | Nation | Yakubu | Eastern | Nation | Nigeria | World | Government | |
| Minister | Nigeria | People | Africa | Office | Elections | British | Federal | Federal | Republic | Office | Administration | |
| Police | Past | Fight | Nigerian | World | Democratic | Country | Comprising | Forces | Task | Human | Service | |
| Nigeria | Sector | Fellow | Policies | Person | Note | Nigerian | Nigeria | Armed | Housing | United | People | |
| Dissolved | Nation | Women | Continue | National | Vision | Nigerians | Nigerian | Council | God | Colonial | Country | |
| Chance | Step | Build | Progress | Nations | Political | Nigeria | Revenue | Military | Nigerian | Status | Ensure |
Fig. 9.
Topic association plot in the ascension/inaugural corpus
Exploring cohesion in the ascension/inaugural addresses (corpus) using cosine similarity
A cosine similarity analysis was performed to examine how each of the ascension/inaugural addresses is related to each other. In this current analysis, two addresses are similar if they share a similar topic distribution with a large cosine similarity measure between them. Although all the ascension/inaugural addresses are somewhat similar due to the large cosine similarity measure, Abacha’s address is very similar to Muritala’s address (0.91005), Shonekan’s inaugural address was very similar to Balewa’s (0.82407), Azikwe’s (0.85611), and Babangida’s (0.85140) addresses. Also, Babangida's ascension address (0.81303), Abdulsalam’s 1998 ascension address (0.83005), Jonathan’s 2010 inaugural address (0.83656) and Buhari’s 2015 inaugural address (0.81213) are very similar to Obasanjo’s 1976 ascension address (Table 7).
Table 7.
Cosine similarity matrix of the ascension/inaugural addresses (corpus)
| S/N | Names | Balewa | Azikwe | Gowon | Ironsi | Muritala | Obasanjo | Shagari | Buhari | Babangida | Abacha | Shonekan | Abdulsalam | Obasanjo | Obasanjo | Yaradua | Jonathan | Buhari | Buhari | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Balewa | 1 | ||||||||||||||||||||||||||||||||||
| 2 | Azikwe | 0.77086 | 1 | |||||||||||||||||||||||||||||||||
| 3 | Gowon | 0.7209 | 0.65882 | 1 | ||||||||||||||||||||||||||||||||
| 4 | Ironsi | 0.72109 | 0.79367 | 0.80652 | 1 | |||||||||||||||||||||||||||||||
| 5 | Muritala | 0.81701 | 0.76941 | 0.78317 | 0.72597 | 1 | ||||||||||||||||||||||||||||||
| 6 | Obasanjo | 0.72796 | 0.71113 | 0.78317 | 0.78697 | 0.73357 | 1 | |||||||||||||||||||||||||||||
| 7 | Shagari | 0.75403 | 0.79584 | 0.74317 | 0.73068 | 0.86751 | 0.71403 | 1 | ||||||||||||||||||||||||||||
| 8 | Buhari | 0.75873 | 0.76163 | 0.74081 | 0.80036 | 0.78968 | 0.75439 | 0.71258 | 1 | |||||||||||||||||||||||||||
| 9 | Babangida | 0.77575 | 0.77629 | 0.78896 | 0.76706 | 0.69774 | 0.81303 | 0.78932 | 0.67964 | 1 | ||||||||||||||||||||||||||
| 10 | Abacha | 0.77339 | 0.74462 | 0.81466 | 0.76923 | 0.91005 | 0.71294 | 0.82751 | 0.80416 | 0.699 | 1 | |||||||||||||||||||||||||
| 11 | Shonekan | 0.82407 | 0.85611 | 0.69158 | 0.75548 | 0.73195 | 0.73937 | 0.78081 | 0.7486 | 0.8514 | 0.73937 | 1 | ||||||||||||||||||||||||
| 12 | Abdulsalam | 0.77701 | 0.66136 | 0.7524 | 0.70353 | 0.74118 | 0.83005 | 0.71348 | 0.75819 | 0.77882 | 0.68181 | 0.70896 | 1 | |||||||||||||||||||||||
| 13 | Obasanjo | 0.77068 | 0.66027 | 0.7571 | 0.65484 | 0.78552 | 0.7933 | 0.7886 | 0.72163 | 0.70733 | 0.78534 | 0.71095 | 0.85032 | 1 | ||||||||||||||||||||||
| 14 | Obasanjo | 0.71113 | 0.82624 | 0.79529 | 0.7314 | 0.78027 | 0.75113 | 0.87511 | 0.68833 | 0.8333 | 0.8114 | 0.82733 | 0.73647 | 0.75403 | 1 | |||||||||||||||||||||
| 15 | Yaradua | 0.68253 | 0.81937 | 0.75493 | 0.82842 | 0.78262 | 0.73357 | 0.74715 | 0.86371 | 0.62462 | 0.79638 | 0.69701 | 0.69828 | 0.71493 | 0.74027 | 1 | ||||||||||||||||||||
| 16 | Jonathan | 0.75457 | 0.75222 | 0.73557 | 0.68995 | 0.7714 | 0.83656 | 0.78534 | 0.71873 | 0.82914 | 0.76633 | 0.80217 | 0.86697 | 0.79113 | 0.79348 | 0.67656 | 1 | |||||||||||||||||||
| 17 | Buhari | 0.7638 | 0.75493 | 0.77629 | 0.85919 | 0.73104 | 0.81213 | 0.76851 | 0.8067 | 0.71747 | 0.76344 | 0.84706 | 0.7095 | 0.80941 | 0.74281 | 0.79548 | 0.70643 | 1 | ||||||||||||||||||
| 18 | Buhari | 0.74081 | 0.84833 | 0.71529 | 0.78244 | 0.70407 | 0.72977 | 0.80977 | 0.63819 | 0.79131 | 0.7229 | 0.80416 | 0.7457 | 0.74715 | 0.8543 | 0.73774 | 0.82281 | 0.74136 | 1 | |||||||||||||||||
Clustering in the ascension/inaugural addresses (corpus) using a heatmap
We used the cosine similarity matrix constructed earlier to map the different similarities in the ascension/inaugural addresses on a heatmap and visualize groups of addresses that are more likely to cluster together using the default hierarchical clustering method for the heatmap function. In our current analysis, a yellow square indicates strong similarity in addresses and a red square indicates dissimilar addresses. The dendrogram in Fig. 10 shows the steps in the hierarchical clustering algorithm and is similar and confirms the results of the cosine matrix output in Table 7. The numbers in Fig. 10 represent the presidents(see Table 7).
Fig. 10.
Heatmap of similarities between the ascension/inaugural addresses
Discussion
In this research, text-mining techniques were used to analyse textual data and sentiment. We have investigated the publicly accessible ascension and inaugural addresses of Nigerian Heads of State and Presidents from 1960 to 2019. These techniques stipulate a significant tool to appraise the consequences and expectations of Nigerians from a political speech. The sentiments of the speech were answered based on the results of the ascension and inaugural addresses of Nigeria's Heads of State and Presidents.
The sum of characters for the civilian and military were 146,649 and 92,104 respectively. The sum of words used by the civilian president and military heads of state were 14,665 and 11,513 respectively. The average number of words used by the civilian presidents and military heads of state was 1820. We found out that the civilian presidents used more words and characters than the military heads of state. Shonekan, in the year 1993, used the lowest number of characters of 3909 and the number of words of 669 in the ascension/Inaugural addresses. Azikiwe in 1963 used the highest number of characters of 28,267 and several words of 4725. This is expected because of his work experience in the journalism industry. In 2003, Obasanjo used 25,098 characters with 4160 words, this was unexpected because of his military background, though he served as a civilian president in 2003. In 1983, Buhari used 8501 characters and 1330 words, in 2015, he used 11,617 characters and 1915 words, and in 2019, he used 19,014 characters and 3021 words.
By comparing the leaders who served the country three times, Buhari and Obasanjo fulfilled this condition, Obasanjo used more characters and words than President Buhari in the ascension/inaugural Addresses. The total sentiment score for the Civilian Presidents and Military Heads of States were compared. For the Civilian Presidents, the sentiment analysis for the ascension/inaugural addresses was based on the count and the graph. Most of the emotions from the speeches were positive, and the positive emotions were higher when compared to emotions that showed anticipation, disgust, fear, joy, sadness, surprise, trust and negative.
For the Military Heads of States, the sentiment analysis for the ascension/inaugural addresses based on the count and the graph, positive emotions were not up to half, the positive emotions were really small. We still observed a few emotions that indicated anticipation, disgust, fear, joy, sadness, surprise, trust and negative emotions for the military regime. Generally, Civilian Presidents had more positive emotions than Military Heads of States. The civilian presidents expressed more joy and trust emotions.
The sentiment score in the ascension/inaugural addresses from 1960 to 2019 showed that the Civilian Presidents express more positive, trust, joy and anticipation emotions in their addresses than the Military Heads of State. This is similar to Permana and Mauriyat (2021b), where the speaker intended to create a future. Obasanjo’s 2003 inaugural address recorded the highest positive and trust emotions followed by Azikwe’s inaugural address in 1963. Abacha’s ascension address in 1993 had no disgust emotion, while Obasanjo’s address in 1999 had the highest disgust when compared with other Heads-of-State/Presidents' addresses. The highest average sentiment score was observed in Obasanjo’s 2003 inaugural address and the lowest score was in Buhari’s 1983 ascension address.
Our model findings showed that the civilian presidents used more words and characters in the ascension and inaugural addresses than the military heads of states. This is in support of Enyi (2016a), where President Buhari's utterances were direct when he was a military Head of State. The leaders who had served the nation had an increment in the number of words and characters in their ascension and inaugural addresses. The highest average words and characters were recorded by Obasanjo.
The word cloud indicated the importance and the words that appeared frequently in the ascension and inaugural addresses of Nigerian Heads of State and Presidents. The word “government” was more frequent than other words in the ascension and inaugural addresses.
Conclusion
In this study, we were able to establish that the length and number of words in ascension/inaugural speeches delivered by civilian presidents are greater than the military heads of state. In addition, among past Presidents and Heads of State who had served at least twice, Obasanjo had the highest average number of characters (length of words) and number of words.
The research also showed that the ascension/inaugural addresses share similar topic distribution: as seen in Abacha’s and Muritala’s addresses; and Shonekan’s inaugural address was very similar to Balewa, Azikwe and Babangida's addresses; Babangida's ascension, Abdulsalam’s 1998 ascension, Jonathan’s 2010 inaugural and Buhari’s 2015 inaugural addresses discussed similar topics to Obasanjo’s 1976 ascension.
However, Obasanjo’s 2003 inaugural address has the highest positive and trust emotions followed by Azikwe’s inaugural address in 1963. Surprisingly, Abacha’s ascension address in 1993 has no disgust emotion while Obasanjo’s address in 1999 had the highest disgust emotion when compared with other Heads-of-State/Presidents' addresses. Furthermore, civilian presidents express more emotions such as Trust, Surprise, Sadness, Joy, Fear, Disgust and Anticipation in their addresses than the military heads of state in terms of the sentiment scores. The highest average sentiment score was observed in Obasanjo’s 2003 inaugural address and the lowest score was in Buhari’s 1983 ascension address.
Acknowledgements
The authors are grateful to the anonymous reviewers for their comments to improve the clarity and quality of the paper.
Author contributions
KRF contributed to the conceptualization, LOM contributed to the review of literature, KRF, LOM, ME, OJI contributed to the manuscript preparation. ME contributed to the logic and planning. KRF, LOM, ME, OJI contributed to the study design. KFR contributed to the data analysis. OJI contributed to the discussion of the findings. KRF, LOM, ME, OJI read and approved the final manuscript.
Funding
No funds, grants, or other support was received.
Data availability
The datasets analysed during the current study are available from www.dawodu.com.
Declarations
Conflict of interest
On behalf of all the authors, the corresponding author states that there is no conflict of interest.
Ethical approval
We used secondary data which are available for public use.
Informed consent
This article does not contain any studies with human participants performed by any of the authors.
Contributor Information
Lawal Olumuyiwa Mashood, Email: maslaw008@gmail.com.
Olayemi Joshua Ibidoja, Email: ojibidoja@fugusau.edu.ng.
References
- Adegbija E. I, major-general X: discourse tacts in military coup speeches in Nigeria. Interdiscip J Study Discourse. 1995;15(2):253–270. [Google Scholar]
- Adetunji A. The Speech Acts and Rhetorics in the Second Inaugural Addresses of Nigeria’s President Olusegun Obasanjo and America’s President George W Bush. In: Odebunmi A, Arua AE, Arimi S, editors. Language Gender and Politics A Festschrift for Yisa Kehinde Yusuf. Lagos: Concept Publications Limited; 2009. pp. 275–296. [Google Scholar]
- Adeyanju D. Pragmatic features of political speeches in english by some prominent nigerian leaders. Ife Stud Eng Lang. 2006;6(1):48–63. [Google Scholar]
- Akinwotu SA. A speech act analysis of the acceptance of nomination speeches of chief obafemi awolowo and chief M.K.O. Abiola. Eng Linguist Res. 2013;2(1):43–51. [Google Scholar]
- Aremu, M. (2017) Pragmatic analysis of conceptual mappings in inaugural speeches of Nigerian presidents. Covenant J Lang Stud (CJLS) 5(2):32–47
- Balogun S, Amodu J. Parallelism in Inaugural Speeches of Presidents Barrack Obama and Goodluck Jonathan. Nigeria: LAPAI RESEARCH IN HUMANITIES; 2018. [Google Scholar]
- Barkur G, Vibha, & Kamath, G. B. Sentiment analysis of nationwide lockdown due to COVID 19 outbreak: evidence from India. Asian J Psychiatry. 2020;51:102089. doi: 10.1016/j.ajp.2020.102089. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Botha A, Abdile M. Reality versus perception: toward understanding Boko Haram in Nigeria. Stud Confl Terror. 2019;42(5):493–519. doi: 10.1080/1057610X.2018.1403152. [DOI] [Google Scholar]
- Charteris-Black J. Analysing political speeches: Rhetoric, discourse and metaphor. Basingstoke: Palgrave Macmillan; 2018. [Google Scholar]
- Ellah SM. Intertextuality in political discourse: a study of President Mohammadu Buhari’s 2015 inaugural speech. J Lang Linguist Lit Stud. 2022;11(1):9–17. [Google Scholar]
- Ellah SM, Nta NG. Illocutionary function of language in the healing miracles of jesus christ. LWATI: J Contemp Res. 2020;17(3):148–163. [Google Scholar]
- Enyi AU. Pragmatic analysis OF Nigeria’s president Muhammadu Buhari’s maiden coup address of january 1, 1984 and his inaugural address of may 29, 2015: a comparative appraISAL. Int J Eng Lang Linguist Res. 2016;4(5):47–64. [Google Scholar]
- Enyi AU. Pragmatic analysis of Nigeria’s president Muhammadu Buhari’s maiden coup address of January 1, 1984 and his inaugural address of May 29, 2015: a comparative appraisal. Int J Engl Lang Linguist Res. 2016;4(5):47–64. [Google Scholar]
- Han H, Lim S. Inaugural Speech classification with named entities and key phrases. IEEE Int Conf Big Data Smart Comput (BigComp) 2021 doi: 10.1109/BigComp51126.2021.00029. [DOI] [Google Scholar]
- Jockers, M. M. (2020). Package “syuzhet” Type Package Title Extracts Sentiment and Sentiment-Derived Plot Arcs from Text. https://github.com/mjockers/syuzhet
- Lim S, Han H. Distinguishing political parties from inaugural speeches using machine learning. IEEE Conf Multimed Inf Proc Retr (MIPR) 2020 doi: 10.1109/MIPR49039.2020.00026. [DOI] [Google Scholar]
- Mey JL. Pragmatics: an introduction. 2. Oxford: Blackwell; 2001. [Google Scholar]
- Nnamdi-Eruchalu GI. A critical discourse analysis of Muhammadu Buhari’s inaugural speeches with a focus on pronouns. Int J Lit Lang Linguist. 2017;4(2):156–164. [Google Scholar]
- Ogungbe EO (2021) Lexico-syntactic Stratagems in President Muhammadu Buhari’s Inaugural Speeches. ANSU J Lang Lit Stud (AJLLS) 2(1):1–16
- Onyenwe, I., Nwagbo, S., Mbeledogu, N., Onyedinma, E., Onyenwe, I., Nwagbo, S., Mbeledogu, N., & Onyedinma, E. (2020). The Impact of Political Party/Candidate on the Election Results From A Sentiment Analysis Perspective Using #AnambraDecides2017 Tweets. https://courses.lumenlearning.com/americangovernment/chapter/introduction-9/
- Osisanwo A. I Belong to Everybody yet to nobody: pragmatic acts in president Muhammadu Buhari’s inaugural speech. Athens J Mass Med Commun. 2017;3(4):297–320. doi: 10.30958/ajmmc/3.4.2. [DOI] [Google Scholar]
- Oyebode O, Orji R. Social Media and Sentiment Analysis: the Nigeria Presidential Election 2019. Piscataway: IEEE; 2019. [Google Scholar]
- Pak A, Paroubek P. Twitter as a corpus for sentiment analysis and opinion mining. Lrec. 2010;10:1320–1326. [Google Scholar]
- Permana TJ, Mauriyat A. Analysis of speech acts in susilo bambang yudhoyono’s political speech. Prof J Eng Educ. 2021;4(1):62–68. [Google Scholar]
- Permana TJ, Mauriyat A. Analysis of speech acts in susilo bambang yudhoyono’s political speech. PROJECT (prof J Eng Educ) 2021;4(1):62. [Google Scholar]
- Rudkowsky E, Haselmayer M, Wastian M, Jenny M, Emrich Š, Sedlmair M. More than bags of words: sentiment analysis with word embeddings. Commun Methods Meas. 2018;12(3):140–157. doi: 10.1080/19312458.2018.1455817. [DOI] [Google Scholar]
- Yang L, Li Y, Wang J, Sherratt RS. Sentiment analysis for e-commerce product reviews in chinese based on sentiment lexicon and deep learning. IEEE Access. 2020;8:23522–23530. doi: 10.1109/ACCESS.2020.2969854. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets analysed during the current study are available from www.dawodu.com.